diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,40232 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999850703930966,
+  "eval_steps": 500,
+  "global_step": 33490,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.971931919976113e-08,
+      "loss": 1.4585,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9859659599880566e-07,
+      "loss": 1.6256,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.971931919976113e-07,
+      "loss": 1.7173,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.957897879964169e-07,
+      "loss": 1.7771,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1943863839952226e-06,
+      "loss": 1.7536,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4929829799940282e-06,
+      "loss": 1.6147,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7915795759928338e-06,
+      "loss": 1.6947,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0901761719916392e-06,
+      "loss": 1.5887,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3887727679904453e-06,
+      "loss": 1.6785,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6873693639892504e-06,
+      "loss": 1.6417,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9859659599880565e-06,
+      "loss": 1.7147,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.284562555986862e-06,
+      "loss": 1.7063,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5831591519856677e-06,
+      "loss": 1.6572,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.881755747984473e-06,
+      "loss": 1.5187,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1803523439832785e-06,
+      "loss": 1.62,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.478948939982085e-06,
+      "loss": 1.6468,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7775455359808905e-06,
+      "loss": 1.5068,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.076142131979695e-06,
+      "loss": 1.5015,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.374738727978501e-06,
+      "loss": 1.508,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.673335323977307e-06,
+      "loss": 1.585,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.971931919976113e-06,
+      "loss": 1.6127,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.2705285159749185e-06,
+      "loss": 1.4654,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.569125111973724e-06,
+      "loss": 1.5766,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.86772170797253e-06,
+      "loss": 1.5265,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.166318303971335e-06,
+      "loss": 1.5008,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.464914899970141e-06,
+      "loss": 1.4499,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.763511495968946e-06,
+      "loss": 1.3783,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.062108091967751e-06,
+      "loss": 1.4394,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.360704687966557e-06,
+      "loss": 1.3813,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.659301283965363e-06,
+      "loss": 1.3199,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.95789787996417e-06,
+      "loss": 1.3726,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.256494475962975e-06,
+      "loss": 1.3185,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.555091071961781e-06,
+      "loss": 1.3442,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.853687667960585e-06,
+      "loss": 1.3801,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.015228426395939e-05,
+      "loss": 1.39,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0450880859958196e-05,
+      "loss": 1.3099,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0749477455957002e-05,
+      "loss": 1.2959,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1048074051955807e-05,
+      "loss": 1.3294,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1346670647954615e-05,
+      "loss": 1.4781,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.164526724395342e-05,
+      "loss": 1.2775,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1943863839952226e-05,
+      "loss": 1.2934,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2242460435951031e-05,
+      "loss": 1.3089,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2541057031949837e-05,
+      "loss": 1.283,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2839653627948641e-05,
+      "loss": 1.4208,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3138250223947448e-05,
+      "loss": 1.2776,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3436846819946252e-05,
+      "loss": 1.2515,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.373544341594506e-05,
+      "loss": 1.2587,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4034040011943863e-05,
+      "loss": 1.2766,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.433263660794267e-05,
+      "loss": 1.1937,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4631233203941475e-05,
+      "loss": 1.2291,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4929829799940282e-05,
+      "loss": 1.2244,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5228426395939088e-05,
+      "loss": 1.2706,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.552702299193789e-05,
+      "loss": 1.242,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.58256195879367e-05,
+      "loss": 1.2132,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6124216183935503e-05,
+      "loss": 1.2591,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.642281277993431e-05,
+      "loss": 1.2301,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6721409375933114e-05,
+      "loss": 1.2511,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.702000597193192e-05,
+      "loss": 1.2778,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7318602567930725e-05,
+      "loss": 1.2718,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.761719916392953e-05,
+      "loss": 1.2428,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.791579575992834e-05,
+      "loss": 1.2418,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8214392355927142e-05,
+      "loss": 1.2644,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.851298895192595e-05,
+      "loss": 1.2524,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8811585547924753e-05,
+      "loss": 1.3222,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9110182143923562e-05,
+      "loss": 1.1725,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9408778739922364e-05,
+      "loss": 1.2395,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.970737533592117e-05,
+      "loss": 1.179,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.000597193191998e-05,
+      "loss": 1.1765,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.030456852791878e-05,
+      "loss": 1.2232,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.060316512391759e-05,
+      "loss": 1.1905,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0901761719916392e-05,
+      "loss": 1.2126,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.12003583159152e-05,
+      "loss": 1.1298,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1498954911914004e-05,
+      "loss": 1.1882,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1797551507912813e-05,
+      "loss": 1.2301,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2096148103911615e-05,
+      "loss": 1.1727,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2394744699910424e-05,
+      "loss": 1.0941,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.269334129590923e-05,
+      "loss": 1.224,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.299193789190803e-05,
+      "loss": 1.1372,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.329053448790684e-05,
+      "loss": 1.2249,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3589131083905643e-05,
+      "loss": 1.2718,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3887727679904452e-05,
+      "loss": 1.0996,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4186324275903254e-05,
+      "loss": 1.1589,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4484920871902063e-05,
+      "loss": 1.2235,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4783517467900865e-05,
+      "loss": 1.1143,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5082114063899674e-05,
+      "loss": 1.1644,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5380710659898476e-05,
+      "loss": 1.163,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5679307255897282e-05,
+      "loss": 1.086,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.597790385189609e-05,
+      "loss": 1.1865,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6276500447894897e-05,
+      "loss": 1.1587,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.65750970438937e-05,
+      "loss": 1.1743,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6873693639892504e-05,
+      "loss": 1.2178,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7172290235891313e-05,
+      "loss": 1.1798,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.747088683189012e-05,
+      "loss": 1.1444,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7769483427888925e-05,
+      "loss": 1.2425,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8068080023887727e-05,
+      "loss": 1.1264,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8366676619886532e-05,
+      "loss": 1.1555,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.866527321588534e-05,
+      "loss": 1.2075,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8963869811884147e-05,
+      "loss": 1.1457,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.926246640788295e-05,
+      "loss": 1.1104,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9561063003881755e-05,
+      "loss": 1.2528,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9859659599880564e-05,
+      "loss": 1.1005,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.015825619587937e-05,
+      "loss": 1.1275,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0456852791878175e-05,
+      "loss": 1.1357,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.075544938787698e-05,
+      "loss": 1.1721,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.105404598387578e-05,
+      "loss": 1.1692,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.135264257987459e-05,
+      "loss": 1.142,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.16512391758734e-05,
+      "loss": 1.1517,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1949835771872196e-05,
+      "loss": 1.1611,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2248432367871005e-05,
+      "loss": 1.2085,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2547028963869814e-05,
+      "loss": 1.0626,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.284562555986862e-05,
+      "loss": 1.1229,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3144222155867426e-05,
+      "loss": 1.151,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.344281875186623e-05,
+      "loss": 1.0884,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.374141534786504e-05,
+      "loss": 1.1217,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.404001194386384e-05,
+      "loss": 1.151,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.433860853986265e-05,
+      "loss": 1.1489,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.463720513586145e-05,
+      "loss": 1.1657,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.493580173186026e-05,
+      "loss": 1.1612,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.523439832785906e-05,
+      "loss": 1.1955,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.553299492385787e-05,
+      "loss": 1.047,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.583159151985668e-05,
+      "loss": 1.0708,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.613018811585548e-05,
+      "loss": 1.2311,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6428784711854284e-05,
+      "loss": 1.2616,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.672738130785309e-05,
+      "loss": 1.1666,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.70259779038519e-05,
+      "loss": 1.1123,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7324574499850704e-05,
+      "loss": 1.1402,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7623171095849506e-05,
+      "loss": 1.1215,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7921767691848315e-05,
+      "loss": 1.0725,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8220364287847124e-05,
+      "loss": 1.1213,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8518960883845926e-05,
+      "loss": 1.1224,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.881755747984473e-05,
+      "loss": 1.1197,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.911615407584354e-05,
+      "loss": 1.1283,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.941475067184234e-05,
+      "loss": 1.1655,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.971334726784115e-05,
+      "loss": 1.0951,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.001194386383996e-05,
+      "loss": 1.1543,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.031054045983876e-05,
+      "loss": 1.0536,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.060913705583756e-05,
+      "loss": 1.1131,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.090773365183637e-05,
+      "loss": 1.0949,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.120633024783518e-05,
+      "loss": 1.1285,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.150492684383398e-05,
+      "loss": 1.1501,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1803523439832785e-05,
+      "loss": 1.1675,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2102120035831594e-05,
+      "loss": 1.1537,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.24007166318304e-05,
+      "loss": 1.0697,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2699313227829205e-05,
+      "loss": 1.0955,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.299790982382801e-05,
+      "loss": 1.0604,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3296506419826816e-05,
+      "loss": 1.0595,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3595103015825625e-05,
+      "loss": 1.064,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.389369961182443e-05,
+      "loss": 1.1457,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.419229620782323e-05,
+      "loss": 1.2189,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.449089280382204e-05,
+      "loss": 1.1428,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.478948939982085e-05,
+      "loss": 1.0917,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.508808599581965e-05,
+      "loss": 1.0878,
+      "step": 755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.538668259181846e-05,
+      "loss": 1.1408,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.568527918781726e-05,
+      "loss": 1.1336,
+      "step": 765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.598387578381606e-05,
+      "loss": 1.1708,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.628247237981487e-05,
+      "loss": 1.1405,
+      "step": 775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.658106897581368e-05,
+      "loss": 1.1571,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.687966557181248e-05,
+      "loss": 1.1289,
+      "step": 785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7178262167811286e-05,
+      "loss": 1.1346,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7476858763810095e-05,
+      "loss": 1.1744,
+      "step": 795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7775455359808904e-05,
+      "loss": 1.1404,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8074051955807706e-05,
+      "loss": 1.1259,
+      "step": 805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.837264855180651e-05,
+      "loss": 1.0889,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.867124514780532e-05,
+      "loss": 1.1308,
+      "step": 815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8969841743804126e-05,
+      "loss": 1.2417,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.926843833980293e-05,
+      "loss": 1.0697,
+      "step": 825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.956703493580173e-05,
+      "loss": 1.0881,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986563153180054e-05,
+      "loss": 1.1786,
+      "step": 835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.016422812779935e-05,
+      "loss": 1.1313,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.046282472379815e-05,
+      "loss": 1.193,
+      "step": 845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.076142131979695e-05,
+      "loss": 1.0761,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.106001791579577e-05,
+      "loss": 1.1518,
+      "step": 855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1358614511794564e-05,
+      "loss": 1.0664,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1657211107793366e-05,
+      "loss": 1.1468,
+      "step": 865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.195580770379218e-05,
+      "loss": 1.1731,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2254404299790984e-05,
+      "loss": 1.0429,
+      "step": 875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.255300089578979e-05,
+      "loss": 1.0916,
+      "step": 880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2851597491788595e-05,
+      "loss": 1.1151,
+      "step": 885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.31501940877874e-05,
+      "loss": 1.1738,
+      "step": 890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.344879068378621e-05,
+      "loss": 1.1225,
+      "step": 895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.374738727978501e-05,
+      "loss": 1.1228,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4045983875783825e-05,
+      "loss": 1.1076,
+      "step": 905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.434458047178263e-05,
+      "loss": 1.0883,
+      "step": 910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.464317706778143e-05,
+      "loss": 1.0699,
+      "step": 915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.494177366378024e-05,
+      "loss": 1.0786,
+      "step": 920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.524037025977904e-05,
+      "loss": 1.0524,
+      "step": 925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.553896685577785e-05,
+      "loss": 1.1435,
+      "step": 930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.583756345177665e-05,
+      "loss": 1.0922,
+      "step": 935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6136160047775454e-05,
+      "loss": 1.0464,
+      "step": 940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.643475664377427e-05,
+      "loss": 1.1485,
+      "step": 945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6733353239773065e-05,
+      "loss": 1.1303,
+      "step": 950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.703194983577187e-05,
+      "loss": 1.1807,
+      "step": 955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.733054643177068e-05,
+      "loss": 1.1524,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7629143027769485e-05,
+      "loss": 1.1062,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7927739623768294e-05,
+      "loss": 1.1626,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8226336219767096e-05,
+      "loss": 1.1152,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.85249328157659e-05,
+      "loss": 1.0504,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.882352941176471e-05,
+      "loss": 1.1192,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.912212600776351e-05,
+      "loss": 1.1953,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9420722603762325e-05,
+      "loss": 1.1602,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.971931919976113e-05,
+      "loss": 1.1221,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.001791579575993e-05,
+      "loss": 1.1316,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.031651239175874e-05,
+      "loss": 1.0403,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.061510898775754e-05,
+      "loss": 1.1145,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.091370558375635e-05,
+      "loss": 1.1008,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.121230217975516e-05,
+      "loss": 1.1407,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.151089877575396e-05,
+      "loss": 1.1395,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.180949537175276e-05,
+      "loss": 1.0828,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.210809196775157e-05,
+      "loss": 1.1553,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.240668856375038e-05,
+      "loss": 1.1672,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.270528515974918e-05,
+      "loss": 1.1025,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.300388175574799e-05,
+      "loss": 1.1906,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.33024783517468e-05,
+      "loss": 1.0398,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.36010749477456e-05,
+      "loss": 1.0955,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.389967154374439e-05,
+      "loss": 1.1328,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.419826813974321e-05,
+      "loss": 1.1968,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.449686473574201e-05,
+      "loss": 1.1023,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.479546133174083e-05,
+      "loss": 1.1678,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.509405792773963e-05,
+      "loss": 1.0655,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.539265452373843e-05,
+      "loss": 1.2225,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.569125111973725e-05,
+      "loss": 1.0962,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.598984771573604e-05,
+      "loss": 1.128,
+      "step": 1105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.628844431173485e-05,
+      "loss": 1.0188,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.658704090773365e-05,
+      "loss": 1.1631,
+      "step": 1115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.688563750373246e-05,
+      "loss": 1.1258,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.718423409973127e-05,
+      "loss": 1.0345,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.748283069573007e-05,
+      "loss": 1.0857,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.778142729172889e-05,
+      "loss": 1.1195,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.808002388772768e-05,
+      "loss": 1.1317,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.837862048372648e-05,
+      "loss": 1.0549,
+      "step": 1145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.86772170797253e-05,
+      "loss": 1.0701,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.89758136757241e-05,
+      "loss": 1.1227,
+      "step": 1155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.92744102717229e-05,
+      "loss": 1.0756,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.957300686772172e-05,
+      "loss": 1.1506,
+      "step": 1165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.987160346372052e-05,
+      "loss": 1.1475,
+      "step": 1170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.017020005971932e-05,
+      "loss": 1.1378,
+      "step": 1175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.046879665571812e-05,
+      "loss": 1.0497,
+      "step": 1180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.076739325171692e-05,
+      "loss": 1.0586,
+      "step": 1185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.106598984771574e-05,
+      "loss": 1.085,
+      "step": 1190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.136458644371454e-05,
+      "loss": 1.1794,
+      "step": 1195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.166318303971336e-05,
+      "loss": 1.0421,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.196177963571216e-05,
+      "loss": 1.0732,
+      "step": 1205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.226037623171096e-05,
+      "loss": 1.0582,
+      "step": 1210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.255897282770977e-05,
+      "loss": 1.0111,
+      "step": 1215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.285756942370857e-05,
+      "loss": 1.0703,
+      "step": 1220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.315616601970738e-05,
+      "loss": 1.1753,
+      "step": 1225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.345476261570619e-05,
+      "loss": 1.1485,
+      "step": 1230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.375335921170499e-05,
+      "loss": 1.0741,
+      "step": 1235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.40519558077038e-05,
+      "loss": 1.1243,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.43505524037026e-05,
+      "loss": 1.0602,
+      "step": 1245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.464914899970141e-05,
+      "loss": 1.1495,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.494774559570021e-05,
+      "loss": 1.1245,
+      "step": 1255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.524634219169901e-05,
+      "loss": 1.0684,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.554493878769783e-05,
+      "loss": 1.1399,
+      "step": 1265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.584353538369663e-05,
+      "loss": 1.0606,
+      "step": 1270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.614213197969543e-05,
+      "loss": 1.107,
+      "step": 1275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.644072857569425e-05,
+      "loss": 1.11,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.673932517169305e-05,
+      "loss": 1.1872,
+      "step": 1285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.703792176769185e-05,
+      "loss": 1.1942,
+      "step": 1290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.733651836369065e-05,
+      "loss": 1.1292,
+      "step": 1295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.763511495968946e-05,
+      "loss": 1.0424,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.793371155568827e-05,
+      "loss": 1.1061,
+      "step": 1305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.823230815168708e-05,
+      "loss": 1.1405,
+      "step": 1310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.853090474768589e-05,
+      "loss": 1.1527,
+      "step": 1315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.882950134368468e-05,
+      "loss": 1.0768,
+      "step": 1320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.912809793968348e-05,
+      "loss": 1.1521,
+      "step": 1325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.94266945356823e-05,
+      "loss": 1.1386,
+      "step": 1330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.97252911316811e-05,
+      "loss": 1.1183,
+      "step": 1335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.002388772767992e-05,
+      "loss": 1.1109,
+      "step": 1340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.032248432367872e-05,
+      "loss": 1.0986,
+      "step": 1345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.062108091967752e-05,
+      "loss": 1.1451,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.091967751567632e-05,
+      "loss": 1.1205,
+      "step": 1355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.121827411167512e-05,
+      "loss": 1.0236,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.151687070767393e-05,
+      "loss": 1.0733,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.181546730367274e-05,
+      "loss": 1.0487,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.211406389967154e-05,
+      "loss": 1.0805,
+      "step": 1375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.241266049567036e-05,
+      "loss": 1.0982,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.271125709166916e-05,
+      "loss": 1.0558,
+      "step": 1385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.300985368766796e-05,
+      "loss": 1.0834,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.330845028366677e-05,
+      "loss": 1.1727,
+      "step": 1395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.360704687966557e-05,
+      "loss": 1.1002,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.390564347566439e-05,
+      "loss": 1.142,
+      "step": 1405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.420424007166319e-05,
+      "loss": 1.0874,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.450283666766199e-05,
+      "loss": 1.0821,
+      "step": 1415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.48014332636608e-05,
+      "loss": 1.0858,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.510002985965961e-05,
+      "loss": 1.0737,
+      "step": 1425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.539862645565841e-05,
+      "loss": 1.0453,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.569722305165721e-05,
+      "loss": 1.1273,
+      "step": 1435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.599581964765601e-05,
+      "loss": 1.0571,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.629441624365483e-05,
+      "loss": 1.0576,
+      "step": 1445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.659301283965363e-05,
+      "loss": 1.1567,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.689160943565243e-05,
+      "loss": 1.157,
+      "step": 1455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.719020603165125e-05,
+      "loss": 1.1469,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.748880262765005e-05,
+      "loss": 1.1209,
+      "step": 1465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.778739922364885e-05,
+      "loss": 1.0979,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.808599581964766e-05,
+      "loss": 1.1423,
+      "step": 1475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.838459241564646e-05,
+      "loss": 1.1072,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.868318901164527e-05,
+      "loss": 1.1439,
+      "step": 1485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.898178560764408e-05,
+      "loss": 1.0568,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.928038220364289e-05,
+      "loss": 1.183,
+      "step": 1495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.95789787996417e-05,
+      "loss": 1.0632,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.987757539564048e-05,
+      "loss": 1.0712,
+      "step": 1505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.01761719916393e-05,
+      "loss": 1.0539,
+      "step": 1510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.04747685876381e-05,
+      "loss": 1.053,
+      "step": 1515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.077336518363692e-05,
+      "loss": 1.0939,
+      "step": 1520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.107196177963572e-05,
+      "loss": 1.0298,
+      "step": 1525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.137055837563452e-05,
+      "loss": 1.114,
+      "step": 1530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.166915497163334e-05,
+      "loss": 1.0674,
+      "step": 1535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.196775156763213e-05,
+      "loss": 1.1114,
+      "step": 1540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.226634816363093e-05,
+      "loss": 1.0624,
+      "step": 1545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.256494475962974e-05,
+      "loss": 1.0214,
+      "step": 1550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.286354135562855e-05,
+      "loss": 1.0792,
+      "step": 1555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.316213795162736e-05,
+      "loss": 1.0711,
+      "step": 1560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.346073454762616e-05,
+      "loss": 1.0985,
+      "step": 1565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.375933114362497e-05,
+      "loss": 1.0999,
+      "step": 1570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.405792773962377e-05,
+      "loss": 1.1437,
+      "step": 1575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.435652433562257e-05,
+      "loss": 1.1081,
+      "step": 1580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.465512093162139e-05,
+      "loss": 1.1424,
+      "step": 1585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.495371752762019e-05,
+      "loss": 1.0566,
+      "step": 1590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.525231412361899e-05,
+      "loss": 1.0845,
+      "step": 1595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.555091071961781e-05,
+      "loss": 1.0832,
+      "step": 1600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.584950731561661e-05,
+      "loss": 1.1048,
+      "step": 1605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.614810391161541e-05,
+      "loss": 1.0849,
+      "step": 1610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.644670050761421e-05,
+      "loss": 1.0663,
+      "step": 1615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.674529710361302e-05,
+      "loss": 1.0951,
+      "step": 1620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.704389369961183e-05,
+      "loss": 1.0655,
+      "step": 1625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.734249029561063e-05,
+      "loss": 1.081,
+      "step": 1630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.764108689160945e-05,
+      "loss": 1.0622,
+      "step": 1635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.793968348760825e-05,
+      "loss": 1.0711,
+      "step": 1640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.823828008360705e-05,
+      "loss": 1.0491,
+      "step": 1645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.853687667960586e-05,
+      "loss": 1.1492,
+      "step": 1650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.883547327560466e-05,
+      "loss": 1.1284,
+      "step": 1655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.913406987160346e-05,
+      "loss": 1.0678,
+      "step": 1660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943266646760228e-05,
+      "loss": 1.1252,
+      "step": 1665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.973126306360108e-05,
+      "loss": 1.1307,
+      "step": 1670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010002985965959988,
+      "loss": 1.0359,
+      "step": 1675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001003284562555987,
+      "loss": 1.1049,
+      "step": 1680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001006270528515975,
+      "loss": 1.1471,
+      "step": 1685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001009256494475963,
+      "loss": 1.1246,
+      "step": 1690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001012242460435951,
+      "loss": 1.076,
+      "step": 1695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001015228426395939,
+      "loss": 1.076,
+      "step": 1700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010182143923559271,
+      "loss": 1.1156,
+      "step": 1705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010212003583159154,
+      "loss": 1.0868,
+      "step": 1710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010241863242759034,
+      "loss": 1.0114,
+      "step": 1715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010271722902358913,
+      "loss": 1.1403,
+      "step": 1720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010301582561958793,
+      "loss": 1.1826,
+      "step": 1725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010331442221558673,
+      "loss": 1.064,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010361301881158556,
+      "loss": 1.0273,
+      "step": 1735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010391161540758436,
+      "loss": 1.1297,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010421021200358317,
+      "loss": 1.16,
+      "step": 1745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010450880859958197,
+      "loss": 1.0746,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010480740519558077,
+      "loss": 1.1209,
+      "step": 1755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010510600179157959,
+      "loss": 1.0853,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010540459838757839,
+      "loss": 1.1119,
+      "step": 1765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010570319498357719,
+      "loss": 1.078,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010600179157957599,
+      "loss": 1.1514,
+      "step": 1775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001063003881755748,
+      "loss": 1.1006,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010659898477157362,
+      "loss": 1.0693,
+      "step": 1785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010689758136757241,
+      "loss": 1.0863,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010719617796357122,
+      "loss": 1.1365,
+      "step": 1795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010749477455957002,
+      "loss": 1.1399,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010779337115556882,
+      "loss": 1.0507,
+      "step": 1805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010809196775156765,
+      "loss": 1.0029,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010839056434756645,
+      "loss": 1.2431,
+      "step": 1815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010868916094356525,
+      "loss": 1.1512,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010898775753956406,
+      "loss": 1.1036,
+      "step": 1825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010928635413556286,
+      "loss": 1.0572,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010958495073156167,
+      "loss": 1.1712,
+      "step": 1835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010988354732756048,
+      "loss": 1.0528,
+      "step": 1840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011018214392355928,
+      "loss": 1.0917,
+      "step": 1845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011048074051955808,
+      "loss": 1.0847,
+      "step": 1850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011077933711555688,
+      "loss": 1.1046,
+      "step": 1855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001110779337115557,
+      "loss": 1.0963,
+      "step": 1860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001113765303075545,
+      "loss": 1.1022,
+      "step": 1865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001116751269035533,
+      "loss": 1.1821,
+      "step": 1870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001119737234995521,
+      "loss": 1.1033,
+      "step": 1875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011227232009555091,
+      "loss": 1.1271,
+      "step": 1880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011257091669154974,
+      "loss": 1.0473,
+      "step": 1885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011286951328754854,
+      "loss": 1.1074,
+      "step": 1890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011316810988354734,
+      "loss": 0.9955,
+      "step": 1895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011346670647954613,
+      "loss": 1.1734,
+      "step": 1900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011376530307554493,
+      "loss": 1.1348,
+      "step": 1905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011406389967154373,
+      "loss": 1.2305,
+      "step": 1910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011436249626754256,
+      "loss": 1.0754,
+      "step": 1915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011466109286354137,
+      "loss": 1.1334,
+      "step": 1920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011495968945954017,
+      "loss": 1.1121,
+      "step": 1925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011525828605553897,
+      "loss": 1.0733,
+      "step": 1930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011555688265153777,
+      "loss": 1.1399,
+      "step": 1935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011585547924753659,
+      "loss": 1.102,
+      "step": 1940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011615407584353539,
+      "loss": 1.0491,
+      "step": 1945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011645267243953419,
+      "loss": 1.0837,
+      "step": 1950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000116751269035533,
+      "loss": 1.0573,
+      "step": 1955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001170498656315318,
+      "loss": 1.1107,
+      "step": 1960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011734846222753063,
+      "loss": 1.1348,
+      "step": 1965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011764705882352942,
+      "loss": 1.0779,
+      "step": 1970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011794565541952822,
+      "loss": 1.0714,
+      "step": 1975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011824425201552702,
+      "loss": 1.0405,
+      "step": 1980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011854284861152582,
+      "loss": 1.1107,
+      "step": 1985
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011884144520752465,
+      "loss": 1.1086,
+      "step": 1990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011914004180352345,
+      "loss": 1.0811,
+      "step": 1995
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011943863839952226,
+      "loss": 1.0469,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011973723499552106,
+      "loss": 1.0867,
+      "step": 2005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012003583159151986,
+      "loss": 1.0944,
+      "step": 2010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012033442818751868,
+      "loss": 1.1057,
+      "step": 2015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012063302478351748,
+      "loss": 1.0371,
+      "step": 2020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012093162137951628,
+      "loss": 1.0589,
+      "step": 2025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012123021797551508,
+      "loss": 1.1048,
+      "step": 2030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012152881457151388,
+      "loss": 1.098,
+      "step": 2035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001218274111675127,
+      "loss": 1.0762,
+      "step": 2040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001221260077635115,
+      "loss": 1.1303,
+      "step": 2045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012242460435951032,
+      "loss": 1.1129,
+      "step": 2050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001227232009555091,
+      "loss": 1.0332,
+      "step": 2055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012302179755150792,
+      "loss": 1.0965,
+      "step": 2060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012332039414750674,
+      "loss": 1.183,
+      "step": 2065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012361899074350553,
+      "loss": 1.013,
+      "step": 2070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012391758733950434,
+      "loss": 1.098,
+      "step": 2075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012421618393550313,
+      "loss": 1.1352,
+      "step": 2080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012451478053150195,
+      "loss": 1.135,
+      "step": 2085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012481337712750076,
+      "loss": 1.0648,
+      "step": 2090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012511197372349955,
+      "loss": 1.0729,
+      "step": 2095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012541057031949837,
+      "loss": 1.1195,
+      "step": 2100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012570916691549716,
+      "loss": 1.0619,
+      "step": 2105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012600776351149597,
+      "loss": 1.0518,
+      "step": 2110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012630636010749476,
+      "loss": 1.107,
+      "step": 2115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001266049567034936,
+      "loss": 1.1509,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001269035532994924,
+      "loss": 1.0284,
+      "step": 2125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001272021498954912,
+      "loss": 1.0268,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012750074649149,
+      "loss": 1.0742,
+      "step": 2135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012779934308748879,
+      "loss": 1.1363,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012809793968348763,
+      "loss": 1.0428,
+      "step": 2145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012839653627948642,
+      "loss": 1.0828,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012869513287548523,
+      "loss": 1.057,
+      "step": 2155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012899372947148402,
+      "loss": 1.1283,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012929232606748284,
+      "loss": 1.0634,
+      "step": 2165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012959092266348165,
+      "loss": 1.04,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012988951925948044,
+      "loss": 1.0368,
+      "step": 2175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013018811585547926,
+      "loss": 1.1117,
+      "step": 2180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013048671245147805,
+      "loss": 1.1447,
+      "step": 2185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013078530904747686,
+      "loss": 1.0725,
+      "step": 2190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013108390564347568,
+      "loss": 1.0388,
+      "step": 2195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001313825022394745,
+      "loss": 1.1068,
+      "step": 2200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013168109883547328,
+      "loss": 1.0089,
+      "step": 2205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013197969543147207,
+      "loss": 1.0564,
+      "step": 2210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013227829202747089,
+      "loss": 1.0147,
+      "step": 2215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001325768886234697,
+      "loss": 1.0147,
+      "step": 2220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013287548521946852,
+      "loss": 1.1182,
+      "step": 2225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001331740818154673,
+      "loss": 1.0915,
+      "step": 2230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013347267841146612,
+      "loss": 1.1495,
+      "step": 2235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001337712750074649,
+      "loss": 1.0869,
+      "step": 2240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013406987160346373,
+      "loss": 1.0828,
+      "step": 2245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013436846819946254,
+      "loss": 1.0157,
+      "step": 2250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013466706479546133,
+      "loss": 1.1523,
+      "step": 2255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013496566139146015,
+      "loss": 1.0472,
+      "step": 2260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013526425798745894,
+      "loss": 1.1676,
+      "step": 2265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013556285458345778,
+      "loss": 1.1472,
+      "step": 2270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013586145117945657,
+      "loss": 1.0972,
+      "step": 2275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013616004777545536,
+      "loss": 1.0558,
+      "step": 2280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013645864437145417,
+      "loss": 1.1049,
+      "step": 2285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013675724096745296,
+      "loss": 1.0733,
+      "step": 2290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013705583756345178,
+      "loss": 1.0504,
+      "step": 2295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001373544341594506,
+      "loss": 1.1005,
+      "step": 2300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001376530307554494,
+      "loss": 1.0958,
+      "step": 2305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001379516273514482,
+      "loss": 1.1191,
+      "step": 2310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000138250223947447,
+      "loss": 1.1083,
+      "step": 2315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001385488205434458,
+      "loss": 1.0621,
+      "step": 2320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013884741713944462,
+      "loss": 1.0963,
+      "step": 2325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013914601373544343,
+      "loss": 1.0541,
+      "step": 2330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013944461033144222,
+      "loss": 1.1256,
+      "step": 2335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013974320692744104,
+      "loss": 1.1294,
+      "step": 2340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014004180352343983,
+      "loss": 1.0728,
+      "step": 2345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014034040011943864,
+      "loss": 1.0333,
+      "step": 2350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014063899671543746,
+      "loss": 1.1385,
+      "step": 2355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014093759331143625,
+      "loss": 1.1034,
+      "step": 2360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014123618990743506,
+      "loss": 1.1511,
+      "step": 2365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014153478650343385,
+      "loss": 1.0998,
+      "step": 2370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001418333830994327,
+      "loss": 1.0593,
+      "step": 2375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014213197969543148,
+      "loss": 1.1902,
+      "step": 2380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001424305762914303,
+      "loss": 1.109,
+      "step": 2385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014272917288742909,
+      "loss": 1.0562,
+      "step": 2390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014302776948342787,
+      "loss": 1.0737,
+      "step": 2395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014332636607942672,
+      "loss": 1.0729,
+      "step": 2400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001436249626754255,
+      "loss": 1.0294,
+      "step": 2405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014392355927142432,
+      "loss": 1.057,
+      "step": 2410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001442221558674231,
+      "loss": 1.0926,
+      "step": 2415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014452075246342193,
+      "loss": 1.0654,
+      "step": 2420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014481934905942074,
+      "loss": 1.0966,
+      "step": 2425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014511794565541953,
+      "loss": 1.0289,
+      "step": 2430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014541654225141835,
+      "loss": 1.0597,
+      "step": 2435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014571513884741714,
+      "loss": 1.0193,
+      "step": 2440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014601373544341595,
+      "loss": 1.0012,
+      "step": 2445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014631233203941477,
+      "loss": 1.0725,
+      "step": 2450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014661092863541358,
+      "loss": 1.0239,
+      "step": 2455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014690952523141237,
+      "loss": 1.1049,
+      "step": 2460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014720812182741116,
+      "loss": 1.0171,
+      "step": 2465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014750671842340998,
+      "loss": 1.0581,
+      "step": 2470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001478053150194088,
+      "loss": 1.1126,
+      "step": 2475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001481039116154076,
+      "loss": 1.054,
+      "step": 2480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001484025082114064,
+      "loss": 1.0213,
+      "step": 2485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001487011048074052,
+      "loss": 1.1767,
+      "step": 2490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000148999701403404,
+      "loss": 1.0246,
+      "step": 2495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014929829799940282,
+      "loss": 1.153,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014959689459540163,
+      "loss": 1.0123,
+      "step": 2505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014989549119140042,
+      "loss": 1.054,
+      "step": 2510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015019408778739924,
+      "loss": 1.145,
+      "step": 2515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015049268438339802,
+      "loss": 1.0201,
+      "step": 2520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015079128097939684,
+      "loss": 1.0261,
+      "step": 2525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015108987757539566,
+      "loss": 1.0733,
+      "step": 2530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015138847417139444,
+      "loss": 1.1479,
+      "step": 2535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015168707076739326,
+      "loss": 1.0782,
+      "step": 2540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015198566736339205,
+      "loss": 1.1269,
+      "step": 2545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015228426395939087,
+      "loss": 1.1491,
+      "step": 2550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015258286055538968,
+      "loss": 1.1119,
+      "step": 2555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001528814571513885,
+      "loss": 1.1064,
+      "step": 2560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015318005374738729,
+      "loss": 1.1304,
+      "step": 2565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001534786503433861,
+      "loss": 1.062,
+      "step": 2570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001537772469393849,
+      "loss": 1.1846,
+      "step": 2575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001540758435353837,
+      "loss": 1.1742,
+      "step": 2580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015437444013138252,
+      "loss": 1.0942,
+      "step": 2585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001546730367273813,
+      "loss": 1.1366,
+      "step": 2590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015497163332338013,
+      "loss": 1.1467,
+      "step": 2595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015527022991937891,
+      "loss": 1.0348,
+      "step": 2600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015556882651537773,
+      "loss": 1.0252,
+      "step": 2605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015586742311137655,
+      "loss": 1.0639,
+      "step": 2610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015616601970737533,
+      "loss": 1.0559,
+      "step": 2615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015646461630337415,
+      "loss": 1.0854,
+      "step": 2620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015676321289937294,
+      "loss": 1.0716,
+      "step": 2625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015706180949537178,
+      "loss": 1.0255,
+      "step": 2630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015736040609137057,
+      "loss": 1.0305,
+      "step": 2635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015765900268736936,
+      "loss": 1.0547,
+      "step": 2640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015795759928336818,
+      "loss": 1.1318,
+      "step": 2645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015825619587936696,
+      "loss": 1.0905,
+      "step": 2650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001585547924753658,
+      "loss": 1.0716,
+      "step": 2655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001588533890713646,
+      "loss": 1.1074,
+      "step": 2660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001591519856673634,
+      "loss": 1.1428,
+      "step": 2665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001594505822633622,
+      "loss": 1.0202,
+      "step": 2670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015974917885936102,
+      "loss": 1.042,
+      "step": 2675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016004777545535983,
+      "loss": 1.1255,
+      "step": 2680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016034637205135862,
+      "loss": 1.0699,
+      "step": 2685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016064496864735744,
+      "loss": 1.1759,
+      "step": 2690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016094356524335622,
+      "loss": 1.0454,
+      "step": 2695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016124216183935504,
+      "loss": 1.0747,
+      "step": 2700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016154075843535383,
+      "loss": 1.071,
+      "step": 2705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016183935503135264,
+      "loss": 1.0476,
+      "step": 2710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016213795162735146,
+      "loss": 1.155,
+      "step": 2715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016243654822335025,
+      "loss": 1.0276,
+      "step": 2720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016273514481934906,
+      "loss": 1.0591,
+      "step": 2725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016303374141534785,
+      "loss": 1.0769,
+      "step": 2730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001633323380113467,
+      "loss": 1.097,
+      "step": 2735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016363093460734548,
+      "loss": 1.0199,
+      "step": 2740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001639295312033443,
+      "loss": 1.0936,
+      "step": 2745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001642281277993431,
+      "loss": 1.086,
+      "step": 2750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016452672439534188,
+      "loss": 1.1206,
+      "step": 2755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016482532099134072,
+      "loss": 1.1686,
+      "step": 2760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001651239175873395,
+      "loss": 1.1004,
+      "step": 2765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016542251418333833,
+      "loss": 1.0702,
+      "step": 2770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016572111077933711,
+      "loss": 1.0575,
+      "step": 2775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016601970737533593,
+      "loss": 1.0978,
+      "step": 2780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016631830397133475,
+      "loss": 1.1407,
+      "step": 2785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016661690056733353,
+      "loss": 1.1511,
+      "step": 2790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016691549716333235,
+      "loss": 1.2012,
+      "step": 2795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016721409375933114,
+      "loss": 1.1247,
+      "step": 2800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016751269035532995,
+      "loss": 1.1194,
+      "step": 2805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016781128695132877,
+      "loss": 1.0245,
+      "step": 2810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016810988354732759,
+      "loss": 1.0469,
+      "step": 2815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016840848014332637,
+      "loss": 1.0938,
+      "step": 2820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016870707673932516,
+      "loss": 1.1133,
+      "step": 2825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016900567333532398,
+      "loss": 1.023,
+      "step": 2830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001693042699313228,
+      "loss": 1.1506,
+      "step": 2835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001696028665273216,
+      "loss": 1.1429,
+      "step": 2840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001699014631233204,
+      "loss": 1.0354,
+      "step": 2845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017020005971931921,
+      "loss": 1.0391,
+      "step": 2850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000170498656315318,
+      "loss": 1.0791,
+      "step": 2855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017079725291131682,
+      "loss": 1.1075,
+      "step": 2860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017109584950731564,
+      "loss": 1.1059,
+      "step": 2865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017139444610331442,
+      "loss": 1.0659,
+      "step": 2870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017169304269931324,
+      "loss": 1.1229,
+      "step": 2875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017199163929531203,
+      "loss": 1.0697,
+      "step": 2880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017229023589131087,
+      "loss": 1.0755,
+      "step": 2885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017258883248730966,
+      "loss": 1.1134,
+      "step": 2890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017288742908330845,
+      "loss": 1.0988,
+      "step": 2895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017318602567930726,
+      "loss": 1.0538,
+      "step": 2900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017348462227530605,
+      "loss": 1.0201,
+      "step": 2905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017378321887130487,
+      "loss": 1.141,
+      "step": 2910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017408181546730368,
+      "loss": 1.1097,
+      "step": 2915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001743804120633025,
+      "loss": 1.066,
+      "step": 2920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001746790086593013,
+      "loss": 1.0917,
+      "step": 2925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001749776052553001,
+      "loss": 1.1559,
+      "step": 2930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001752762018512989,
+      "loss": 1.1427,
+      "step": 2935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001755747984472977,
+      "loss": 1.0514,
+      "step": 2940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017587339504329652,
+      "loss": 1.0922,
+      "step": 2945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001761719916392953,
+      "loss": 1.0638,
+      "step": 2950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017647058823529413,
+      "loss": 1.1429,
+      "step": 2955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017676918483129292,
+      "loss": 1.0247,
+      "step": 2960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017706778142729173,
+      "loss": 1.0268,
+      "step": 2965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017736637802329055,
+      "loss": 1.1297,
+      "step": 2970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017766497461928934,
+      "loss": 1.041,
+      "step": 2975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017796357121528815,
+      "loss": 1.0589,
+      "step": 2980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017826216781128694,
+      "loss": 1.0459,
+      "step": 2985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017856076440728579,
+      "loss": 1.1031,
+      "step": 2990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017885936100328457,
+      "loss": 1.1287,
+      "step": 2995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001791579575992834,
+      "loss": 1.0877,
+      "step": 3000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017945655419528218,
+      "loss": 1.0522,
+      "step": 3005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017975515079128097,
+      "loss": 1.1099,
+      "step": 3010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001800537473872798,
+      "loss": 1.0418,
+      "step": 3015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001803523439832786,
+      "loss": 1.0555,
+      "step": 3020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018065094057927741,
+      "loss": 1.1016,
+      "step": 3025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001809495371752762,
+      "loss": 1.1006,
+      "step": 3030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018124813377127502,
+      "loss": 1.0478,
+      "step": 3035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018154673036727383,
+      "loss": 1.1453,
+      "step": 3040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018184532696327262,
+      "loss": 1.0999,
+      "step": 3045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018214392355927144,
+      "loss": 1.0918,
+      "step": 3050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018244252015527023,
+      "loss": 1.1002,
+      "step": 3055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018274111675126904,
+      "loss": 1.0768,
+      "step": 3060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018303971334726786,
+      "loss": 1.1154,
+      "step": 3065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018333830994326668,
+      "loss": 1.0765,
+      "step": 3070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018363690653926546,
+      "loss": 1.1575,
+      "step": 3075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018393550313526425,
+      "loss": 1.1789,
+      "step": 3080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018423409973126307,
+      "loss": 1.1259,
+      "step": 3085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018453269632726186,
+      "loss": 1.1294,
+      "step": 3090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001848312929232607,
+      "loss": 1.129,
+      "step": 3095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001851298895192595,
+      "loss": 1.0955,
+      "step": 3100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001854284861152583,
+      "loss": 1.1485,
+      "step": 3105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001857270827112571,
+      "loss": 1.0539,
+      "step": 3110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001860256793072559,
+      "loss": 1.1297,
+      "step": 3115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018632427590325472,
+      "loss": 1.0511,
+      "step": 3120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001866228724992535,
+      "loss": 1.1104,
+      "step": 3125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018692146909525233,
+      "loss": 1.1456,
+      "step": 3130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018722006569125112,
+      "loss": 1.0181,
+      "step": 3135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018751866228724993,
+      "loss": 1.1191,
+      "step": 3140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018781725888324875,
+      "loss": 1.0653,
+      "step": 3145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018811585547924754,
+      "loss": 1.2023,
+      "step": 3150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018841445207524635,
+      "loss": 1.0825,
+      "step": 3155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018871304867124514,
+      "loss": 1.0431,
+      "step": 3160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018901164526724396,
+      "loss": 1.1006,
+      "step": 3165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018931024186324277,
+      "loss": 1.149,
+      "step": 3170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001896088384592416,
+      "loss": 1.0379,
+      "step": 3175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018990743505524038,
+      "loss": 1.0628,
+      "step": 3180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001902060316512392,
+      "loss": 1.0216,
+      "step": 3185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019050462824723798,
+      "loss": 1.0309,
+      "step": 3190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001908032248432368,
+      "loss": 1.0195,
+      "step": 3195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019110182143923561,
+      "loss": 1.0972,
+      "step": 3200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001914004180352344,
+      "loss": 1.1071,
+      "step": 3205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019169901463123322,
+      "loss": 1.0252,
+      "step": 3210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000191997611227232,
+      "loss": 1.0535,
+      "step": 3215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019229620782323082,
+      "loss": 1.1252,
+      "step": 3220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019259480441922964,
+      "loss": 1.0827,
+      "step": 3225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019289340101522843,
+      "loss": 1.1525,
+      "step": 3230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019319199761122724,
+      "loss": 1.0793,
+      "step": 3235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019349059420722603,
+      "loss": 1.093,
+      "step": 3240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019378919080322487,
+      "loss": 1.1159,
+      "step": 3245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019408778739922366,
+      "loss": 1.076,
+      "step": 3250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019438638399522245,
+      "loss": 1.0385,
+      "step": 3255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019468498059122127,
+      "loss": 1.0041,
+      "step": 3260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019498357718722006,
+      "loss": 1.0801,
+      "step": 3265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001952821737832189,
+      "loss": 1.1181,
+      "step": 3270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001955807703792177,
+      "loss": 1.0675,
+      "step": 3275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001958793669752165,
+      "loss": 1.0912,
+      "step": 3280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001961779635712153,
+      "loss": 1.1491,
+      "step": 3285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001964765601672141,
+      "loss": 1.1827,
+      "step": 3290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001967751567632129,
+      "loss": 1.1387,
+      "step": 3295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001970737533592117,
+      "loss": 1.094,
+      "step": 3300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019737234995521053,
+      "loss": 1.0641,
+      "step": 3305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019767094655120932,
+      "loss": 1.0688,
+      "step": 3310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019796954314720813,
+      "loss": 1.0866,
+      "step": 3315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019826813974320692,
+      "loss": 1.0183,
+      "step": 3320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019856673633920574,
+      "loss": 1.0316,
+      "step": 3325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019886533293520455,
+      "loss": 1.0154,
+      "step": 3330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019916392953120334,
+      "loss": 1.1154,
+      "step": 3335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019946252612720216,
+      "loss": 1.0715,
+      "step": 3340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019976112272320095,
+      "loss": 1.1234,
+      "step": 3345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999999945680667,
+      "loss": 1.1382,
+      "step": 3350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999998044504044,
+      "loss": 1.1766,
+      "step": 3355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999999342736132,
+      "loss": 1.0361,
+      "step": 3360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999986094253744,
+      "loss": 1.0135,
+      "step": 3365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999976045183312,
+      "loss": 1.0417,
+      "step": 3370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999996328015275,
+      "loss": 1.0215,
+      "step": 3375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999994779916553,
+      "loss": 1.0643,
+      "step": 3380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999929602225854,
+      "loss": 1.2135,
+      "step": 3385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999908689338663,
+      "loss": 1.1343,
+      "step": 3390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999885060509635,
+      "loss": 1.0845,
+      "step": 3395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999858715745195,
+      "loss": 1.198,
+      "step": 3400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999982965505249,
+      "loss": 1.1071,
+      "step": 3405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999797878439415,
+      "loss": 1.1457,
+      "step": 3410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999763385914604,
+      "loss": 1.0965,
+      "step": 3415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999972617748742,
+      "loss": 1.1234,
+      "step": 3420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999686253167976,
+      "loss": 1.1075,
+      "step": 3425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999643612967106,
+      "loss": 1.1002,
+      "step": 3430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199995982568964,
+      "loss": 1.1558,
+      "step": 3435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999955018496817,
+      "loss": 1.2144,
+      "step": 3440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999949939719547,
+      "loss": 1.1135,
+      "step": 3445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999445893592106,
+      "loss": 1.1371,
+      "step": 3450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999389674172594,
+      "loss": 1.0312,
+      "step": 3455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999330738952216,
+      "loss": 1.1131,
+      "step": 3460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999269087946973,
+      "loss": 1.104,
+      "step": 3465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999204721173609,
+      "loss": 1.1234,
+      "step": 3470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199991376386496,
+      "loss": 1.1372,
+      "step": 3475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999906784039318,
+      "loss": 1.141,
+      "step": 3480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999899532642329,
+      "loss": 1.1108,
+      "step": 3485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019998920096759638,
+      "loss": 1.0414,
+      "step": 3490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019998842151422647,
+      "loss": 1.07,
+      "step": 3495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999876149043349,
+      "loss": 1.017,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019998678113814075,
+      "loss": 1.1036,
+      "step": 3505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019998592021587045,
+      "loss": 1.0656,
+      "step": 3510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019998503213775783,
+      "loss": 1.0716,
+      "step": 3515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998411690404412,
+      "loss": 1.0759,
+      "step": 3520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998317451497784,
+      "loss": 1.0861,
+      "step": 3525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998220497081498,
+      "loss": 1.0306,
+      "step": 3530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998120827181883,
+      "loss": 1.1173,
+      "step": 3535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998018441826014,
+      "loss": 1.0895,
+      "step": 3540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997913341041694,
+      "loss": 1.1571,
+      "step": 3545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999780552485747,
+      "loss": 1.0958,
+      "step": 3550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999769499330262,
+      "loss": 1.0831,
+      "step": 3555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997581746407174,
+      "loss": 1.1515,
+      "step": 3560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997465784201882,
+      "loss": 1.048,
+      "step": 3565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999734710671824,
+      "loss": 1.0868,
+      "step": 3570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999722571398848,
+      "loss": 1.0319,
+      "step": 3575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997101606045577,
+      "loss": 1.0128,
+      "step": 3580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999697478292323,
+      "loss": 1.0703,
+      "step": 3585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019996845244655892,
+      "loss": 1.1442,
+      "step": 3590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019996712991278735,
+      "loss": 1.1207,
+      "step": 3595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019996578022827687,
+      "loss": 1.047,
+      "step": 3600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019996440339339406,
+      "loss": 1.0814,
+      "step": 3605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019996299940851277,
+      "loss": 1.0791,
+      "step": 3610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019996156827401445,
+      "loss": 1.1105,
+      "step": 3615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019996010999028766,
+      "loss": 1.0932,
+      "step": 3620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019995862455772856,
+      "loss": 1.1267,
+      "step": 3625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019995711197674053,
+      "loss": 1.1348,
+      "step": 3630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999555722477344,
+      "loss": 1.0513,
+      "step": 3635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019995400537112836,
+      "loss": 1.0844,
+      "step": 3640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000199952411347348,
+      "loss": 1.1631,
+      "step": 3645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999507901768262,
+      "loss": 1.1233,
+      "step": 3650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019994914186000328,
+      "loss": 1.088,
+      "step": 3655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019994746639732692,
+      "loss": 1.0896,
+      "step": 3660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019994576378925215,
+      "loss": 1.1069,
+      "step": 3665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019994403403624146,
+      "loss": 1.1112,
+      "step": 3670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999422771387646,
+      "loss": 1.0745,
+      "step": 3675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999404930972987,
+      "loss": 1.1232,
+      "step": 3680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019993868191232834,
+      "loss": 1.0648,
+      "step": 3685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019993684358434545,
+      "loss": 1.171,
+      "step": 3690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019993497811384927,
+      "loss": 1.0275,
+      "step": 3695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999330855013465,
+      "loss": 1.1045,
+      "step": 3700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019993116574735113,
+      "loss": 1.1415,
+      "step": 3705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999292188523846,
+      "loss": 1.0976,
+      "step": 3710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999272448169756,
+      "loss": 1.133,
+      "step": 3715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019992524364166036,
+      "loss": 0.9992,
+      "step": 3720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019992321532698236,
+      "loss": 1.0332,
+      "step": 3725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999211598734925,
+      "loss": 1.1238,
+      "step": 3730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000199919077281749,
+      "loss": 1.0483,
+      "step": 3735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019991696755231747,
+      "loss": 1.0224,
+      "step": 3740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019991483068577097,
+      "loss": 1.0906,
+      "step": 3745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019991266668268983,
+      "loss": 1.1205,
+      "step": 3750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019991047554366177,
+      "loss": 1.0803,
+      "step": 3755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999082572692819,
+      "loss": 1.1077,
+      "step": 3760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019990601186015275,
+      "loss": 1.1253,
+      "step": 3765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999037393168841,
+      "loss": 1.1671,
+      "step": 3770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999014396400932,
+      "loss": 1.0707,
+      "step": 3775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019989911283040462,
+      "loss": 1.0694,
+      "step": 3780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019989675888845034,
+      "loss": 1.0888,
+      "step": 3785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019989437781486963,
+      "loss": 1.1157,
+      "step": 3790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019989196961030925,
+      "loss": 1.0365,
+      "step": 3795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001998895342754232,
+      "loss": 1.0746,
+      "step": 3800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019988707181087291,
+      "loss": 1.0803,
+      "step": 3805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019988458221732725,
+      "loss": 1.0568,
+      "step": 3810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001998820654954623,
+      "loss": 1.071,
+      "step": 3815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019987952164596164,
+      "loss": 1.0827,
+      "step": 3820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019987695066951618,
+      "loss": 1.0955,
+      "step": 3825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019987435256682413,
+      "loss": 1.0791,
+      "step": 3830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001998717273385912,
+      "loss": 1.0974,
+      "step": 3835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019986907498553032,
+      "loss": 1.0471,
+      "step": 3840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019986639550836191,
+      "loss": 1.0482,
+      "step": 3845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019986368890781368,
+      "loss": 1.0733,
+      "step": 3850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019986095518462077,
+      "loss": 1.0684,
+      "step": 3855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019985819433952564,
+      "loss": 1.1437,
+      "step": 3860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019985540637327808,
+      "loss": 1.1195,
+      "step": 3865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019985259128663536,
+      "loss": 1.1315,
+      "step": 3870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000199849749080362,
+      "loss": 1.098,
+      "step": 3875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019984687975522993,
+      "loss": 1.0512,
+      "step": 3880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998439833120185,
+      "loss": 1.0707,
+      "step": 3885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019984105975151434,
+      "loss": 1.0753,
+      "step": 3890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019983810907451145,
+      "loss": 1.1256,
+      "step": 3895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019983513128181128,
+      "loss": 1.0676,
+      "step": 3900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019983212637422258,
+      "loss": 0.9838,
+      "step": 3905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019982909435256144,
+      "loss": 1.0884,
+      "step": 3910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019982603521765134,
+      "loss": 1.1129,
+      "step": 3915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998229489703232,
+      "loss": 1.1245,
+      "step": 3920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019981983561141516,
+      "loss": 1.0162,
+      "step": 3925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019981669514177282,
+      "loss": 1.0832,
+      "step": 3930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019981352756224917,
+      "loss": 1.0793,
+      "step": 3935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019981033287370443,
+      "loss": 1.0764,
+      "step": 3940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019980711107700632,
+      "loss": 1.1139,
+      "step": 3945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019980386217302985,
+      "loss": 1.0912,
+      "step": 3950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019980058616265743,
+      "loss": 1.0632,
+      "step": 3955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019979728304677882,
+      "loss": 1.0528,
+      "step": 3960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997939528262911,
+      "loss": 1.0304,
+      "step": 3965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019979059550209876,
+      "loss": 1.1203,
+      "step": 3970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019978721107511365,
+      "loss": 1.1041,
+      "step": 3975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019978379954625495,
+      "loss": 1.1427,
+      "step": 3980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019978036091644925,
+      "loss": 1.007,
+      "step": 3985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019977689518663042,
+      "loss": 1.0742,
+      "step": 3990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019977340235773983,
+      "loss": 1.0635,
+      "step": 3995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019976988243072604,
+      "loss": 1.0774,
+      "step": 4000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019976633540654505,
+      "loss": 1.0696,
+      "step": 4005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019976276128616028,
+      "loss": 1.134,
+      "step": 4010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997591600705424,
+      "loss": 1.0233,
+      "step": 4015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997555317606695,
+      "loss": 1.0641,
+      "step": 4020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000199751876357527,
+      "loss": 0.9879,
+      "step": 4025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019974819386210774,
+      "loss": 1.0829,
+      "step": 4030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019974448427541184,
+      "loss": 1.0976,
+      "step": 4035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019974074759844678,
+      "loss": 1.089,
+      "step": 4040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019973698383222752,
+      "loss": 1.0393,
+      "step": 4045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997331929777762,
+      "loss": 1.0799,
+      "step": 4050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997293750361225,
+      "loss": 1.084,
+      "step": 4055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019972553000830324,
+      "loss": 1.0347,
+      "step": 4060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019972165789536276,
+      "loss": 1.1614,
+      "step": 4065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019971775869835276,
+      "loss": 1.0459,
+      "step": 4070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019971383241833222,
+      "loss": 1.0581,
+      "step": 4075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997098790563675,
+      "loss": 1.0846,
+      "step": 4080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019970589861353227,
+      "loss": 1.078,
+      "step": 4085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019970189109090773,
+      "loss": 1.1187,
+      "step": 4090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996978564895822,
+      "loss": 1.0696,
+      "step": 4095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019969379481065149,
+      "loss": 1.1331,
+      "step": 4100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019968970605521876,
+      "loss": 1.1067,
+      "step": 4105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019968559022439452,
+      "loss": 1.1005,
+      "step": 4110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019968144731929654,
+      "loss": 1.0483,
+      "step": 4115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019967727734105007,
+      "loss": 1.068,
+      "step": 4120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019967308029078767,
+      "loss": 1.0471,
+      "step": 4125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019966885616964924,
+      "loss": 1.0161,
+      "step": 4130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000199664604978782,
+      "loss": 1.1243,
+      "step": 4135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019966032671934062,
+      "loss": 1.1375,
+      "step": 4140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019965602139248703,
+      "loss": 1.0319,
+      "step": 4145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019965168899939052,
+      "loss": 1.1084,
+      "step": 4150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996473295412278,
+      "loss": 1.1189,
+      "step": 4155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019964294301918285,
+      "loss": 1.0556,
+      "step": 4160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019963852943444702,
+      "loss": 1.0722,
+      "step": 4165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996340887882191,
+      "loss": 1.0373,
+      "step": 4170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019962962108170508,
+      "loss": 1.0413,
+      "step": 4175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996251263161184,
+      "loss": 1.1025,
+      "step": 4180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996206044926798,
+      "loss": 1.1334,
+      "step": 4185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019961605561261744,
+      "loss": 1.0642,
+      "step": 4190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019961147967716673,
+      "loss": 1.065,
+      "step": 4195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019960687668757053,
+      "loss": 1.0835,
+      "step": 4200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019960224664507896,
+      "loss": 1.1132,
+      "step": 4205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019959758955094957,
+      "loss": 1.1363,
+      "step": 4210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019959290540644712,
+      "loss": 1.0778,
+      "step": 4215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995881942128439,
+      "loss": 1.1452,
+      "step": 4220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995834559714194,
+      "loss": 1.0822,
+      "step": 4225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019957869068346052,
+      "loss": 1.1601,
+      "step": 4230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995738983502615,
+      "loss": 1.0361,
+      "step": 4235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000199569078973124,
+      "loss": 1.0654,
+      "step": 4240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019956423255335678,
+      "loss": 1.1747,
+      "step": 4245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019955935909227624,
+      "loss": 1.0899,
+      "step": 4250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019955445859120595,
+      "loss": 1.0899,
+      "step": 4255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019954953105147692,
+      "loss": 1.1152,
+      "step": 4260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995445764744274,
+      "loss": 1.0925,
+      "step": 4265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019953959486140302,
+      "loss": 1.1296,
+      "step": 4270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995345862137568,
+      "loss": 1.0637,
+      "step": 4275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995295505328491,
+      "loss": 1.0797,
+      "step": 4280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019952448782004756,
+      "loss": 1.0704,
+      "step": 4285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995193980767272,
+      "loss": 1.1394,
+      "step": 4290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019951428130427038,
+      "loss": 1.0521,
+      "step": 4295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950913750406678,
+      "loss": 1.1602,
+      "step": 4300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950396667751344,
+      "loss": 1.1037,
+      "step": 4305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994987688260148,
+      "loss": 1.0254,
+      "step": 4310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019949354395098253,
+      "loss": 1.0934,
+      "step": 4315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948829205383568,
+      "loss": 1.1284,
+      "step": 4320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994830131360007,
+      "loss": 1.0931,
+      "step": 4325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994777071989112,
+      "loss": 1.105,
+      "step": 4330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994723742440084,
+      "loss": 1.0215,
+      "step": 4335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019946701427274067,
+      "loss": 1.1079,
+      "step": 4340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019946162728656374,
+      "loss": 1.0819,
+      "step": 4345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994562132869407,
+      "loss": 1.0461,
+      "step": 4350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019945077227534195,
+      "loss": 1.0445,
+      "step": 4355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994453042532453,
+      "loss": 1.1002,
+      "step": 4360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019943980922213582,
+      "loss": 1.1169,
+      "step": 4365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019943428718350591,
+      "loss": 1.0898,
+      "step": 4370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994287381388554,
+      "loss": 1.068,
+      "step": 4375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994231620896914,
+      "loss": 1.1344,
+      "step": 4380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019941755903752828,
+      "loss": 1.1067,
+      "step": 4385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019941192898388784,
+      "loss": 1.0948,
+      "step": 4390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994062719302992,
+      "loss": 1.1338,
+      "step": 4395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019940058787829877,
+      "loss": 0.9293,
+      "step": 4400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019939487682943033,
+      "loss": 1.0346,
+      "step": 4405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000199389138785245,
+      "loss": 1.0545,
+      "step": 4410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001993833737473012,
+      "loss": 1.0076,
+      "step": 4415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019937758171716468,
+      "loss": 1.0951,
+      "step": 4420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019937176269640855,
+      "loss": 1.044,
+      "step": 4425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019936591668661324,
+      "loss": 1.1229,
+      "step": 4430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001993600436893665,
+      "loss": 0.9646,
+      "step": 4435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019935414370626342,
+      "loss": 1.0925,
+      "step": 4440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019934821673890643,
+      "loss": 1.0375,
+      "step": 4445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019934226278890525,
+      "loss": 1.0375,
+      "step": 4450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019933628185787695,
+      "loss": 1.0886,
+      "step": 4455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019933027394744595,
+      "loss": 1.1145,
+      "step": 4460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000199324239059244,
+      "loss": 1.1474,
+      "step": 4465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019931817719491006,
+      "loss": 1.1015,
+      "step": 4470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019931208835609064,
+      "loss": 1.068,
+      "step": 4475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019930597254443937,
+      "loss": 1.1134,
+      "step": 4480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019929982976161728,
+      "loss": 1.0303,
+      "step": 4485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019929366000929276,
+      "loss": 1.0893,
+      "step": 4490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019928746328914148,
+      "loss": 1.0784,
+      "step": 4495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001992812396028465,
+      "loss": 1.071,
+      "step": 4500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019927498895209807,
+      "loss": 1.1054,
+      "step": 4505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019926871133859388,
+      "loss": 1.0855,
+      "step": 4510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019926240676403892,
+      "loss": 1.0519,
+      "step": 4515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019925607523014547,
+      "loss": 1.0823,
+      "step": 4520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019924971673863318,
+      "loss": 1.0435,
+      "step": 4525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019924333129122895,
+      "loss": 1.1264,
+      "step": 4530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019923691888966714,
+      "loss": 1.0691,
+      "step": 4535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019923047953568924,
+      "loss": 1.0732,
+      "step": 4540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019922401323104415,
+      "loss": 1.0695,
+      "step": 4545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001992175199774882,
+      "loss": 1.1293,
+      "step": 4550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019921099977678484,
+      "loss": 1.053,
+      "step": 4555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019920445263070494,
+      "loss": 1.0902,
+      "step": 4560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019919787854102676,
+      "loss": 1.1227,
+      "step": 4565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019919127750953573,
+      "loss": 1.0759,
+      "step": 4570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019918464953802472,
+      "loss": 1.0376,
+      "step": 4575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001991779946282938,
+      "loss": 1.0365,
+      "step": 4580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001991713127821505,
+      "loss": 1.0152,
+      "step": 4585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001991646040014095,
+      "loss": 1.0486,
+      "step": 4590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019915786828789297,
+      "loss": 1.0088,
+      "step": 4595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001991511056434302,
+      "loss": 1.0394,
+      "step": 4600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000199144316069858,
+      "loss": 1.1839,
+      "step": 4605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019913749956902038,
+      "loss": 1.0251,
+      "step": 4610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019913065614276862,
+      "loss": 1.0267,
+      "step": 4615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019912378579296146,
+      "loss": 1.0685,
+      "step": 4620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019911688852146476,
+      "loss": 1.1112,
+      "step": 4625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019910996433015189,
+      "loss": 1.013,
+      "step": 4630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019910301322090338,
+      "loss": 1.0462,
+      "step": 4635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019909603519560712,
+      "loss": 1.0048,
+      "step": 4640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019908903025615842,
+      "loss": 1.0672,
+      "step": 4645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019908199840445968,
+      "loss": 1.2213,
+      "step": 4650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019907493964242074,
+      "loss": 1.0137,
+      "step": 4655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001990678539719588,
+      "loss": 1.0608,
+      "step": 4660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019906074139499827,
+      "loss": 1.0901,
+      "step": 4665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019905360191347092,
+      "loss": 1.0626,
+      "step": 4670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001990464355293158,
+      "loss": 1.0526,
+      "step": 4675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001990392422444792,
+      "loss": 1.1368,
+      "step": 4680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019903202206091495,
+      "loss": 1.071,
+      "step": 4685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019902477498058387,
+      "loss": 1.0363,
+      "step": 4690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019901750100545434,
+      "loss": 1.0452,
+      "step": 4695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001990102001375019,
+      "loss": 1.204,
+      "step": 4700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001990028723787095,
+      "loss": 1.1102,
+      "step": 4705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019899551773106726,
+      "loss": 1.1238,
+      "step": 4710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019898813619657272,
+      "loss": 1.1649,
+      "step": 4715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989807277772307,
+      "loss": 1.1148,
+      "step": 4720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019897329247505323,
+      "loss": 1.0787,
+      "step": 4725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989658302920598,
+      "loss": 1.0655,
+      "step": 4730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019895834123027703,
+      "loss": 1.0368,
+      "step": 4735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000198950825291739,
+      "loss": 1.0231,
+      "step": 4740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019894328247848698,
+      "loss": 1.1136,
+      "step": 4745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019893571279256956,
+      "loss": 1.0852,
+      "step": 4750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019892811623604267,
+      "loss": 1.0852,
+      "step": 4755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019892049281096944,
+      "loss": 1.0646,
+      "step": 4760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019891284251942048,
+      "loss": 1.059,
+      "step": 4765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019890516536347352,
+      "loss": 1.0181,
+      "step": 4770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019889746134521365,
+      "loss": 1.0821,
+      "step": 4775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019888973046673324,
+      "loss": 1.0235,
+      "step": 4780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000198881972730132,
+      "loss": 1.0848,
+      "step": 4785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001988741881375169,
+      "loss": 1.1709,
+      "step": 4790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019886637669100222,
+      "loss": 1.1003,
+      "step": 4795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019885853839270948,
+      "loss": 1.0877,
+      "step": 4800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001988506732447676,
+      "loss": 1.0929,
+      "step": 4805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019884278124931266,
+      "loss": 1.1319,
+      "step": 4810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019883486240848816,
+      "loss": 0.9905,
+      "step": 4815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019882691672444476,
+      "loss": 1.1209,
+      "step": 4820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019881894419934056,
+      "loss": 1.0181,
+      "step": 4825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019881094483534083,
+      "loss": 1.0518,
+      "step": 4830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019880291863461817,
+      "loss": 1.0303,
+      "step": 4835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019879486559935247,
+      "loss": 1.0751,
+      "step": 4840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001987867857317309,
+      "loss": 1.0305,
+      "step": 4845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019877867903394795,
+      "loss": 1.052,
+      "step": 4850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001987705455082054,
+      "loss": 1.1089,
+      "step": 4855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987623851567122,
+      "loss": 1.1282,
+      "step": 4860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019875419798168472,
+      "loss": 1.025,
+      "step": 4865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019874598398534658,
+      "loss": 1.0621,
+      "step": 4870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987377431699287,
+      "loss": 1.0967,
+      "step": 4875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987294755376692,
+      "loss": 1.1446,
+      "step": 4880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019872118109081355,
+      "loss": 1.0542,
+      "step": 4885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019871285983161452,
+      "loss": 1.0261,
+      "step": 4890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019870451176233212,
+      "loss": 1.0884,
+      "step": 4895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019869613688523367,
+      "loss": 1.0708,
+      "step": 4900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019868773520259375,
+      "loss": 1.0196,
+      "step": 4905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019867930671669422,
+      "loss": 1.0376,
+      "step": 4910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019867085142982426,
+      "loss": 1.1106,
+      "step": 4915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019866236934428026,
+      "loss": 0.9639,
+      "step": 4920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019865386046236596,
+      "loss": 1.047,
+      "step": 4925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019864532478639234,
+      "loss": 1.0621,
+      "step": 4930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001986367623186776,
+      "loss": 1.0867,
+      "step": 4935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019862817306154736,
+      "loss": 1.1177,
+      "step": 4940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001986195570173344,
+      "loss": 1.1182,
+      "step": 4945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019861091418837878,
+      "loss": 1.1056,
+      "step": 4950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001986022445770279,
+      "loss": 1.0565,
+      "step": 4955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019859354818563643,
+      "loss": 1.0851,
+      "step": 4960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001985848250165662,
+      "loss": 1.0588,
+      "step": 4965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019857607507218648,
+      "loss": 1.0763,
+      "step": 4970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019856729835487363,
+      "loss": 1.0228,
+      "step": 4975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019855849486701143,
+      "loss": 1.0125,
+      "step": 4980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001985496646109909,
+      "loss": 1.127,
+      "step": 4985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019854080758921025,
+      "loss": 1.0706,
+      "step": 4990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001985319238040751,
+      "loss": 1.0399,
+      "step": 4995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019852301325799817,
+      "loss": 1.0319,
+      "step": 5000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019851407595339958,
+      "loss": 1.0729,
+      "step": 5005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019850511189270663,
+      "loss": 1.1789,
+      "step": 5010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000198496121078354,
+      "loss": 1.0771,
+      "step": 5015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019848710351278354,
+      "loss": 1.0313,
+      "step": 5020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019847805919844436,
+      "loss": 1.0176,
+      "step": 5025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001984689881377929,
+      "loss": 1.1203,
+      "step": 5030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019845989033329277,
+      "loss": 1.0339,
+      "step": 5035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000198450765787415,
+      "loss": 1.0743,
+      "step": 5040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001984416145026377,
+      "loss": 1.0582,
+      "step": 5045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019843243648144635,
+      "loss": 1.0595,
+      "step": 5050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001984232317263337,
+      "loss": 1.0853,
+      "step": 5055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019841400023979975,
+      "loss": 1.1285,
+      "step": 5060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019840474202435164,
+      "loss": 1.1024,
+      "step": 5065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019839545708250398,
+      "loss": 1.0348,
+      "step": 5070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019838614541677844,
+      "loss": 1.0373,
+      "step": 5075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001983768070297041,
+      "loss": 1.1146,
+      "step": 5080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019836744192381725,
+      "loss": 1.1097,
+      "step": 5085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019835805010166136,
+      "loss": 1.0626,
+      "step": 5090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019834863156578724,
+      "loss": 0.9856,
+      "step": 5095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019833918631875292,
+      "loss": 1.1017,
+      "step": 5100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019832971436312374,
+      "loss": 1.0223,
+      "step": 5105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001983202157014722,
+      "loss": 1.0143,
+      "step": 5110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019831069033637817,
+      "loss": 1.0442,
+      "step": 5115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019830113827042865,
+      "loss": 1.1573,
+      "step": 5120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019829155950621796,
+      "loss": 1.0452,
+      "step": 5125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001982819540463477,
+      "loss": 1.1239,
+      "step": 5130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001982723218934266,
+      "loss": 1.0805,
+      "step": 5135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001982626630500708,
+      "loss": 1.1184,
+      "step": 5140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019825297751890358,
+      "loss": 0.976,
+      "step": 5145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019824326530255547,
+      "loss": 1.0452,
+      "step": 5150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019823352640366434,
+      "loss": 1.0741,
+      "step": 5155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019822376082487516,
+      "loss": 1.0332,
+      "step": 5160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019821396856884032,
+      "loss": 1.102,
+      "step": 5165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001982041496382193,
+      "loss": 1.1308,
+      "step": 5170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019819430403567888,
+      "loss": 1.0755,
+      "step": 5175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019818443176389315,
+      "loss": 1.1165,
+      "step": 5180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019817453282554333,
+      "loss": 1.129,
+      "step": 5185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000198164607223318,
+      "loss": 1.1256,
+      "step": 5190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019815465495991288,
+      "loss": 1.0689,
+      "step": 5195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019814467603803095,
+      "loss": 1.0428,
+      "step": 5200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001981346704603825,
+      "loss": 1.1491,
+      "step": 5205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019812463822968497,
+      "loss": 1.1501,
+      "step": 5210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001981145793486631,
+      "loss": 1.0981,
+      "step": 5215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019810449382004887,
+      "loss": 1.069,
+      "step": 5220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019809438164658145,
+      "loss": 1.0498,
+      "step": 5225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019808424283100726,
+      "loss": 1.0319,
+      "step": 5230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019807407737608,
+      "loss": 1.1154,
+      "step": 5235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019806388528456056,
+      "loss": 1.0625,
+      "step": 5240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019805366655921707,
+      "loss": 1.1906,
+      "step": 5245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001980434212028249,
+      "loss": 1.0769,
+      "step": 5250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019803314921816668,
+      "loss": 1.0853,
+      "step": 5255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001980228506080322,
+      "loss": 1.0669,
+      "step": 5260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001980125253752186,
+      "loss": 1.0237,
+      "step": 5265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019800217352253013,
+      "loss": 1.0978,
+      "step": 5270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001979917950527783,
+      "loss": 1.0863,
+      "step": 5275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019798138996878192,
+      "loss": 1.0266,
+      "step": 5280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001979709582733669,
+      "loss": 1.0415,
+      "step": 5285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019796049996936656,
+      "loss": 1.021,
+      "step": 5290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019795001505962126,
+      "loss": 1.0419,
+      "step": 5295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001979395035469787,
+      "loss": 1.0269,
+      "step": 5300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019792896543429375,
+      "loss": 1.0916,
+      "step": 5305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019791840072442855,
+      "loss": 1.1082,
+      "step": 5310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001979078094202524,
+      "loss": 1.1307,
+      "step": 5315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019789719152464192,
+      "loss": 1.0728,
+      "step": 5320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019788654704048088,
+      "loss": 1.0874,
+      "step": 5325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019787587597066024,
+      "loss": 1.1152,
+      "step": 5330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019786517831807827,
+      "loss": 1.0201,
+      "step": 5335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001978544540856404,
+      "loss": 1.1357,
+      "step": 5340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019784370327625931,
+      "loss": 1.1342,
+      "step": 5345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001978329258928549,
+      "loss": 0.9875,
+      "step": 5350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019782212193835422,
+      "loss": 1.1138,
+      "step": 5355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001978112914156916,
+      "loss": 1.0139,
+      "step": 5360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001978004343278086,
+      "loss": 1.0608,
+      "step": 5365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019778955067765398,
+      "loss": 1.0497,
+      "step": 5370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019777864046818366,
+      "loss": 1.1149,
+      "step": 5375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019776770370236082,
+      "loss": 1.1508,
+      "step": 5380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001977567403831559,
+      "loss": 1.1393,
+      "step": 5385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019774575051354645,
+      "loss": 0.991,
+      "step": 5390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001977347340965173,
+      "loss": 1.0558,
+      "step": 5395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001977236911350605,
+      "loss": 1.1075,
+      "step": 5400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001977126216321752,
+      "loss": 1.0433,
+      "step": 5405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019770152559086794,
+      "loss": 1.1164,
+      "step": 5410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019769040301415228,
+      "loss": 1.0112,
+      "step": 5415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019767925390504914,
+      "loss": 1.0151,
+      "step": 5420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019766807826658656,
+      "loss": 1.058,
+      "step": 5425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001976568761017998,
+      "loss": 1.0654,
+      "step": 5430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019764564741373133,
+      "loss": 1.072,
+      "step": 5435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019763439220543084,
+      "loss": 1.0295,
+      "step": 5440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019762311047995517,
+      "loss": 1.1007,
+      "step": 5445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019761180224036844,
+      "loss": 1.063,
+      "step": 5450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001976004674897419,
+      "loss": 1.0566,
+      "step": 5455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019758910623115405,
+      "loss": 1.073,
+      "step": 5460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019757771846769058,
+      "loss": 1.0545,
+      "step": 5465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019756630420244433,
+      "loss": 1.0589,
+      "step": 5470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019755486343851541,
+      "loss": 1.0361,
+      "step": 5475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019754339617901107,
+      "loss": 1.0427,
+      "step": 5480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001975319024270458,
+      "loss": 1.0694,
+      "step": 5485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019752038218574124,
+      "loss": 1.0612,
+      "step": 5490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001975088354582263,
+      "loss": 1.0657,
+      "step": 5495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019749726224763697,
+      "loss": 1.1525,
+      "step": 5500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019748566255711652,
+      "loss": 1.1052,
+      "step": 5505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001974740363898154,
+      "loss": 1.0906,
+      "step": 5510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019746238374889125,
+      "loss": 1.0021,
+      "step": 5515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019745070463750885,
+      "loss": 1.1803,
+      "step": 5520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019743899905884022,
+      "loss": 1.0912,
+      "step": 5525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019742726701606458,
+      "loss": 1.0049,
+      "step": 5530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019741550851236825,
+      "loss": 1.138,
+      "step": 5535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001974037235509449,
+      "loss": 1.0151,
+      "step": 5540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001973919121349952,
+      "loss": 1.0765,
+      "step": 5545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001973800742677271,
+      "loss": 1.0131,
+      "step": 5550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001973682099523558,
+      "loss": 1.0848,
+      "step": 5555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001973563191921035,
+      "loss": 1.0464,
+      "step": 5560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019734440199019982,
+      "loss": 0.9965,
+      "step": 5565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001973324583498813,
+      "loss": 1.1585,
+      "step": 5570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019732048827439188,
+      "loss": 1.0624,
+      "step": 5575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019730849176698254,
+      "loss": 1.022,
+      "step": 5580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019729646883091153,
+      "loss": 1.0323,
+      "step": 5585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019728441946944422,
+      "loss": 1.1031,
+      "step": 5590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019727234368585318,
+      "loss": 1.0758,
+      "step": 5595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019726024148341816,
+      "loss": 1.099,
+      "step": 5600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001972481128654261,
+      "loss": 1.1152,
+      "step": 5605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019723595783517102,
+      "loss": 1.0904,
+      "step": 5610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019722377639595423,
+      "loss": 1.0967,
+      "step": 5615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001972115685510842,
+      "loss": 1.0337,
+      "step": 5620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001971993343038765,
+      "loss": 1.1187,
+      "step": 5625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001971870736576539,
+      "loss": 1.0138,
+      "step": 5630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019717478661574638,
+      "loss": 1.0121,
+      "step": 5635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019716247318149103,
+      "loss": 1.084,
+      "step": 5640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001971501333582322,
+      "loss": 1.1152,
+      "step": 5645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019713776714932126,
+      "loss": 1.0011,
+      "step": 5650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019712537455811687,
+      "loss": 1.0922,
+      "step": 5655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019711295558798487,
+      "loss": 1.0545,
+      "step": 5660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001971005102422981,
+      "loss": 1.0796,
+      "step": 5665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019708803852443678,
+      "loss": 1.1074,
+      "step": 5670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019707554043778811,
+      "loss": 1.1439,
+      "step": 5675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001970630159857466,
+      "loss": 0.9909,
+      "step": 5680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019705046517171378,
+      "loss": 1.0679,
+      "step": 5685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001970378879990984,
+      "loss": 1.0229,
+      "step": 5690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019702528447131646,
+      "loss": 1.0642,
+      "step": 5695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000197012654591791,
+      "loss": 1.0415,
+      "step": 5700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019699999836395225,
+      "loss": 1.1193,
+      "step": 5705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019698731579123754,
+      "loss": 1.1045,
+      "step": 5710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019697460687709154,
+      "loss": 1.0317,
+      "step": 5715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019696187162496583,
+      "loss": 1.0721,
+      "step": 5720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019694911003831933,
+      "loss": 1.0118,
+      "step": 5725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000196936322120618,
+      "loss": 1.0759,
+      "step": 5730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019692350787533505,
+      "loss": 1.0189,
+      "step": 5735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019691066730595077,
+      "loss": 1.1207,
+      "step": 5740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019689780041595256,
+      "loss": 1.0603,
+      "step": 5745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001968849072088351,
+      "loss": 0.9801,
+      "step": 5750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019687198768810008,
+      "loss": 1.0711,
+      "step": 5755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019685904185725645,
+      "loss": 1.0938,
+      "step": 5760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019684606971982023,
+      "loss": 1.0413,
+      "step": 5765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019683307127931458,
+      "loss": 1.0297,
+      "step": 5770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019682004653926992,
+      "loss": 1.0707,
+      "step": 5775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001968069955032236,
+      "loss": 1.1785,
+      "step": 5780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019679391817472035,
+      "loss": 1.0661,
+      "step": 5785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019678081455731187,
+      "loss": 1.1462,
+      "step": 5790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001967676846545571,
+      "loss": 1.1029,
+      "step": 5795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019675452847002205,
+      "loss": 1.077,
+      "step": 5800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019674134600727988,
+      "loss": 1.1297,
+      "step": 5805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019672813726991092,
+      "loss": 1.0142,
+      "step": 5810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019671490226150264,
+      "loss": 1.0151,
+      "step": 5815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001967016409856496,
+      "loss": 1.0813,
+      "step": 5820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019668835344595353,
+      "loss": 1.0649,
+      "step": 5825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019667503964602325,
+      "loss": 1.0729,
+      "step": 5830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001966616995894748,
+      "loss": 1.084,
+      "step": 5835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019664833327993125,
+      "loss": 1.1161,
+      "step": 5840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019663494072102284,
+      "loss": 1.062,
+      "step": 5845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000196621521916387,
+      "loss": 1.1028,
+      "step": 5850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019660807686966818,
+      "loss": 1.1866,
+      "step": 5855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000196594605584518,
+      "loss": 1.0915,
+      "step": 5860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019658110806459526,
+      "loss": 1.0741,
+      "step": 5865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019656758431356582,
+      "loss": 0.9842,
+      "step": 5870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019655403433510268,
+      "loss": 1.0389,
+      "step": 5875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000196540458132886,
+      "loss": 0.9803,
+      "step": 5880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000196526855710603,
+      "loss": 1.0118,
+      "step": 5885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019651322707194804,
+      "loss": 1.0507,
+      "step": 5890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019649957222062266,
+      "loss": 1.0712,
+      "step": 5895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019648589116033543,
+      "loss": 0.9621,
+      "step": 5900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001964721838948021,
+      "loss": 1.0436,
+      "step": 5905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019645845042774553,
+      "loss": 1.1087,
+      "step": 5910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019644469076289563,
+      "loss": 1.074,
+      "step": 5915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019643090490398954,
+      "loss": 1.0794,
+      "step": 5920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019641709285477145,
+      "loss": 1.0439,
+      "step": 5925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019640325461899263,
+      "loss": 1.1111,
+      "step": 5930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001963893902004115,
+      "loss": 1.0678,
+      "step": 5935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001963754996027936,
+      "loss": 1.0038,
+      "step": 5940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001963615828299116,
+      "loss": 1.0867,
+      "step": 5945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019634763988554522,
+      "loss": 1.0338,
+      "step": 5950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001963336707734813,
+      "loss": 1.1219,
+      "step": 5955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019631967549751386,
+      "loss": 1.1158,
+      "step": 5960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001963056540614439,
+      "loss": 1.0276,
+      "step": 5965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019629160646907965,
+      "loss": 1.0499,
+      "step": 5970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019627753272423638,
+      "loss": 0.9816,
+      "step": 5975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019626343283073644,
+      "loss": 1.0808,
+      "step": 5980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019624930679240933,
+      "loss": 1.0684,
+      "step": 5985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019623515461309165,
+      "loss": 1.0807,
+      "step": 5990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019622097629662708,
+      "loss": 1.0857,
+      "step": 5995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001962067718468664,
+      "loss": 1.107,
+      "step": 6000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001961925412676675,
+      "loss": 1.0473,
+      "step": 6005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019617828456289533,
+      "loss": 1.1362,
+      "step": 6010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000196164001736422,
+      "loss": 0.9959,
+      "step": 6015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019614969279212665,
+      "loss": 1.0562,
+      "step": 6020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019613535773389554,
+      "loss": 1.1135,
+      "step": 6025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019612099656562203,
+      "loss": 1.0536,
+      "step": 6030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019610660929120658,
+      "loss": 1.0403,
+      "step": 6035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019609219591455675,
+      "loss": 1.1076,
+      "step": 6040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001960777564395871,
+      "loss": 1.1449,
+      "step": 6045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019606329087021936,
+      "loss": 1.0475,
+      "step": 6050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019604879921038238,
+      "loss": 1.0646,
+      "step": 6055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019603428146401197,
+      "loss": 1.1391,
+      "step": 6060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019601973763505118,
+      "loss": 1.109,
+      "step": 6065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019600516772745,
+      "loss": 1.035,
+      "step": 6070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001959905717451656,
+      "loss": 1.0514,
+      "step": 6075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001959759496921622,
+      "loss": 1.0511,
+      "step": 6080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019596130157241112,
+      "loss": 1.0169,
+      "step": 6085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001959466273898907,
+      "loss": 1.0509,
+      "step": 6090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001959319271485864,
+      "loss": 1.1169,
+      "step": 6095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019591720085249082,
+      "loss": 1.0444,
+      "step": 6100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019590244850560347,
+      "loss": 1.1796,
+      "step": 6105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019588767011193112,
+      "loss": 1.0999,
+      "step": 6110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019587286567548752,
+      "loss": 1.0524,
+      "step": 6115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019585803520029348,
+      "loss": 1.0346,
+      "step": 6120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001958431786903769,
+      "loss": 1.086,
+      "step": 6125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019582829614977283,
+      "loss": 1.0802,
+      "step": 6130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001958133875825232,
+      "loss": 1.0511,
+      "step": 6135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019579845299267726,
+      "loss": 1.1482,
+      "step": 6140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019578349238429107,
+      "loss": 1.1057,
+      "step": 6145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000195768505761428,
+      "loss": 0.996,
+      "step": 6150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019575349312815824,
+      "loss": 1.032,
+      "step": 6155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001957384544885593,
+      "loss": 1.0306,
+      "step": 6160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019572338984671553,
+      "loss": 1.1493,
+      "step": 6165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019570829920671846,
+      "loss": 1.0969,
+      "step": 6170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019569318257266666,
+      "loss": 1.094,
+      "step": 6175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001956780399486658,
+      "loss": 1.1324,
+      "step": 6180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019566287133882852,
+      "loss": 1.0979,
+      "step": 6185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019564767674727456,
+      "loss": 1.01,
+      "step": 6190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019563245617813072,
+      "loss": 1.0648,
+      "step": 6195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001956172096355309,
+      "loss": 1.0528,
+      "step": 6200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019560193712361596,
+      "loss": 1.0651,
+      "step": 6205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019558663864653388,
+      "loss": 1.1136,
+      "step": 6210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001955713142084397,
+      "loss": 1.1032,
+      "step": 6215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019555596381349547,
+      "loss": 1.0741,
+      "step": 6220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019554058746587026,
+      "loss": 1.0934,
+      "step": 6225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001955251851697403,
+      "loss": 1.1092,
+      "step": 6230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001955097569292888,
+      "loss": 1.0315,
+      "step": 6235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019549430274870596,
+      "loss": 1.0634,
+      "step": 6240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019547882263218914,
+      "loss": 1.1065,
+      "step": 6245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019546331658394265,
+      "loss": 1.0974,
+      "step": 6250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019544778460817792,
+      "loss": 1.0553,
+      "step": 6255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019543222670911336,
+      "loss": 1.0858,
+      "step": 6260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001954166428909744,
+      "loss": 1.0592,
+      "step": 6265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019540103315799368,
+      "loss": 1.0981,
+      "step": 6270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019538539751441062,
+      "loss": 1.1368,
+      "step": 6275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019536973596447185,
+      "loss": 1.0476,
+      "step": 6280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000195354048512431,
+      "loss": 1.075,
+      "step": 6285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019533833516254876,
+      "loss": 1.0192,
+      "step": 6290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001953225959190928,
+      "loss": 1.0262,
+      "step": 6295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001953068307863378,
+      "loss": 1.0238,
+      "step": 6300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001952910397685656,
+      "loss": 0.9787,
+      "step": 6305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019527522287006495,
+      "loss": 0.9966,
+      "step": 6310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019525938009513162,
+      "loss": 1.1149,
+      "step": 6315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019524351144806855,
+      "loss": 1.059,
+      "step": 6320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019522761693318552,
+      "loss": 1.101,
+      "step": 6325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001952116965547995,
+      "loss": 1.1542,
+      "step": 6330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019519575031723437,
+      "loss": 1.0849,
+      "step": 6335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001951797782248211,
+      "loss": 1.0792,
+      "step": 6340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001951637802818976,
+      "loss": 1.0201,
+      "step": 6345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001951477564928089,
+      "loss": 1.0709,
+      "step": 6350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019513170686190706,
+      "loss": 1.069,
+      "step": 6355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019511563139355099,
+      "loss": 1.0179,
+      "step": 6360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019509953009210683,
+      "loss": 1.0761,
+      "step": 6365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001950834029619476,
+      "loss": 1.1206,
+      "step": 6370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019506725000745337,
+      "loss": 1.0902,
+      "step": 6375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019505107123301128,
+      "loss": 1.0567,
+      "step": 6380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019503486664301534,
+      "loss": 0.9986,
+      "step": 6385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019501863624186676,
+      "loss": 1.1395,
+      "step": 6390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019500238003397358,
+      "loss": 1.0032,
+      "step": 6395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019498609802375097,
+      "loss": 0.9435,
+      "step": 6400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001949697902156211,
+      "loss": 1.0698,
+      "step": 6405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019495345661401305,
+      "loss": 1.175,
+      "step": 6410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019493709722336302,
+      "loss": 1.1948,
+      "step": 6415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019492071204811414,
+      "loss": 1.0924,
+      "step": 6420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001949043010927166,
+      "loss": 1.0597,
+      "step": 6425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019488786436162753,
+      "loss": 1.1123,
+      "step": 6430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001948714018593111,
+      "loss": 1.0924,
+      "step": 6435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019485491359023848,
+      "loss": 1.1378,
+      "step": 6440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019483839955888784,
+      "loss": 1.1007,
+      "step": 6445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001948218597697443,
+      "loss": 1.0708,
+      "step": 6450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019480529422730004,
+      "loss": 1.1415,
+      "step": 6455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019478870293605416,
+      "loss": 0.9645,
+      "step": 6460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019477208590051287,
+      "loss": 1.1225,
+      "step": 6465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019475544312518923,
+      "loss": 1.0065,
+      "step": 6470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019473877461460344,
+      "loss": 1.0561,
+      "step": 6475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019472208037328256,
+      "loss": 1.0846,
+      "step": 6480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019470536040576067,
+      "loss": 1.1218,
+      "step": 6485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001946886147165789,
+      "loss": 1.1326,
+      "step": 6490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019467184331028534,
+      "loss": 1.143,
+      "step": 6495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000194655046191435,
+      "loss": 0.9918,
+      "step": 6500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019463822336458994,
+      "loss": 1.0854,
+      "step": 6505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001946213748343192,
+      "loss": 1.1453,
+      "step": 6510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019460450060519876,
+      "loss": 1.0637,
+      "step": 6515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001945876006818116,
+      "loss": 1.0748,
+      "step": 6520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001945706750687477,
+      "loss": 1.0499,
+      "step": 6525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000194553723770604,
+      "loss": 1.0572,
+      "step": 6530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001945367467919844,
+      "loss": 1.0715,
+      "step": 6535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019451974413749981,
+      "loss": 1.0852,
+      "step": 6540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001945027158117681,
+      "loss": 1.1172,
+      "step": 6545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001944856618194141,
+      "loss": 1.0434,
+      "step": 6550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001944685821650696,
+      "loss": 1.1306,
+      "step": 6555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019445147685337338,
+      "loss": 1.081,
+      "step": 6560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001944343458889712,
+      "loss": 0.9441,
+      "step": 6565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019441718927651577,
+      "loss": 1.0113,
+      "step": 6570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019440000702066677,
+      "loss": 1.0873,
+      "step": 6575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019438279912609082,
+      "loss": 1.0596,
+      "step": 6580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019436556559746158,
+      "loss": 1.0867,
+      "step": 6585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019434830643945953,
+      "loss": 1.0603,
+      "step": 6590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019433102165677228,
+      "loss": 1.1341,
+      "step": 6595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001943137112540943,
+      "loss": 1.1013,
+      "step": 6600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019429637523612705,
+      "loss": 1.0778,
+      "step": 6605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001942790136075789,
+      "loss": 0.9962,
+      "step": 6610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001942616263731652,
+      "loss": 1.0115,
+      "step": 6615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019424421353760832,
+      "loss": 1.0727,
+      "step": 6620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019422677510563748,
+      "loss": 1.1123,
+      "step": 6625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019420931108198893,
+      "loss": 1.0533,
+      "step": 6630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019419182147140582,
+      "loss": 1.1516,
+      "step": 6635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019417430627863828,
+      "loss": 1.0532,
+      "step": 6640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001941567655084434,
+      "loss": 1.1083,
+      "step": 6645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019413919916558514,
+      "loss": 1.0768,
+      "step": 6650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001941216072548345,
+      "loss": 1.0665,
+      "step": 6655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001941039897809694,
+      "loss": 0.9859,
+      "step": 6660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019408634674877463,
+      "loss": 1.0962,
+      "step": 6665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019406867816304204,
+      "loss": 1.0535,
+      "step": 6670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019405098402857033,
+      "loss": 1.1598,
+      "step": 6675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019403326435016513,
+      "loss": 1.1472,
+      "step": 6680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001940155191326391,
+      "loss": 1.0788,
+      "step": 6685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019399774838081178,
+      "loss": 1.0634,
+      "step": 6690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019397995209950963,
+      "loss": 1.0429,
+      "step": 6695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019396213029356608,
+      "loss": 1.0056,
+      "step": 6700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019394428296782144,
+      "loss": 1.1964,
+      "step": 6705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000193926410127123,
+      "loss": 1.0845,
+      "step": 6710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019390851177632497,
+      "loss": 1.1189,
+      "step": 6715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019389058792028846,
+      "loss": 1.005,
+      "step": 6720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019387263856388156,
+      "loss": 1.1551,
+      "step": 6725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019385466371197923,
+      "loss": 1.0764,
+      "step": 6730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001938366633694634,
+      "loss": 1.052,
+      "step": 6735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019381863754122292,
+      "loss": 1.1638,
+      "step": 6740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001938005862321535,
+      "loss": 1.1142,
+      "step": 6745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019378250944715784,
+      "loss": 1.1115,
+      "step": 6750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001937644071911455,
+      "loss": 1.0206,
+      "step": 6755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019374627946903305,
+      "loss": 1.0454,
+      "step": 6760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019372812628574385,
+      "loss": 1.0975,
+      "step": 6765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019370994764620833,
+      "loss": 1.0171,
+      "step": 6770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001936917435553637,
+      "loss": 1.0474,
+      "step": 6775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019367351401815413,
+      "loss": 1.0654,
+      "step": 6780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019365525903953067,
+      "loss": 1.0801,
+      "step": 6785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019363697862445137,
+      "loss": 1.0106,
+      "step": 6790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019361867277788112,
+      "loss": 1.0548,
+      "step": 6795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001936003415047917,
+      "loss": 0.9941,
+      "step": 6800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019358198481016182,
+      "loss": 1.0454,
+      "step": 6805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019356360269897713,
+      "loss": 1.1103,
+      "step": 6810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019354519517623012,
+      "loss": 1.0954,
+      "step": 6815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019352676224692022,
+      "loss": 1.0873,
+      "step": 6820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019350830391605376,
+      "loss": 1.1164,
+      "step": 6825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019348982018864397,
+      "loss": 1.0316,
+      "step": 6830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019347131106971094,
+      "loss": 1.0965,
+      "step": 6835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001934527765642817,
+      "loss": 1.111,
+      "step": 6840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001934342166773902,
+      "loss": 1.1366,
+      "step": 6845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019341563141407714,
+      "loss": 1.0946,
+      "step": 6850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001933970207793903,
+      "loss": 1.0986,
+      "step": 6855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019337838477838428,
+      "loss": 1.0962,
+      "step": 6860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019335972341612048,
+      "loss": 1.0555,
+      "step": 6865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001933410366976673,
+      "loss": 1.0412,
+      "step": 6870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019332232462810004,
+      "loss": 1.0282,
+      "step": 6875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019330358721250074,
+      "loss": 1.1347,
+      "step": 6880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001932848244559585,
+      "loss": 1.0403,
+      "step": 6885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019326603636356918,
+      "loss": 1.0159,
+      "step": 6890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019324722294043558,
+      "loss": 1.0784,
+      "step": 6895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019322838419166737,
+      "loss": 1.1826,
+      "step": 6900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019320952012238108,
+      "loss": 1.033,
+      "step": 6905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001931906307377001,
+      "loss": 1.1023,
+      "step": 6910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019317171604275478,
+      "loss": 1.1325,
+      "step": 6915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019315277604268225,
+      "loss": 1.1143,
+      "step": 6920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019313381074262655,
+      "loss": 1.0861,
+      "step": 6925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019311482014773863,
+      "loss": 1.0554,
+      "step": 6930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019309580426317622,
+      "loss": 1.0179,
+      "step": 6935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000193076763094104,
+      "loss": 1.1075,
+      "step": 6940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019305769664569348,
+      "loss": 1.0568,
+      "step": 6945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019303860492312307,
+      "loss": 1.1335,
+      "step": 6950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019301948793157796,
+      "loss": 1.0968,
+      "step": 6955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019300034567625032,
+      "loss": 1.0492,
+      "step": 6960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001929811781623391,
+      "loss": 1.0504,
+      "step": 6965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019296198539505013,
+      "loss": 1.0451,
+      "step": 6970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001929427673795961,
+      "loss": 1.0916,
+      "step": 6975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019292352412119657,
+      "loss": 1.0767,
+      "step": 6980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019290425562507792,
+      "loss": 1.044,
+      "step": 6985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019288496189647346,
+      "loss": 1.048,
+      "step": 6990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019286564294062323,
+      "loss": 1.1075,
+      "step": 6995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019284629876277425,
+      "loss": 1.0377,
+      "step": 7000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019282692936818031,
+      "loss": 1.0853,
+      "step": 7005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001928075347621021,
+      "loss": 1.0473,
+      "step": 7010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001927881149498071,
+      "loss": 1.079,
+      "step": 7015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001927686699365697,
+      "loss": 1.0701,
+      "step": 7020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019274919972767104,
+      "loss": 1.0047,
+      "step": 7025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019272970432839922,
+      "loss": 1.0778,
+      "step": 7030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001927101837440491,
+      "loss": 1.0638,
+      "step": 7035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019269063797992242,
+      "loss": 1.1327,
+      "step": 7040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019267106704132773,
+      "loss": 1.0794,
+      "step": 7045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019265147093358044,
+      "loss": 0.9926,
+      "step": 7050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019263184966200278,
+      "loss": 1.0287,
+      "step": 7055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019261220323192385,
+      "loss": 1.052,
+      "step": 7060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001925925316486795,
+      "loss": 1.0628,
+      "step": 7065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019257283491761254,
+      "loss": 0.9556,
+      "step": 7070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019255311304407243,
+      "loss": 1.1124,
+      "step": 7075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001925333660334157,
+      "loss": 1.0557,
+      "step": 7080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019251359389100547,
+      "loss": 1.1685,
+      "step": 7085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019249379662221181,
+      "loss": 1.1111,
+      "step": 7090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019247397423241162,
+      "loss": 1.1248,
+      "step": 7095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019245412672698857,
+      "loss": 0.9916,
+      "step": 7100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019243425411133317,
+      "loss": 1.0128,
+      "step": 7105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001924143563908428,
+      "loss": 1.0776,
+      "step": 7110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019239443357092157,
+      "loss": 1.0848,
+      "step": 7115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019237448565698046,
+      "loss": 1.1113,
+      "step": 7120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019235451265443726,
+      "loss": 1.0992,
+      "step": 7125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019233451456871658,
+      "loss": 1.0531,
+      "step": 7130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019231449140524984,
+      "loss": 1.0496,
+      "step": 7135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019229444316947524,
+      "loss": 1.0508,
+      "step": 7140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019227436986683782,
+      "loss": 1.0668,
+      "step": 7145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019225427150278943,
+      "loss": 1.0097,
+      "step": 7150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001922341480827887,
+      "loss": 1.0788,
+      "step": 7155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019221399961230113,
+      "loss": 1.0439,
+      "step": 7160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019219382609679894,
+      "loss": 1.0776,
+      "step": 7165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001921736275417612,
+      "loss": 1.1287,
+      "step": 7170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019215340395267375,
+      "loss": 1.0896,
+      "step": 7175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019213315533502928,
+      "loss": 1.0834,
+      "step": 7180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019211288169432723,
+      "loss": 1.0397,
+      "step": 7185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019209258303607384,
+      "loss": 1.1279,
+      "step": 7190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019207225936578216,
+      "loss": 1.0745,
+      "step": 7195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001920519106889721,
+      "loss": 0.9946,
+      "step": 7200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001920315370111702,
+      "loss": 1.0628,
+      "step": 7205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001920111383379099,
+      "loss": 1.0947,
+      "step": 7210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001919907146747315,
+      "loss": 1.009,
+      "step": 7215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019197026602718186,
+      "loss": 1.0339,
+      "step": 7220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001919497924008149,
+      "loss": 1.0508,
+      "step": 7225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019192929380119108,
+      "loss": 1.0446,
+      "step": 7230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001919087702338778,
+      "loss": 1.0769,
+      "step": 7235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019188822170444918,
+      "loss": 1.0916,
+      "step": 7240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019186764821848616,
+      "loss": 0.9858,
+      "step": 7245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019184704978157638,
+      "loss": 1.0113,
+      "step": 7250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019182642639931438,
+      "loss": 1.1053,
+      "step": 7255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019180577807730134,
+      "loss": 1.0728,
+      "step": 7260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001917851048211453,
+      "loss": 1.1403,
+      "step": 7265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019176440663646102,
+      "loss": 1.1083,
+      "step": 7270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019174368352887008,
+      "loss": 1.0488,
+      "step": 7275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019172293550400084,
+      "loss": 1.1161,
+      "step": 7280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001917021625674883,
+      "loss": 1.0215,
+      "step": 7285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019168136472497445,
+      "loss": 1.0804,
+      "step": 7290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001916605419821078,
+      "loss": 1.0917,
+      "step": 7295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019163969434454382,
+      "loss": 1.0453,
+      "step": 7300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019161882181794457,
+      "loss": 1.0144,
+      "step": 7305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019159792440797904,
+      "loss": 1.091,
+      "step": 7310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001915770021203229,
+      "loss": 0.9946,
+      "step": 7315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019155605496065848,
+      "loss": 1.0256,
+      "step": 7320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019153508293467503,
+      "loss": 1.0741,
+      "step": 7325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019151408604806846,
+      "loss": 1.0902,
+      "step": 7330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001914930643065415,
+      "loss": 1.0409,
+      "step": 7335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019147201771580355,
+      "loss": 1.0306,
+      "step": 7340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019145094628157078,
+      "loss": 1.0588,
+      "step": 7345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019142985000956614,
+      "loss": 1.0233,
+      "step": 7350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001914087289055193,
+      "loss": 1.0628,
+      "step": 7355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019138758297516668,
+      "loss": 1.0893,
+      "step": 7360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019136641222425148,
+      "loss": 1.0242,
+      "step": 7365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019134521665852352,
+      "loss": 1.0498,
+      "step": 7370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001913239962837395,
+      "loss": 1.1378,
+      "step": 7375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019130275110566285,
+      "loss": 1.0647,
+      "step": 7380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019128148113006362,
+      "loss": 1.0919,
+      "step": 7385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019126018636271867,
+      "loss": 1.0667,
+      "step": 7390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019123886680941162,
+      "loss": 1.0603,
+      "step": 7395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019121752247593275,
+      "loss": 1.105,
+      "step": 7400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019119615336807917,
+      "loss": 1.068,
+      "step": 7405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001911747594916546,
+      "loss": 1.0312,
+      "step": 7410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019115334085246957,
+      "loss": 1.1432,
+      "step": 7415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001911318974563413,
+      "loss": 1.033,
+      "step": 7420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019111042930909376,
+      "loss": 1.0094,
+      "step": 7425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019108893641655761,
+      "loss": 0.9923,
+      "step": 7430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019106741878457026,
+      "loss": 1.048,
+      "step": 7435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019104587641897582,
+      "loss": 0.9729,
+      "step": 7440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019102430932562515,
+      "loss": 1.1082,
+      "step": 7445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019100271751037575,
+      "loss": 0.9876,
+      "step": 7450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019098110097909193,
+      "loss": 1.1132,
+      "step": 7455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019095945973764464,
+      "loss": 1.0795,
+      "step": 7460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001909377937919116,
+      "loss": 1.094,
+      "step": 7465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019091610314777717,
+      "loss": 1.0292,
+      "step": 7470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019089438781113248,
+      "loss": 1.049,
+      "step": 7475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019087264778787534,
+      "loss": 1.1129,
+      "step": 7480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019085088308391028,
+      "loss": 1.0573,
+      "step": 7485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001908290937051485,
+      "loss": 1.0603,
+      "step": 7490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001908072796575079,
+      "loss": 1.0293,
+      "step": 7495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001907854409469132,
+      "loss": 1.0986,
+      "step": 7500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001907635775792956,
+      "loss": 1.1646,
+      "step": 7505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019074168956059316,
+      "loss": 1.0537,
+      "step": 7510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019071977689675065,
+      "loss": 1.0483,
+      "step": 7515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019069783959371943,
+      "loss": 1.1612,
+      "step": 7520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019067587765745757,
+      "loss": 1.0754,
+      "step": 7525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001906538910939299,
+      "loss": 1.1354,
+      "step": 7530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001906318799091079,
+      "loss": 1.0109,
+      "step": 7535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019060984410896968,
+      "loss": 1.0934,
+      "step": 7540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019058778369950016,
+      "loss": 1.1387,
+      "step": 7545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019056569868669082,
+      "loss": 1.1044,
+      "step": 7550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001905435890765399,
+      "loss": 1.0475,
+      "step": 7555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019052145487505234,
+      "loss": 1.0782,
+      "step": 7560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019049929608823963,
+      "loss": 1.0668,
+      "step": 7565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019047711272212006,
+      "loss": 1.1293,
+      "step": 7570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019045490478271856,
+      "loss": 1.1171,
+      "step": 7575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019043267227606675,
+      "loss": 1.1093,
+      "step": 7580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001904104152082029,
+      "loss": 1.0032,
+      "step": 7585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019038813358517192,
+      "loss": 1.0426,
+      "step": 7590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019036582741302548,
+      "loss": 1.1175,
+      "step": 7595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019034349669782178,
+      "loss": 1.0634,
+      "step": 7600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019032114144562582,
+      "loss": 0.9559,
+      "step": 7605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019029876166250923,
+      "loss": 1.0568,
+      "step": 7610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019027635735455027,
+      "loss": 1.1106,
+      "step": 7615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019025392852783387,
+      "loss": 1.0473,
+      "step": 7620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019023147518845162,
+      "loss": 1.0409,
+      "step": 7625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019020899734250178,
+      "loss": 1.01,
+      "step": 7630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019018649499608921,
+      "loss": 1.0053,
+      "step": 7635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019016396815532555,
+      "loss": 1.1377,
+      "step": 7640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000190141416826329,
+      "loss": 0.9869,
+      "step": 7645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019011884101522437,
+      "loss": 1.1015,
+      "step": 7650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019009624072814326,
+      "loss": 1.0546,
+      "step": 7655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019007361597122373,
+      "loss": 1.0258,
+      "step": 7660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019005096675061065,
+      "loss": 1.0672,
+      "step": 7665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019002829307245548,
+      "loss": 1.0932,
+      "step": 7670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001900055949429163,
+      "loss": 1.0997,
+      "step": 7675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018998287236815785,
+      "loss": 1.1079,
+      "step": 7680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018996012535435147,
+      "loss": 1.0259,
+      "step": 7685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018993735390767527,
+      "loss": 1.0451,
+      "step": 7690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018991455803431378,
+      "loss": 1.1488,
+      "step": 7695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018989173774045835,
+      "loss": 1.1009,
+      "step": 7700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001898688930323069,
+      "loss": 1.0727,
+      "step": 7705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018984602391606395,
+      "loss": 1.1288,
+      "step": 7710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018982313039794067,
+      "loss": 0.9979,
+      "step": 7715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018980021248415487,
+      "loss": 1.104,
+      "step": 7720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000189777270180931,
+      "loss": 1.0332,
+      "step": 7725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001897543034945001,
+      "loss": 1.1018,
+      "step": 7730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018973131243109988,
+      "loss": 1.0029,
+      "step": 7735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018970829699697456,
+      "loss": 1.0737,
+      "step": 7740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001896852571983751,
+      "loss": 1.1074,
+      "step": 7745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018966219304155906,
+      "loss": 1.0807,
+      "step": 7750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001896391045327905,
+      "loss": 1.1124,
+      "step": 7755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001896159916783403,
+      "loss": 1.0085,
+      "step": 7760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018959285448448573,
+      "loss": 1.0485,
+      "step": 7765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018956969295751084,
+      "loss": 1.0846,
+      "step": 7770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018954650710370619,
+      "loss": 1.0411,
+      "step": 7775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018952329692936899,
+      "loss": 1.0668,
+      "step": 7780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018950006244080307,
+      "loss": 1.0369,
+      "step": 7785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018947680364431882,
+      "loss": 1.1261,
+      "step": 7790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001894535205462332,
+      "loss": 1.0033,
+      "step": 7795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001894302131528699,
+      "loss": 1.059,
+      "step": 7800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001894068814705591,
+      "loss": 1.1585,
+      "step": 7805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018938352550563762,
+      "loss": 1.0501,
+      "step": 7810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018936014526444884,
+      "loss": 1.0361,
+      "step": 7815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018933674075334274,
+      "loss": 0.931,
+      "step": 7820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000189313311978676,
+      "loss": 1.0988,
+      "step": 7825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001892898589468117,
+      "loss": 1.0829,
+      "step": 7830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018926638166411958,
+      "loss": 1.1268,
+      "step": 7835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001892428801369761,
+      "loss": 1.0373,
+      "step": 7840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018921935437176416,
+      "loss": 1.0947,
+      "step": 7845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001891958043748732,
+      "loss": 1.0747,
+      "step": 7850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018917223015269943,
+      "loss": 1.0562,
+      "step": 7855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018914863171164546,
+      "loss": 1.0468,
+      "step": 7860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018912500905812058,
+      "loss": 1.0434,
+      "step": 7865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018910136219854061,
+      "loss": 1.0742,
+      "step": 7870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018907769113932798,
+      "loss": 1.0173,
+      "step": 7875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018905399588691163,
+      "loss": 1.0872,
+      "step": 7880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018903027644772715,
+      "loss": 1.0333,
+      "step": 7885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018900653282821666,
+      "loss": 1.0405,
+      "step": 7890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001889827650348288,
+      "loss": 1.0363,
+      "step": 7895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018895897307401888,
+      "loss": 0.9642,
+      "step": 7900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001889351569522487,
+      "loss": 1.041,
+      "step": 7905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018891131667598662,
+      "loss": 1.0259,
+      "step": 7910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001888874522517076,
+      "loss": 1.0784,
+      "step": 7915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001888635636858931,
+      "loss": 1.0443,
+      "step": 7920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001888396509850312,
+      "loss": 1.094,
+      "step": 7925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018881571415561656,
+      "loss": 1.0991,
+      "step": 7930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018879175320415028,
+      "loss": 1.0486,
+      "step": 7935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001887677681371401,
+      "loss": 1.0873,
+      "step": 7940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018874375896110027,
+      "loss": 1.0915,
+      "step": 7945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001887197256825516,
+      "loss": 1.0575,
+      "step": 7950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018869566830802148,
+      "loss": 1.1026,
+      "step": 7955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018867158684404375,
+      "loss": 1.0201,
+      "step": 7960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018864748129715896,
+      "loss": 1.0621,
+      "step": 7965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000188623351673914,
+      "loss": 1.1016,
+      "step": 7970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018859919798086238,
+      "loss": 1.0232,
+      "step": 7975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018857502022456423,
+      "loss": 1.0368,
+      "step": 7980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018855081841158614,
+      "loss": 1.0235,
+      "step": 7985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018852659254850126,
+      "loss": 1.0425,
+      "step": 7990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001885023426418892,
+      "loss": 1.0677,
+      "step": 7995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018847806869833614,
+      "loss": 1.0818,
+      "step": 8000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001884537707244349,
+      "loss": 1.0938,
+      "step": 8005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018842944872678462,
+      "loss": 1.0528,
+      "step": 8010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018840510271199112,
+      "loss": 1.0229,
+      "step": 8015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018838073268666672,
+      "loss": 1.0359,
+      "step": 8020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001883563386574302,
+      "loss": 1.0493,
+      "step": 8025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018833192063090693,
+      "loss": 1.0767,
+      "step": 8030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018830747861372873,
+      "loss": 1.0127,
+      "step": 8035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018828301261253398,
+      "loss": 1.1243,
+      "step": 8040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018825852263396758,
+      "loss": 1.0462,
+      "step": 8045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018823400868468092,
+      "loss": 1.0391,
+      "step": 8050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018820947077133186,
+      "loss": 1.0506,
+      "step": 8055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001881849089005849,
+      "loss": 1.0589,
+      "step": 8060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018816032307911092,
+      "loss": 1.0248,
+      "step": 8065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018813571331358734,
+      "loss": 1.0853,
+      "step": 8070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018811107961069808,
+      "loss": 1.0075,
+      "step": 8075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018808642197713362,
+      "loss": 1.0741,
+      "step": 8080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018806174041959085,
+      "loss": 1.0568,
+      "step": 8085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018803703494477323,
+      "loss": 1.0012,
+      "step": 8090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018801230555939063,
+      "loss": 1.0675,
+      "step": 8095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001879875522701595,
+      "loss": 1.0536,
+      "step": 8100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001879627750838028,
+      "loss": 1.0919,
+      "step": 8105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018793797400704985,
+      "loss": 1.0975,
+      "step": 8110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018791314904663655,
+      "loss": 1.1052,
+      "step": 8115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018788830020930535,
+      "loss": 1.0374,
+      "step": 8120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000187863427501805,
+      "loss": 1.0021,
+      "step": 8125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018783853093089095,
+      "loss": 1.0822,
+      "step": 8130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018781361050332498,
+      "loss": 1.0305,
+      "step": 8135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001877886662258754,
+      "loss": 1.1041,
+      "step": 8140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000187763698105317,
+      "loss": 1.0321,
+      "step": 8145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000187738706148431,
+      "loss": 1.0294,
+      "step": 8150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018771369036200513,
+      "loss": 1.0225,
+      "step": 8155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018768865075283366,
+      "loss": 1.0785,
+      "step": 8160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018766358732771722,
+      "loss": 1.0822,
+      "step": 8165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000187638500093463,
+      "loss": 1.0403,
+      "step": 8170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018761338905688452,
+      "loss": 1.0497,
+      "step": 8175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018758825422480192,
+      "loss": 1.0821,
+      "step": 8180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018756309560404174,
+      "loss": 1.0261,
+      "step": 8185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018753791320143693,
+      "loss": 1.0271,
+      "step": 8190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018751270702382697,
+      "loss": 1.0672,
+      "step": 8195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001874874770780578,
+      "loss": 1.0327,
+      "step": 8200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018746222337098173,
+      "loss": 1.0406,
+      "step": 8205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018743694590945767,
+      "loss": 1.0679,
+      "step": 8210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001874116447003508,
+      "loss": 0.9973,
+      "step": 8215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001873863197505329,
+      "loss": 1.0753,
+      "step": 8220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018736097106688215,
+      "loss": 1.0561,
+      "step": 8225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001873355986562831,
+      "loss": 1.0845,
+      "step": 8230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018731020252562693,
+      "loss": 1.0671,
+      "step": 8235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000187284782681811,
+      "loss": 1.0487,
+      "step": 8240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018725933913173938,
+      "loss": 1.0248,
+      "step": 8245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001872338718823224,
+      "loss": 0.9849,
+      "step": 8250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018720838094047685,
+      "loss": 1.1194,
+      "step": 8255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018718286631312604,
+      "loss": 0.9687,
+      "step": 8260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018715732800719963,
+      "loss": 0.9876,
+      "step": 8265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018713176602963374,
+      "loss": 1.0317,
+      "step": 8270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001871061803873709,
+      "loss": 1.112,
+      "step": 8275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018708057108736013,
+      "loss": 1.0606,
+      "step": 8280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001870549381365568,
+      "loss": 1.0588,
+      "step": 8285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018702928154192274,
+      "loss": 1.1281,
+      "step": 8290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018700360131042621,
+      "loss": 1.0934,
+      "step": 8295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018697789744904185,
+      "loss": 1.0193,
+      "step": 8300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018695216996475074,
+      "loss": 1.0087,
+      "step": 8305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001869264188645404,
+      "loss": 0.9902,
+      "step": 8310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018690064415540473,
+      "loss": 1.1592,
+      "step": 8315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018687484584434407,
+      "loss": 1.062,
+      "step": 8320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018684902393836515,
+      "loss": 1.1288,
+      "step": 8325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001868231784444811,
+      "loss": 1.0769,
+      "step": 8330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018679730936971148,
+      "loss": 1.101,
+      "step": 8335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018677141672108225,
+      "loss": 1.1072,
+      "step": 8340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018674550050562576,
+      "loss": 1.1311,
+      "step": 8345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018671956073038076,
+      "loss": 1.117,
+      "step": 8350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001866935974023924,
+      "loss": 1.1281,
+      "step": 8355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018666761052871228,
+      "loss": 1.0334,
+      "step": 8360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001866416001163983,
+      "loss": 1.0716,
+      "step": 8365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018661556617251482,
+      "loss": 1.0865,
+      "step": 8370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018658950870413255,
+      "loss": 1.1423,
+      "step": 8375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018656342771832866,
+      "loss": 1.0613,
+      "step": 8380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001865373232221866,
+      "loss": 1.0547,
+      "step": 8385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018651119522279633,
+      "loss": 1.1119,
+      "step": 8390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001864850437272541,
+      "loss": 1.0498,
+      "step": 8395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018645886874266252,
+      "loss": 1.0417,
+      "step": 8400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018643267027613068,
+      "loss": 1.0226,
+      "step": 8405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000186406448334774,
+      "loss": 1.1333,
+      "step": 8410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018638020292571422,
+      "loss": 1.0347,
+      "step": 8415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018635393405607958,
+      "loss": 1.1059,
+      "step": 8420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018632764173300457,
+      "loss": 1.0667,
+      "step": 8425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018630132596363007,
+      "loss": 1.0696,
+      "step": 8430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001862749867551034,
+      "loss": 1.069,
+      "step": 8435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018624862411457824,
+      "loss": 1.0165,
+      "step": 8440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018622223804921453,
+      "loss": 1.0232,
+      "step": 8445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018619582856617864,
+      "loss": 1.0774,
+      "step": 8450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001861693956726433,
+      "loss": 1.0932,
+      "step": 8455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018614293937578764,
+      "loss": 1.0742,
+      "step": 8460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018611645968279707,
+      "loss": 1.0233,
+      "step": 8465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018608995660086337,
+      "loss": 1.0958,
+      "step": 8470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018606343013718472,
+      "loss": 1.1775,
+      "step": 8475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001860368802989656,
+      "loss": 1.0299,
+      "step": 8480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018601030709341685,
+      "loss": 1.0571,
+      "step": 8485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018598371052775566,
+      "loss": 1.0272,
+      "step": 8490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018595709060920563,
+      "loss": 1.0953,
+      "step": 8495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018593044734499655,
+      "loss": 1.0284,
+      "step": 8500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018590378074236473,
+      "loss": 1.0557,
+      "step": 8505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018587709080855266,
+      "loss": 1.0895,
+      "step": 8510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018585037755080926,
+      "loss": 1.044,
+      "step": 8515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018582364097638978,
+      "loss": 1.0315,
+      "step": 8520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018579688109255575,
+      "loss": 1.0601,
+      "step": 8525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001857700979065751,
+      "loss": 1.0313,
+      "step": 8530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018574329142572206,
+      "loss": 1.0689,
+      "step": 8535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018571646165727713,
+      "loss": 1.1464,
+      "step": 8540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001856896086085272,
+      "loss": 1.0778,
+      "step": 8545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018566273228676553,
+      "loss": 1.0185,
+      "step": 8550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018563583269929157,
+      "loss": 0.9554,
+      "step": 8555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001856089098534112,
+      "loss": 1.0443,
+      "step": 8560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018558196375643654,
+      "loss": 1.0408,
+      "step": 8565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018555499441568607,
+      "loss": 1.1809,
+      "step": 8570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018552800183848457,
+      "loss": 0.9881,
+      "step": 8575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018550098603216318,
+      "loss": 1.0447,
+      "step": 8580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018547394700405925,
+      "loss": 1.0367,
+      "step": 8585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001854468847615165,
+      "loss": 1.053,
+      "step": 8590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018541979931188494,
+      "loss": 1.1557,
+      "step": 8595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018539269066252095,
+      "loss": 1.0831,
+      "step": 8600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018536555882078706,
+      "loss": 1.0323,
+      "step": 8605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001853384037940522,
+      "loss": 1.0565,
+      "step": 8610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018531122558969166,
+      "loss": 1.0819,
+      "step": 8615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018528402421508688,
+      "loss": 1.096,
+      "step": 8620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018525679967762568,
+      "loss": 1.0257,
+      "step": 8625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018522955198470214,
+      "loss": 1.0263,
+      "step": 8630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018520228114371669,
+      "loss": 0.9635,
+      "step": 8635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018517498716207596,
+      "loss": 1.0757,
+      "step": 8640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001851476700471929,
+      "loss": 1.0224,
+      "step": 8645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018512032980648674,
+      "loss": 1.1312,
+      "step": 8650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018509296644738304,
+      "loss": 1.0558,
+      "step": 8655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001850655799773136,
+      "loss": 1.1064,
+      "step": 8660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018503817040371642,
+      "loss": 1.1367,
+      "step": 8665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018501073773403593,
+      "loss": 1.1039,
+      "step": 8670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001849832819757227,
+      "loss": 1.0733,
+      "step": 8675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018495580313623365,
+      "loss": 1.1023,
+      "step": 8680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018492830122303193,
+      "loss": 1.0357,
+      "step": 8685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018490077624358698,
+      "loss": 0.9803,
+      "step": 8690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018487322820537444,
+      "loss": 1.0748,
+      "step": 8695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018484565711587636,
+      "loss": 0.994,
+      "step": 8700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018481806298258088,
+      "loss": 0.9863,
+      "step": 8705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001847904458129825,
+      "loss": 1.0737,
+      "step": 8710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018476280561458195,
+      "loss": 1.041,
+      "step": 8715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018473514239488622,
+      "loss": 1.0958,
+      "step": 8720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018470745616140855,
+      "loss": 1.1243,
+      "step": 8725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001846797469216684,
+      "loss": 1.0914,
+      "step": 8730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018465201468319156,
+      "loss": 1.069,
+      "step": 8735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018462425945350996,
+      "loss": 1.1206,
+      "step": 8740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018459648124016185,
+      "loss": 1.0171,
+      "step": 8745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001845686800506917,
+      "loss": 1.1079,
+      "step": 8750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001845408558926502,
+      "loss": 1.0418,
+      "step": 8755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018451300877359437,
+      "loss": 1.0546,
+      "step": 8760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001844851387010873,
+      "loss": 1.0015,
+      "step": 8765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018445724568269845,
+      "loss": 1.0903,
+      "step": 8770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001844293297260035,
+      "loss": 1.0207,
+      "step": 8775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001844013908385843,
+      "loss": 1.1067,
+      "step": 8780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018437342902802897,
+      "loss": 1.0979,
+      "step": 8785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018434544430193182,
+      "loss": 1.0979,
+      "step": 8790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018431743666789342,
+      "loss": 1.047,
+      "step": 8795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018428940613352058,
+      "loss": 1.0804,
+      "step": 8800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001842613527064262,
+      "loss": 1.043,
+      "step": 8805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018423327639422964,
+      "loss": 1.0372,
+      "step": 8810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001842051772045563,
+      "loss": 1.1403,
+      "step": 8815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001841770551450377,
+      "loss": 1.0577,
+      "step": 8820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018414891022331184,
+      "loss": 1.1079,
+      "step": 8825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018412074244702274,
+      "loss": 1.0391,
+      "step": 8830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018409255182382065,
+      "loss": 0.9934,
+      "step": 8835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018406433836136205,
+      "loss": 1.0312,
+      "step": 8840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018403610206730968,
+      "loss": 1.0983,
+      "step": 8845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018400784294933238,
+      "loss": 0.9886,
+      "step": 8850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018397956101510522,
+      "loss": 1.0446,
+      "step": 8855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018395125627230949,
+      "loss": 1.1094,
+      "step": 8860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018392292872863267,
+      "loss": 1.0722,
+      "step": 8865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018389457839176844,
+      "loss": 1.0138,
+      "step": 8870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018386620526941662,
+      "loss": 1.0333,
+      "step": 8875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001838378093692833,
+      "loss": 1.1228,
+      "step": 8880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018380939069908067,
+      "loss": 1.1053,
+      "step": 8885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018378094926652718,
+      "loss": 1.0006,
+      "step": 8890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001837524850793474,
+      "loss": 1.0861,
+      "step": 8895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001837239981452721,
+      "loss": 1.0508,
+      "step": 8900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001836954884720383,
+      "loss": 1.1438,
+      "step": 8905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018366695606738908,
+      "loss": 1.0971,
+      "step": 8910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018363840093907376,
+      "loss": 1.0189,
+      "step": 8915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001836098230948478,
+      "loss": 1.0605,
+      "step": 8920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018358122254247286,
+      "loss": 1.0267,
+      "step": 8925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018355259928971678,
+      "loss": 1.086,
+      "step": 8930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001835239533443535,
+      "loss": 1.1447,
+      "step": 8935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018349528471416315,
+      "loss": 1.0877,
+      "step": 8940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001834665934069321,
+      "loss": 1.0966,
+      "step": 8945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018343787943045275,
+      "loss": 1.0805,
+      "step": 8950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018340914279252376,
+      "loss": 1.0764,
+      "step": 8955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018338038350094985,
+      "loss": 1.0628,
+      "step": 8960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018335160156354206,
+      "loss": 1.056,
+      "step": 8965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018332279698811735,
+      "loss": 1.1383,
+      "step": 8970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000183293969782499,
+      "loss": 1.1231,
+      "step": 8975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018326511995451637,
+      "loss": 1.0614,
+      "step": 8980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018323624751200498,
+      "loss": 1.0065,
+      "step": 8985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018320735246280654,
+      "loss": 1.1525,
+      "step": 8990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018317843481476876,
+      "loss": 1.0112,
+      "step": 8995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001831494945757456,
+      "loss": 1.1021,
+      "step": 9000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018312053175359718,
+      "loss": 1.0999,
+      "step": 9005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018309154635618965,
+      "loss": 1.0045,
+      "step": 9010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001830625383913954,
+      "loss": 1.0683,
+      "step": 9015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018303350786709287,
+      "loss": 1.0421,
+      "step": 9020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018300445479116663,
+      "loss": 0.9429,
+      "step": 9025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018297537917150747,
+      "loss": 1.045,
+      "step": 9030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018294628101601213,
+      "loss": 1.1988,
+      "step": 9035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018291716033258366,
+      "loss": 1.1228,
+      "step": 9040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001828880171291311,
+      "loss": 1.0659,
+      "step": 9045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018285885141356963,
+      "loss": 0.9692,
+      "step": 9050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018282966319382057,
+      "loss": 1.075,
+      "step": 9055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001828004524778114,
+      "loss": 1.0313,
+      "step": 9060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018277121927347557,
+      "loss": 1.0588,
+      "step": 9065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001827419635887528,
+      "loss": 1.1489,
+      "step": 9070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018271268543158876,
+      "loss": 1.1478,
+      "step": 9075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018268338480993535,
+      "loss": 1.0938,
+      "step": 9080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018265406173175052,
+      "loss": 1.0669,
+      "step": 9085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018262471620499832,
+      "loss": 1.0314,
+      "step": 9090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018259534823764886,
+      "loss": 1.0462,
+      "step": 9095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018256595783767843,
+      "loss": 1.1094,
+      "step": 9100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001825365450130693,
+      "loss": 1.0894,
+      "step": 9105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018250710977180993,
+      "loss": 0.9947,
+      "step": 9110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001824776521218949,
+      "loss": 0.9896,
+      "step": 9115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001824481720713247,
+      "loss": 0.9662,
+      "step": 9120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018241866962810609,
+      "loss": 1.0737,
+      "step": 9125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001823891448002518,
+      "loss": 1.1008,
+      "step": 9130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018235959759578066,
+      "loss": 1.1576,
+      "step": 9135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018233002802271763,
+      "loss": 1.0693,
+      "step": 9140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001823004360890937,
+      "loss": 1.0303,
+      "step": 9145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018227082180294593,
+      "loss": 1.1354,
+      "step": 9150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018224118517231745,
+      "loss": 0.9801,
+      "step": 9155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001822115262052575,
+      "loss": 1.1462,
+      "step": 9160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018218184490982134,
+      "loss": 1.089,
+      "step": 9165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001821521412940703,
+      "loss": 1.0919,
+      "step": 9170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018212241536607178,
+      "loss": 1.0244,
+      "step": 9175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018209266713389928,
+      "loss": 1.0901,
+      "step": 9180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018206289660563227,
+      "loss": 1.0607,
+      "step": 9185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018203310378935638,
+      "loss": 1.129,
+      "step": 9190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001820032886931632,
+      "loss": 1.0977,
+      "step": 9195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018197345132515044,
+      "loss": 1.0385,
+      "step": 9200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001819435916934218,
+      "loss": 1.0308,
+      "step": 9205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018191370980608705,
+      "loss": 1.1238,
+      "step": 9210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018188380567126205,
+      "loss": 1.0373,
+      "step": 9215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018185387929706866,
+      "loss": 1.069,
+      "step": 9220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018182393069163477,
+      "loss": 1.107,
+      "step": 9225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018179395986309428,
+      "loss": 1.0122,
+      "step": 9230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018176396681958724,
+      "loss": 1.0674,
+      "step": 9235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001817339515692596,
+      "loss": 1.0335,
+      "step": 9240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018170391412026344,
+      "loss": 0.9976,
+      "step": 9245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018167385448075683,
+      "loss": 1.09,
+      "step": 9250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018164377265890384,
+      "loss": 1.1133,
+      "step": 9255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018161366866287462,
+      "loss": 1.1153,
+      "step": 9260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018158354250084527,
+      "loss": 0.9831,
+      "step": 9265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018155339418099802,
+      "loss": 1.0122,
+      "step": 9270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018152322371152102,
+      "loss": 1.0698,
+      "step": 9275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018149303110060846,
+      "loss": 1.0615,
+      "step": 9280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018146281635646059,
+      "loss": 1.0508,
+      "step": 9285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018143257948728358,
+      "loss": 1.0151,
+      "step": 9290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001814023205012897,
+      "loss": 1.0845,
+      "step": 9295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018137203940669715,
+      "loss": 1.0291,
+      "step": 9300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018134173621173023,
+      "loss": 1.0044,
+      "step": 9305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018131141092461917,
+      "loss": 1.0676,
+      "step": 9310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001812810635536002,
+      "loss": 1.0104,
+      "step": 9315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018125069410691554,
+      "loss": 1.0304,
+      "step": 9320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001812203025928135,
+      "loss": 1.0728,
+      "step": 9325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018118988901954825,
+      "loss": 1.0693,
+      "step": 9330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018115945339538007,
+      "loss": 1.1296,
+      "step": 9335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018112899572857513,
+      "loss": 1.0553,
+      "step": 9340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018109851602740563,
+      "loss": 1.0454,
+      "step": 9345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001810680143001498,
+      "loss": 1.1059,
+      "step": 9350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018103749055509177,
+      "loss": 1.0721,
+      "step": 9355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018100694480052167,
+      "loss": 1.0436,
+      "step": 9360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001809763770447357,
+      "loss": 1.0557,
+      "step": 9365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018094578729603583,
+      "loss": 1.0236,
+      "step": 9370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018091517556273029,
+      "loss": 1.0323,
+      "step": 9375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018088454185313303,
+      "loss": 1.0606,
+      "step": 9380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018085388617556408,
+      "loss": 1.0591,
+      "step": 9385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001808232085383494,
+      "loss": 1.0361,
+      "step": 9390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000180792508949821,
+      "loss": 1.0301,
+      "step": 9395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018076178741831668,
+      "loss": 1.0147,
+      "step": 9400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001807310439521804,
+      "loss": 1.1033,
+      "step": 9405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018070027855976198,
+      "loss": 1.123,
+      "step": 9410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018066949124941712,
+      "loss": 1.1029,
+      "step": 9415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018063868202950763,
+      "loss": 1.0138,
+      "step": 9420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018060785090840114,
+      "loss": 1.03,
+      "step": 9425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018057699789447132,
+      "loss": 0.9927,
+      "step": 9430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001805461229960977,
+      "loss": 1.0876,
+      "step": 9435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018051522622166585,
+      "loss": 1.0637,
+      "step": 9440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001804843075795672,
+      "loss": 1.035,
+      "step": 9445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018045336707819912,
+      "loss": 1.0558,
+      "step": 9450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018042240472596502,
+      "loss": 1.0774,
+      "step": 9455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018039142053127412,
+      "loss": 1.0296,
+      "step": 9460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018036041450254165,
+      "loss": 1.0875,
+      "step": 9465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018032938664818871,
+      "loss": 1.1091,
+      "step": 9470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001802983369766424,
+      "loss": 1.0048,
+      "step": 9475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018026726549633573,
+      "loss": 1.0819,
+      "step": 9480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001802361722157075,
+      "loss": 1.0892,
+      "step": 9485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018020505714320267,
+      "loss": 0.9882,
+      "step": 9490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001801739202872719,
+      "loss": 1.0099,
+      "step": 9495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018014276165637187,
+      "loss": 1.0049,
+      "step": 9500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018011158125896522,
+      "loss": 0.9832,
+      "step": 9505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018008037910352038,
+      "loss": 0.9622,
+      "step": 9510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018004915519851178,
+      "loss": 1.0992,
+      "step": 9515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018001790955241972,
+      "loss": 1.0749,
+      "step": 9520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001799866421737304,
+      "loss": 1.0929,
+      "step": 9525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017995535307093597,
+      "loss": 1.0834,
+      "step": 9530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001799240422525344,
+      "loss": 1.0512,
+      "step": 9535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017989270972702966,
+      "loss": 1.0923,
+      "step": 9540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001798613555029315,
+      "loss": 1.1023,
+      "step": 9545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017982997958875565,
+      "loss": 1.0273,
+      "step": 9550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017979858199302374,
+      "loss": 1.091,
+      "step": 9555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017976716272426318,
+      "loss": 1.1111,
+      "step": 9560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001797357217910074,
+      "loss": 1.071,
+      "step": 9565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017970425920179562,
+      "loss": 1.0787,
+      "step": 9570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017967277496517298,
+      "loss": 1.0982,
+      "step": 9575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017964126908969047,
+      "loss": 1.1133,
+      "step": 9580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017960974158390505,
+      "loss": 1.0864,
+      "step": 9585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017957819245637943,
+      "loss": 1.1479,
+      "step": 9590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017954662171568226,
+      "loss": 1.0617,
+      "step": 9595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017951502937038803,
+      "loss": 1.0668,
+      "step": 9600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017948341542907713,
+      "loss": 0.9874,
+      "step": 9605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017945177990033584,
+      "loss": 1.0718,
+      "step": 9610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001794201227927562,
+      "loss": 1.0583,
+      "step": 9615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017938844411493623,
+      "loss": 1.0286,
+      "step": 9620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001793567438754797,
+      "loss": 1.0601,
+      "step": 9625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001793250220829963,
+      "loss": 0.981,
+      "step": 9630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001792932787461016,
+      "loss": 1.097,
+      "step": 9635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017926151387341697,
+      "loss": 1.0536,
+      "step": 9640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017922972747356963,
+      "loss": 1.1318,
+      "step": 9645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017919791955519268,
+      "loss": 1.0816,
+      "step": 9650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000179166090126925,
+      "loss": 0.9683,
+      "step": 9655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001791342391974114,
+      "loss": 1.0296,
+      "step": 9660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017910236677530248,
+      "loss": 1.0833,
+      "step": 9665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017907047286925466,
+      "loss": 1.0989,
+      "step": 9670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017903855748793027,
+      "loss": 1.0359,
+      "step": 9675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017900662063999735,
+      "loss": 0.9875,
+      "step": 9680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001789746623341299,
+      "loss": 1.1201,
+      "step": 9685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017894268257900766,
+      "loss": 1.0228,
+      "step": 9690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017891068138331623,
+      "loss": 1.0629,
+      "step": 9695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017887865875574702,
+      "loss": 1.0374,
+      "step": 9700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017884661470499733,
+      "loss": 1.0532,
+      "step": 9705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001788145492397701,
+      "loss": 1.007,
+      "step": 9710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017878246236877431,
+      "loss": 1.1285,
+      "step": 9715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017875035410072458,
+      "loss": 1.0121,
+      "step": 9720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017871822444434144,
+      "loss": 1.0373,
+      "step": 9725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001786860734083512,
+      "loss": 1.0169,
+      "step": 9730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000178653901001486,
+      "loss": 1.0481,
+      "step": 9735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017862170723248371,
+      "loss": 1.0503,
+      "step": 9740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017858949211008802,
+      "loss": 1.0542,
+      "step": 9745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017855725564304854,
+      "loss": 1.0643,
+      "step": 9750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017852499784012048,
+      "loss": 1.0859,
+      "step": 9755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017849271871006508,
+      "loss": 0.9938,
+      "step": 9760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017846041826164916,
+      "loss": 1.0597,
+      "step": 9765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017842809650364541,
+      "loss": 1.1135,
+      "step": 9770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017839575344483238,
+      "loss": 0.991,
+      "step": 9775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017836338909399423,
+      "loss": 1.0387,
+      "step": 9780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017833100345992112,
+      "loss": 1.0582,
+      "step": 9785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001782985965514088,
+      "loss": 0.9943,
+      "step": 9790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017826616837725894,
+      "loss": 1.0798,
+      "step": 9795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017823371894627884,
+      "loss": 1.0353,
+      "step": 9800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017820124826728172,
+      "loss": 1.14,
+      "step": 9805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017816875634908653,
+      "loss": 1.1337,
+      "step": 9810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017813624320051793,
+      "loss": 1.0926,
+      "step": 9815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017810370883040636,
+      "loss": 1.0967,
+      "step": 9820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001780711532475881,
+      "loss": 1.1147,
+      "step": 9825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001780385764609051,
+      "loss": 1.126,
+      "step": 9830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001780059784792051,
+      "loss": 1.0112,
+      "step": 9835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017797335931134164,
+      "loss": 1.1067,
+      "step": 9840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017794071896617397,
+      "loss": 1.0192,
+      "step": 9845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017790805745256704,
+      "loss": 1.0365,
+      "step": 9850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017787537477939168,
+      "loss": 1.0413,
+      "step": 9855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017784267095552437,
+      "loss": 1.0762,
+      "step": 9860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017780994598984737,
+      "loss": 1.1271,
+      "step": 9865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017777719989124866,
+      "loss": 0.9765,
+      "step": 9870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017774443266862196,
+      "loss": 1.1807,
+      "step": 9875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017771164433086676,
+      "loss": 1.1155,
+      "step": 9880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017767883488688824,
+      "loss": 1.1016,
+      "step": 9885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017764600434559734,
+      "loss": 1.0402,
+      "step": 9890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017761315271591074,
+      "loss": 1.0907,
+      "step": 9895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017758028000675085,
+      "loss": 1.04,
+      "step": 9900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017754738622704575,
+      "loss": 1.0962,
+      "step": 9905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001775144713857293,
+      "loss": 1.0974,
+      "step": 9910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017748153549174101,
+      "loss": 1.1375,
+      "step": 9915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017744857855402625,
+      "loss": 0.9718,
+      "step": 9920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000177415600581536,
+      "loss": 0.9461,
+      "step": 9925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017738260158322688,
+      "loss": 1.0948,
+      "step": 9930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017734958156806138,
+      "loss": 0.9827,
+      "step": 9935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001773165405450076,
+      "loss": 1.0447,
+      "step": 9940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017728347852303942,
+      "loss": 1.0658,
+      "step": 9945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017725039551113632,
+      "loss": 1.0579,
+      "step": 9950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017721729151828355,
+      "loss": 1.0644,
+      "step": 9955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017718416655347204,
+      "loss": 1.0203,
+      "step": 9960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017715102062569843,
+      "loss": 1.0373,
+      "step": 9965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017711785374396506,
+      "loss": 1.0775,
+      "step": 9970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017708466591727994,
+      "loss": 1.0262,
+      "step": 9975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001770514571546567,
+      "loss": 1.0232,
+      "step": 9980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001770182274651148,
+      "loss": 1.0877,
+      "step": 9985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001769849768576793,
+      "loss": 0.999,
+      "step": 9990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017695170534138099,
+      "loss": 1.1102,
+      "step": 9995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017691841292525625,
+      "loss": 1.0055,
+      "step": 10000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017688509961834718,
+      "loss": 1.0518,
+      "step": 10005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017685176542970164,
+      "loss": 1.0937,
+      "step": 10010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000176818410368373,
+      "loss": 1.1077,
+      "step": 10015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017678503444342041,
+      "loss": 1.0592,
+      "step": 10020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017675163766390868,
+      "loss": 1.0446,
+      "step": 10025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017671822003890823,
+      "loss": 1.1023,
+      "step": 10030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017668478157749522,
+      "loss": 1.0301,
+      "step": 10035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017665132228875138,
+      "loss": 1.0979,
+      "step": 10040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001766178421817642,
+      "loss": 0.9809,
+      "step": 10045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017658434126562668,
+      "loss": 1.1656,
+      "step": 10050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017655081954943763,
+      "loss": 1.0992,
+      "step": 10055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001765172770423014,
+      "loss": 1.0201,
+      "step": 10060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017648371375332804,
+      "loss": 1.0051,
+      "step": 10065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017645012969163323,
+      "loss": 1.0845,
+      "step": 10070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017641652486633826,
+      "loss": 0.9871,
+      "step": 10075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001763828992865701,
+      "loss": 1.105,
+      "step": 10080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017634925296146136,
+      "loss": 0.976,
+      "step": 10085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017631558590015028,
+      "loss": 1.1195,
+      "step": 10090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001762818981117807,
+      "loss": 1.0407,
+      "step": 10095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001762481896055021,
+      "loss": 1.0617,
+      "step": 10100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017621446039046962,
+      "loss": 1.0943,
+      "step": 10105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000176180710475844,
+      "loss": 1.0255,
+      "step": 10110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017614693987079162,
+      "loss": 1.0764,
+      "step": 10115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001761131485844844,
+      "loss": 1.0446,
+      "step": 10120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001760793366261,
+      "loss": 1.0476,
+      "step": 10125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017604550400482163,
+      "loss": 1.0885,
+      "step": 10130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001760116507298381,
+      "loss": 1.0298,
+      "step": 10135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017597777681034385,
+      "loss": 1.0527,
+      "step": 10140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017594388225553894,
+      "loss": 1.074,
+      "step": 10145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017590996707462898,
+      "loss": 1.098,
+      "step": 10150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017587603127682525,
+      "loss": 1.027,
+      "step": 10155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001758420748713446,
+      "loss": 1.0795,
+      "step": 10160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017580809786740947,
+      "loss": 1.0218,
+      "step": 10165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001757741002742479,
+      "loss": 1.1484,
+      "step": 10170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017574008210109355,
+      "loss": 1.1184,
+      "step": 10175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001757060433571856,
+      "loss": 1.0346,
+      "step": 10180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017567198405176883,
+      "loss": 0.9837,
+      "step": 10185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017563790419409373,
+      "loss": 1.1164,
+      "step": 10190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001756038037934162,
+      "loss": 1.1031,
+      "step": 10195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017556968285899785,
+      "loss": 1.057,
+      "step": 10200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017553554140010573,
+      "loss": 1.1212,
+      "step": 10205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017550137942601263,
+      "loss": 1.1493,
+      "step": 10210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017546719694599682,
+      "loss": 1.0712,
+      "step": 10215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017543299396934208,
+      "loss": 1.0703,
+      "step": 10220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001753987705053379,
+      "loss": 1.0419,
+      "step": 10225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001753645265632792,
+      "loss": 1.0542,
+      "step": 10230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017533026215246655,
+      "loss": 1.0051,
+      "step": 10235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017529597728220607,
+      "loss": 0.9742,
+      "step": 10240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017526167196180937,
+      "loss": 1.0363,
+      "step": 10245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017522734620059369,
+      "loss": 1.0681,
+      "step": 10250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017519300000788177,
+      "loss": 1.0744,
+      "step": 10255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000175158633393002,
+      "loss": 0.9856,
+      "step": 10260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017512424636528811,
+      "loss": 0.9904,
+      "step": 10265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017508983893407958,
+      "loss": 1.043,
+      "step": 10270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017505541110872134,
+      "loss": 1.0616,
+      "step": 10275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017502096289856388,
+      "loss": 1.0931,
+      "step": 10280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017498649431296322,
+      "loss": 1.1059,
+      "step": 10285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001749520053612809,
+      "loss": 1.1393,
+      "step": 10290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000174917496052884,
+      "loss": 1.1346,
+      "step": 10295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017488296639714515,
+      "loss": 1.0574,
+      "step": 10300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017484841640344248,
+      "loss": 1.0793,
+      "step": 10305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017481384608115965,
+      "loss": 1.0286,
+      "step": 10310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017477925543968586,
+      "loss": 1.106,
+      "step": 10315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001747446444884158,
+      "loss": 1.0845,
+      "step": 10320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017471001323674968,
+      "loss": 1.0591,
+      "step": 10325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017467536169409322,
+      "loss": 1.0897,
+      "step": 10330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017464068986985774,
+      "loss": 0.9603,
+      "step": 10335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001746059977734599,
+      "loss": 1.1038,
+      "step": 10340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000174571285414322,
+      "loss": 1.1185,
+      "step": 10345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001745365528018718,
+      "loss": 1.0566,
+      "step": 10350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017450179994554253,
+      "loss": 1.0225,
+      "step": 10355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017446702685477303,
+      "loss": 1.113,
+      "step": 10360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001744322335390075,
+      "loss": 0.9604,
+      "step": 10365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017439742000769566,
+      "loss": 1.0882,
+      "step": 10370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001743625862702928,
+      "loss": 1.0483,
+      "step": 10375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017432773233625961,
+      "loss": 0.9613,
+      "step": 10380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017429285821506232,
+      "loss": 0.9571,
+      "step": 10385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017425796391617264,
+      "loss": 1.1105,
+      "step": 10390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001742230494490677,
+      "loss": 1.1211,
+      "step": 10395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017418811482323022,
+      "loss": 1.0489,
+      "step": 10400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001741531600481483,
+      "loss": 1.0653,
+      "step": 10405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017411818513331549,
+      "loss": 1.149,
+      "step": 10410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017408319008823092,
+      "loss": 0.9796,
+      "step": 10415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001740481749223991,
+      "loss": 1.0434,
+      "step": 10420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017401313964533004,
+      "loss": 1.0935,
+      "step": 10425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017397808426653923,
+      "loss": 1.0577,
+      "step": 10430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001739430087955475,
+      "loss": 1.0501,
+      "step": 10435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017390791324188137,
+      "loss": 1.0698,
+      "step": 10440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017387279761507261,
+      "loss": 1.0553,
+      "step": 10445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017383766192465846,
+      "loss": 1.1076,
+      "step": 10450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001738025061801817,
+      "loss": 1.1294,
+      "step": 10455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017376733039119055,
+      "loss": 1.093,
+      "step": 10460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017373213456723855,
+      "loss": 1.094,
+      "step": 10465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017369691871788487,
+      "loss": 1.047,
+      "step": 10470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017366168285269393,
+      "loss": 0.9834,
+      "step": 10475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001736264269812357,
+      "loss": 1.0605,
+      "step": 10480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017359115111308557,
+      "loss": 1.0071,
+      "step": 10485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017355585525782435,
+      "loss": 1.0752,
+      "step": 10490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017352053942503828,
+      "loss": 0.9945,
+      "step": 10495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017348520362431898,
+      "loss": 1.0059,
+      "step": 10500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001734498478652636,
+      "loss": 1.0507,
+      "step": 10505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001734144721574746,
+      "loss": 1.117,
+      "step": 10510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017337907651055989,
+      "loss": 1.1572,
+      "step": 10515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017334366093413287,
+      "loss": 1.0091,
+      "step": 10520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017330822543781225,
+      "loss": 1.0622,
+      "step": 10525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017327277003122217,
+      "loss": 1.0245,
+      "step": 10530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017323729472399228,
+      "loss": 1.0173,
+      "step": 10535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001732017995257575,
+      "loss": 1.0354,
+      "step": 10540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017316628444615818,
+      "loss": 1.1327,
+      "step": 10545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017313074949484015,
+      "loss": 1.1031,
+      "step": 10550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001730951946814546,
+      "loss": 1.035,
+      "step": 10555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017305962001565802,
+      "loss": 0.9861,
+      "step": 10560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017302402550711243,
+      "loss": 1.1012,
+      "step": 10565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017298841116548515,
+      "loss": 1.0657,
+      "step": 10570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017295277700044896,
+      "loss": 1.0602,
+      "step": 10575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017291712302168192,
+      "loss": 1.0591,
+      "step": 10580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017288144923886757,
+      "loss": 1.0226,
+      "step": 10585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017284575566169481,
+      "loss": 1.0894,
+      "step": 10590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017281004229985784,
+      "loss": 1.1207,
+      "step": 10595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017277430916305634,
+      "loss": 1.0861,
+      "step": 10600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001727385562609953,
+      "loss": 1.0469,
+      "step": 10605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017270278360338503,
+      "loss": 1.1543,
+      "step": 10610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017266699119994136,
+      "loss": 1.0736,
+      "step": 10615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001726311790603853,
+      "loss": 1.0649,
+      "step": 10620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017259534719444339,
+      "loss": 1.0869,
+      "step": 10625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017255949561184738,
+      "loss": 1.094,
+      "step": 10630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017252362432233447,
+      "loss": 1.0225,
+      "step": 10635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017248773333564715,
+      "loss": 1.0428,
+      "step": 10640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017245182266153336,
+      "loss": 1.0885,
+      "step": 10645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017241589230974626,
+      "loss": 1.0467,
+      "step": 10650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017237994229004445,
+      "loss": 1.129,
+      "step": 10655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001723439726121918,
+      "loss": 1.031,
+      "step": 10660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017230798328595758,
+      "loss": 1.0598,
+      "step": 10665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017227197432111635,
+      "loss": 1.1726,
+      "step": 10670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017223594572744806,
+      "loss": 1.0586,
+      "step": 10675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001721998975147379,
+      "loss": 1.0251,
+      "step": 10680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001721638296927765,
+      "loss": 1.0656,
+      "step": 10685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017212774227135972,
+      "loss": 1.1257,
+      "step": 10690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017209163526028882,
+      "loss": 1.1391,
+      "step": 10695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017205550866937036,
+      "loss": 1.0993,
+      "step": 10700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017201936250841612,
+      "loss": 1.0638,
+      "step": 10705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001719831967872433,
+      "loss": 1.0655,
+      "step": 10710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017194701151567444,
+      "loss": 1.1308,
+      "step": 10715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017191080670353733,
+      "loss": 1.0275,
+      "step": 10720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017187458236066504,
+      "loss": 1.1002,
+      "step": 10725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017183833849689596,
+      "loss": 1.1032,
+      "step": 10730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017180207512207388,
+      "loss": 1.0557,
+      "step": 10735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017176579224604778,
+      "loss": 1.1544,
+      "step": 10740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001717294898786719,
+      "loss": 0.9852,
+      "step": 10745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000171693168029806,
+      "loss": 1.0723,
+      "step": 10750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001716568267093148,
+      "loss": 1.1809,
+      "step": 10755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001716204659270686,
+      "loss": 0.9814,
+      "step": 10760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001715840856929428,
+      "loss": 1.0341,
+      "step": 10765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017154768601681817,
+      "loss": 1.0411,
+      "step": 10770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017151126690858072,
+      "loss": 0.991,
+      "step": 10775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017147482837812184,
+      "loss": 1.0587,
+      "step": 10780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017143837043533802,
+      "loss": 1.0361,
+      "step": 10785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017140189309013115,
+      "loss": 1.0721,
+      "step": 10790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017136539635240837,
+      "loss": 1.1646,
+      "step": 10795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017132888023208207,
+      "loss": 1.0641,
+      "step": 10800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017129234473906986,
+      "loss": 1.0283,
+      "step": 10805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017125578988329472,
+      "loss": 1.1348,
+      "step": 10810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001712192156746848,
+      "loss": 1.0673,
+      "step": 10815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017118262212317354,
+      "loss": 1.0256,
+      "step": 10820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001711460092386996,
+      "loss": 1.027,
+      "step": 10825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017110937703120692,
+      "loss": 1.0684,
+      "step": 10830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017107272551064473,
+      "loss": 1.0521,
+      "step": 10835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017103605468696742,
+      "loss": 1.0737,
+      "step": 10840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001709993645701347,
+      "loss": 1.1187,
+      "step": 10845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017096265517011142,
+      "loss": 1.0724,
+      "step": 10850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001709259264968678,
+      "loss": 1.0554,
+      "step": 10855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017088917856037918,
+      "loss": 1.071,
+      "step": 10860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001708524113706262,
+      "loss": 1.0793,
+      "step": 10865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017081562493759473,
+      "loss": 1.0794,
+      "step": 10870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017077881927127578,
+      "loss": 1.0713,
+      "step": 10875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017074199438166566,
+      "loss": 0.9661,
+      "step": 10880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001707051502787659,
+      "loss": 1.1395,
+      "step": 10885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017066828697258327,
+      "loss": 1.0334,
+      "step": 10890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017063140447312969,
+      "loss": 1.0667,
+      "step": 10895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017059450279042232,
+      "loss": 0.9435,
+      "step": 10900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001705575819344835,
+      "loss": 1.1204,
+      "step": 10905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017052064191534088,
+      "loss": 0.9949,
+      "step": 10910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001704836827430272,
+      "loss": 1.0463,
+      "step": 10915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017044670442758052,
+      "loss": 1.141,
+      "step": 10920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017040970697904394,
+      "loss": 1.1366,
+      "step": 10925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017037269040746583,
+      "loss": 1.0973,
+      "step": 10930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017033565472289984,
+      "loss": 1.0018,
+      "step": 10935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017029859993540473,
+      "loss": 1.0307,
+      "step": 10940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017026152605504442,
+      "loss": 1.114,
+      "step": 10945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017022443309188808,
+      "loss": 1.1354,
+      "step": 10950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017018732105600999,
+      "loss": 0.9942,
+      "step": 10955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017015018995748972,
+      "loss": 1.142,
+      "step": 10960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001701130398064119,
+      "loss": 1.0744,
+      "step": 10965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001700758706128664,
+      "loss": 1.0051,
+      "step": 10970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017003868238694829,
+      "loss": 1.0617,
+      "step": 10975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017000147513875774,
+      "loss": 1.1149,
+      "step": 10980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016996424887840008,
+      "loss": 1.0244,
+      "step": 10985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016992700361598585,
+      "loss": 1.1042,
+      "step": 10990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016988973936163078,
+      "loss": 1.0517,
+      "step": 10995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001698524561254557,
+      "loss": 1.0554,
+      "step": 11000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016981515391758666,
+      "loss": 1.0733,
+      "step": 11005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016977783274815466,
+      "loss": 1.0433,
+      "step": 11010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016974049262729617,
+      "loss": 1.0617,
+      "step": 11015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016970313356515256,
+      "loss": 1.1643,
+      "step": 11020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016966575557187042,
+      "loss": 1.0726,
+      "step": 11025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016962835865760157,
+      "loss": 1.0293,
+      "step": 11030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016959094283250278,
+      "loss": 1.1168,
+      "step": 11035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016955350810673612,
+      "loss": 1.0544,
+      "step": 11040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016951605449046876,
+      "loss": 1.0167,
+      "step": 11045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016947858199387294,
+      "loss": 1.1325,
+      "step": 11050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016944109062712608,
+      "loss": 1.1149,
+      "step": 11055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016940358040041067,
+      "loss": 1.0613,
+      "step": 11060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016936605132391442,
+      "loss": 1.0369,
+      "step": 11065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016932850340783007,
+      "loss": 1.0442,
+      "step": 11070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016929093666235548,
+      "loss": 1.0915,
+      "step": 11075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016925335109769374,
+      "loss": 0.9899,
+      "step": 11080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001692157467240529,
+      "loss": 1.1513,
+      "step": 11085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001691781235516462,
+      "loss": 1.0328,
+      "step": 11090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000169140481590692,
+      "loss": 1.1203,
+      "step": 11095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001691028208514136,
+      "loss": 1.1032,
+      "step": 11100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001690651413440397,
+      "loss": 1.0514,
+      "step": 11105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016902744307880382,
+      "loss": 1.0804,
+      "step": 11110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016898972606594473,
+      "loss": 1.0344,
+      "step": 11115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016895199031570623,
+      "loss": 1.0962,
+      "step": 11120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016891423583833724,
+      "loss": 1.075,
+      "step": 11125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001688764626440917,
+      "loss": 1.0056,
+      "step": 11130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016883867074322874,
+      "loss": 1.0226,
+      "step": 11135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016880086014601248,
+      "loss": 1.0332,
+      "step": 11140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001687630308627122,
+      "loss": 1.0638,
+      "step": 11145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016872518290360214,
+      "loss": 1.0057,
+      "step": 11150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016868731627896173,
+      "loss": 1.0786,
+      "step": 11155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001686494309990754,
+      "loss": 1.0649,
+      "step": 11160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016861152707423264,
+      "loss": 1.0194,
+      "step": 11165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001685736045147281,
+      "loss": 1.1267,
+      "step": 11170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016853566333086136,
+      "loss": 0.9675,
+      "step": 11175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001684977035329371,
+      "loss": 1.1057,
+      "step": 11180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016845972513126518,
+      "loss": 1.0367,
+      "step": 11185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016842172813616029,
+      "loss": 1.0302,
+      "step": 11190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016838371255794234,
+      "loss": 1.0392,
+      "step": 11195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016834567840693623,
+      "loss": 1.0329,
+      "step": 11200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016830762569347191,
+      "loss": 0.9874,
+      "step": 11205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016826955442788438,
+      "loss": 0.9987,
+      "step": 11210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016823146462051364,
+      "loss": 1.0931,
+      "step": 11215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016819335628170476,
+      "loss": 1.06,
+      "step": 11220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016815522942180782,
+      "loss": 1.0694,
+      "step": 11225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016811708405117806,
+      "loss": 1.097,
+      "step": 11230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001680789201801755,
+      "loss": 1.0368,
+      "step": 11235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016804073781916538,
+      "loss": 1.0322,
+      "step": 11240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016800253697851785,
+      "loss": 1.0542,
+      "step": 11245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016796431766860826,
+      "loss": 1.0591,
+      "step": 11250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001679260798998167,
+      "loss": 1.032,
+      "step": 11255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001678878236825285,
+      "loss": 0.9596,
+      "step": 11260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001678495490271339,
+      "loss": 1.1499,
+      "step": 11265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016781125594402817,
+      "loss": 1.0385,
+      "step": 11270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016777294444361157,
+      "loss": 1.0725,
+      "step": 11275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016773461453628943,
+      "loss": 1.0684,
+      "step": 11280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016769626623247197,
+      "loss": 1.0652,
+      "step": 11285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016765789954257447,
+      "loss": 0.9965,
+      "step": 11290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016761951447701718,
+      "loss": 1.0424,
+      "step": 11295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001675811110462254,
+      "loss": 1.0728,
+      "step": 11300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016754268926062938,
+      "loss": 1.0953,
+      "step": 11305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016750424913066427,
+      "loss": 1.1191,
+      "step": 11310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016746579066677036,
+      "loss": 1.0527,
+      "step": 11315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001674273138793928,
+      "loss": 1.0268,
+      "step": 11320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016738881877898174,
+      "loss": 1.0805,
+      "step": 11325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001673503053759924,
+      "loss": 1.032,
+      "step": 11330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016731177368088483,
+      "loss": 0.9756,
+      "step": 11335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001672732237041241,
+      "loss": 1.1145,
+      "step": 11340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001672346554561803,
+      "loss": 1.1231,
+      "step": 11345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001671960689475284,
+      "loss": 1.0336,
+      "step": 11350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016715746418864837,
+      "loss": 1.0601,
+      "step": 11355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016711884119002517,
+      "loss": 1.1045,
+      "step": 11360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016708019996214863,
+      "loss": 1.0638,
+      "step": 11365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001670415405155136,
+      "loss": 1.1021,
+      "step": 11370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016700286286061987,
+      "loss": 1.0261,
+      "step": 11375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016696416700797214,
+      "loss": 1.0588,
+      "step": 11380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001669254529680801,
+      "loss": 1.0451,
+      "step": 11385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016688672075145831,
+      "loss": 1.0668,
+      "step": 11390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016684797036862635,
+      "loss": 1.0468,
+      "step": 11395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016680920183010872,
+      "loss": 1.0722,
+      "step": 11400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016677041514643476,
+      "loss": 1.0765,
+      "step": 11405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001667316103281388,
+      "loss": 1.162,
+      "step": 11410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016669278738576019,
+      "loss": 1.0987,
+      "step": 11415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016665394632984302,
+      "loss": 1.1128,
+      "step": 11420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001666150871709364,
+      "loss": 1.0165,
+      "step": 11425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001665762099195944,
+      "loss": 1.0056,
+      "step": 11430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016653731458637593,
+      "loss": 0.9907,
+      "step": 11435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001664984011818448,
+      "loss": 1.0796,
+      "step": 11440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016645946971656985,
+      "loss": 1.0061,
+      "step": 11445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016642052020112463,
+      "loss": 1.0139,
+      "step": 11450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016638155264608772,
+      "loss": 1.0964,
+      "step": 11455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016634256706204265,
+      "loss": 1.0438,
+      "step": 11460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016630356345957769,
+      "loss": 1.1069,
+      "step": 11465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016626454184928612,
+      "loss": 1.0651,
+      "step": 11470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016622550224176612,
+      "loss": 1.1476,
+      "step": 11475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016618644464762064,
+      "loss": 1.0643,
+      "step": 11480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016614736907745768,
+      "loss": 1.0212,
+      "step": 11485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001661082755418899,
+      "loss": 1.1331,
+      "step": 11490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016606916405153513,
+      "loss": 1.0847,
+      "step": 11495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016603003461701583,
+      "loss": 1.0255,
+      "step": 11500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016599088724895947,
+      "loss": 1.0759,
+      "step": 11505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016595172195799828,
+      "loss": 1.0984,
+      "step": 11510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001659125387547695,
+      "loss": 1.073,
+      "step": 11515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016587333764991508,
+      "loss": 1.0789,
+      "step": 11520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016583411865408196,
+      "loss": 1.0268,
+      "step": 11525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016579488177792187,
+      "loss": 1.0709,
+      "step": 11530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016575562703209144,
+      "loss": 1.0498,
+      "step": 11535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016571635442725208,
+      "loss": 1.0779,
+      "step": 11540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016567706397407016,
+      "loss": 1.0348,
+      "step": 11545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016563775568321677,
+      "loss": 1.0354,
+      "step": 11550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016559842956536799,
+      "loss": 1.0427,
+      "step": 11555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016555908563120457,
+      "loss": 1.1559,
+      "step": 11560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016551972389141227,
+      "loss": 1.0096,
+      "step": 11565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016548034435668154,
+      "loss": 0.9997,
+      "step": 11570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001654409470377078,
+      "loss": 1.0042,
+      "step": 11575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016540153194519115,
+      "loss": 1.0752,
+      "step": 11580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016536209908983668,
+      "loss": 1.0325,
+      "step": 11585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016532264848235416,
+      "loss": 1.1081,
+      "step": 11590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016528318013345832,
+      "loss": 1.0106,
+      "step": 11595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016524369405386853,
+      "loss": 1.092,
+      "step": 11600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016520419025430915,
+      "loss": 1.0487,
+      "step": 11605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016516466874550925,
+      "loss": 1.1423,
+      "step": 11610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016512512953820272,
+      "loss": 1.0689,
+      "step": 11615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016508557264312835,
+      "loss": 1.0832,
+      "step": 11620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016504599807102958,
+      "loss": 1.0373,
+      "step": 11625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001650064058326548,
+      "loss": 1.0593,
+      "step": 11630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016496679593875705,
+      "loss": 1.0384,
+      "step": 11635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001649271684000943,
+      "loss": 1.041,
+      "step": 11640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001648875232274293,
+      "loss": 1.0803,
+      "step": 11645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016484786043152947,
+      "loss": 1.0552,
+      "step": 11650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016480818002316712,
+      "loss": 1.0733,
+      "step": 11655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016476848201311928,
+      "loss": 1.0528,
+      "step": 11660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016472876641216786,
+      "loss": 1.1617,
+      "step": 11665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016468903323109947,
+      "loss": 1.0528,
+      "step": 11670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016464928248070547,
+      "loss": 1.0346,
+      "step": 11675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016460951417178205,
+      "loss": 1.0367,
+      "step": 11680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016456972831513017,
+      "loss": 1.0163,
+      "step": 11685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016452992492155553,
+      "loss": 1.1528,
+      "step": 11690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001644901040018686,
+      "loss": 1.0737,
+      "step": 11695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001644502655668846,
+      "loss": 1.2226,
+      "step": 11700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001644104096274235,
+      "loss": 1.0894,
+      "step": 11705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016437053619431006,
+      "loss": 1.1269,
+      "step": 11710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016433064527837375,
+      "loss": 1.014,
+      "step": 11715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001642907368904489,
+      "loss": 1.0894,
+      "step": 11720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016425081104137434,
+      "loss": 1.0442,
+      "step": 11725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016421086774199396,
+      "loss": 1.0189,
+      "step": 11730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016417090700315606,
+      "loss": 1.0383,
+      "step": 11735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000164130928835714,
+      "loss": 1.0017,
+      "step": 11740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016409093325052562,
+      "loss": 1.0848,
+      "step": 11745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016405092025845362,
+      "loss": 1.05,
+      "step": 11750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016401088987036538,
+      "loss": 1.0706,
+      "step": 11755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016397084209713304,
+      "loss": 1.0387,
+      "step": 11760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016393077694963344,
+      "loss": 1.0628,
+      "step": 11765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016389069443874813,
+      "loss": 1.0059,
+      "step": 11770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016385059457536336,
+      "loss": 1.0952,
+      "step": 11775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001638104773703702,
+      "loss": 1.1202,
+      "step": 11780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016377034283466423,
+      "loss": 0.9865,
+      "step": 11785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016373019097914594,
+      "loss": 0.9898,
+      "step": 11790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016369002181472047,
+      "loss": 1.046,
+      "step": 11795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016364983535229754,
+      "loss": 1.0893,
+      "step": 11800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001636096316027917,
+      "loss": 1.0112,
+      "step": 11805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016356941057712218,
+      "loss": 1.0174,
+      "step": 11810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016352917228621284,
+      "loss": 1.0089,
+      "step": 11815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001634889167409923,
+      "loss": 1.022,
+      "step": 11820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016344864395239378,
+      "loss": 1.0968,
+      "step": 11825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016340835393135528,
+      "loss": 1.1273,
+      "step": 11830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016336804668881945,
+      "loss": 1.1258,
+      "step": 11835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016332772223573352,
+      "loss": 0.9953,
+      "step": 11840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016328738058304956,
+      "loss": 1.0602,
+      "step": 11845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016324702174172416,
+      "loss": 1.0642,
+      "step": 11850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016320664572271873,
+      "loss": 1.0208,
+      "step": 11855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001631662525369992,
+      "loss": 1.0466,
+      "step": 11860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016312584219553626,
+      "loss": 1.1004,
+      "step": 11865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001630854147093052,
+      "loss": 1.1379,
+      "step": 11870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000163044970089286,
+      "loss": 1.0261,
+      "step": 11875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016300450834646327,
+      "loss": 1.0207,
+      "step": 11880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016296402949182628,
+      "loss": 1.1076,
+      "step": 11885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000162923533536369,
+      "loss": 1.0206,
+      "step": 11890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001628830204910899,
+      "loss": 1.0338,
+      "step": 11895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016284249036699228,
+      "loss": 1.0762,
+      "step": 11900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016280194317508398,
+      "loss": 1.0633,
+      "step": 11905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016276137892637743,
+      "loss": 1.0292,
+      "step": 11910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001627207976318898,
+      "loss": 1.049,
+      "step": 11915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001626801993026428,
+      "loss": 1.0657,
+      "step": 11920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001626395839496628,
+      "loss": 1.0183,
+      "step": 11925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001625989515839808,
+      "loss": 1.0882,
+      "step": 11930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016255830221663243,
+      "loss": 1.1033,
+      "step": 11935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016251763585865793,
+      "loss": 1.0454,
+      "step": 11940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001624769525211021,
+      "loss": 1.1028,
+      "step": 11945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001624362522150144,
+      "loss": 1.0967,
+      "step": 11950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016239553495144896,
+      "loss": 1.1508,
+      "step": 11955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001623548007414644,
+      "loss": 0.979,
+      "step": 11960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016231404959612405,
+      "loss": 1.0233,
+      "step": 11965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001622732815264957,
+      "loss": 1.0314,
+      "step": 11970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001622324965436519,
+      "loss": 1.0959,
+      "step": 11975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016219169465866965,
+      "loss": 1.0261,
+      "step": 11980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016215087588263068,
+      "loss": 1.0652,
+      "step": 11985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016211004022662117,
+      "loss": 1.1337,
+      "step": 11990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016206918770173196,
+      "loss": 1.0871,
+      "step": 11995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001620283183190585,
+      "loss": 1.0926,
+      "step": 12000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001619874320897007,
+      "loss": 0.9973,
+      "step": 12005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001619465290247632,
+      "loss": 1.0675,
+      "step": 12010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016190560913535514,
+      "loss": 1.0662,
+      "step": 12015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001618646724325902,
+      "loss": 1.0407,
+      "step": 12020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016182371892758658,
+      "loss": 1.0701,
+      "step": 12025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016178274863146722,
+      "loss": 1.1339,
+      "step": 12030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016174176155535946,
+      "loss": 1.0564,
+      "step": 12035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016170075771039527,
+      "loss": 1.0982,
+      "step": 12040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001616597371077112,
+      "loss": 1.0472,
+      "step": 12045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016161869975844822,
+      "loss": 1.112,
+      "step": 12050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016157764567375197,
+      "loss": 1.0499,
+      "step": 12055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016153657486477264,
+      "loss": 1.1265,
+      "step": 12060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001614954873426649,
+      "loss": 1.0425,
+      "step": 12065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000161454383118588,
+      "loss": 1.0358,
+      "step": 12070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016141326220370564,
+      "loss": 1.1033,
+      "step": 12075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016137212460918624,
+      "loss": 1.0941,
+      "step": 12080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016133097034620253,
+      "loss": 1.0244,
+      "step": 12085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016128979942593193,
+      "loss": 1.0387,
+      "step": 12090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016124861185955628,
+      "loss": 1.091,
+      "step": 12095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016120740765826206,
+      "loss": 1.0611,
+      "step": 12100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001611661868332401,
+      "loss": 1.027,
+      "step": 12105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016112494939568589,
+      "loss": 1.0849,
+      "step": 12110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001610836953567994,
+      "loss": 1.0287,
+      "step": 12115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016104242472778503,
+      "loss": 1.0451,
+      "step": 12120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001610011375198518,
+      "loss": 1.0043,
+      "step": 12125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016095983374421314,
+      "loss": 1.0372,
+      "step": 12130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016091851341208706,
+      "loss": 1.0414,
+      "step": 12135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016087717653469598,
+      "loss": 1.1234,
+      "step": 12140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016083582312326685,
+      "loss": 1.07,
+      "step": 12145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001607944531890312,
+      "loss": 1.0498,
+      "step": 12150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016075306674322488,
+      "loss": 1.066,
+      "step": 12155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016071166379708833,
+      "loss": 1.0404,
+      "step": 12160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016067024436186648,
+      "loss": 1.062,
+      "step": 12165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001606288084488087,
+      "loss": 1.1035,
+      "step": 12170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016058735606916882,
+      "loss": 1.0286,
+      "step": 12175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016054588723420521,
+      "loss": 1.0172,
+      "step": 12180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016050440195518063,
+      "loss": 1.0529,
+      "step": 12185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016046290024336236,
+      "loss": 1.0007,
+      "step": 12190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016042138211002215,
+      "loss": 1.0681,
+      "step": 12195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001603798475664361,
+      "loss": 1.0855,
+      "step": 12200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016033829662388498,
+      "loss": 1.0965,
+      "step": 12205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001602967292936538,
+      "loss": 1.0717,
+      "step": 12210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001602551455870321,
+      "loss": 1.0609,
+      "step": 12215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016021354551531394,
+      "loss": 0.9945,
+      "step": 12220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016017192908979774,
+      "loss": 1.0446,
+      "step": 12225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016013029632178634,
+      "loss": 1.0,
+      "step": 12230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001600886472225871,
+      "loss": 1.0896,
+      "step": 12235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016004698180351176,
+      "loss": 1.0985,
+      "step": 12240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001600053000758765,
+      "loss": 1.0819,
+      "step": 12245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000159963602051002,
+      "loss": 1.0969,
+      "step": 12250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015992188774021325,
+      "loss": 1.0584,
+      "step": 12255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015988015715483967,
+      "loss": 1.0449,
+      "step": 12260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015983841030621526,
+      "loss": 1.0808,
+      "step": 12265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001597966472056782,
+      "loss": 1.1722,
+      "step": 12270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015975486786457137,
+      "loss": 1.0571,
+      "step": 12275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015971307229424174,
+      "loss": 1.0479,
+      "step": 12280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015967126050604093,
+      "loss": 1.1316,
+      "step": 12285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015962943251132488,
+      "loss": 1.1198,
+      "step": 12290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015958758832145388,
+      "loss": 0.9917,
+      "step": 12295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015954572794779273,
+      "loss": 1.0472,
+      "step": 12300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015950385140171057,
+      "loss": 1.0713,
+      "step": 12305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015946195869458093,
+      "loss": 1.0398,
+      "step": 12310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015942004983778168,
+      "loss": 1.06,
+      "step": 12315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015937812484269513,
+      "loss": 1.044,
+      "step": 12320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015933618372070805,
+      "loss": 0.9675,
+      "step": 12325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001592942264832114,
+      "loss": 1.0668,
+      "step": 12330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001592522531416007,
+      "loss": 0.9494,
+      "step": 12335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015921026370727573,
+      "loss": 1.0085,
+      "step": 12340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001591682581916407,
+      "loss": 1.0809,
+      "step": 12345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001591262366061042,
+      "loss": 1.0572,
+      "step": 12350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001590841989620791,
+      "loss": 1.0494,
+      "step": 12355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015904214527098272,
+      "loss": 1.0348,
+      "step": 12360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015900007554423665,
+      "loss": 1.1366,
+      "step": 12365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015895798979326692,
+      "loss": 1.0334,
+      "step": 12370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015891588802950394,
+      "loss": 1.1262,
+      "step": 12375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015887377026438227,
+      "loss": 1.0053,
+      "step": 12380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001588316365093411,
+      "loss": 0.9844,
+      "step": 12385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015878948677582372,
+      "loss": 1.0509,
+      "step": 12390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001587473210752779,
+      "loss": 1.0638,
+      "step": 12395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001587051394191557,
+      "loss": 1.0847,
+      "step": 12400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001586629418189135,
+      "loss": 1.0602,
+      "step": 12405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015862072828601203,
+      "loss": 1.1459,
+      "step": 12410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015857849883191635,
+      "loss": 1.1086,
+      "step": 12415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015853625346809585,
+      "loss": 1.047,
+      "step": 12420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015849399220602424,
+      "loss": 1.0071,
+      "step": 12425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015845171505717947,
+      "loss": 1.0305,
+      "step": 12430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015840942203304396,
+      "loss": 0.9619,
+      "step": 12435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001583671131451043,
+      "loss": 1.0425,
+      "step": 12440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015832478840485147,
+      "loss": 1.1414,
+      "step": 12445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001582824478237807,
+      "loss": 1.0897,
+      "step": 12450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015824009141339156,
+      "loss": 1.1626,
+      "step": 12455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015819771918518794,
+      "loss": 1.1347,
+      "step": 12460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015815533115067794,
+      "loss": 1.1648,
+      "step": 12465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001581129273213741,
+      "loss": 1.0929,
+      "step": 12470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015807050770879304,
+      "loss": 1.0203,
+      "step": 12475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015802807232445586,
+      "loss": 1.0773,
+      "step": 12480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015798562117988785,
+      "loss": 1.1076,
+      "step": 12485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001579431542866186,
+      "loss": 1.0848,
+      "step": 12490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000157900671656182,
+      "loss": 1.0672,
+      "step": 12495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015785817330011615,
+      "loss": 1.0444,
+      "step": 12500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001578156592299635,
+      "loss": 1.0385,
+      "step": 12505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015777312945727072,
+      "loss": 1.1397,
+      "step": 12510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001577305839935887,
+      "loss": 1.0314,
+      "step": 12515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015768802285047272,
+      "loss": 1.061,
+      "step": 12520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015764544603948224,
+      "loss": 1.0831,
+      "step": 12525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001576028535721809,
+      "loss": 1.0409,
+      "step": 12530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015756024546013676,
+      "loss": 1.0264,
+      "step": 12535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000157517621714922,
+      "loss": 1.0486,
+      "step": 12540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001574749823481131,
+      "loss": 1.1393,
+      "step": 12545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001574323273712908,
+      "loss": 0.9918,
+      "step": 12550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015738965679603997,
+      "loss": 0.9591,
+      "step": 12555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015734697063394986,
+      "loss": 1.0308,
+      "step": 12560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015730426889661386,
+      "loss": 1.1057,
+      "step": 12565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015726155159562964,
+      "loss": 1.0179,
+      "step": 12570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015721881874259904,
+      "loss": 1.033,
+      "step": 12575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001571760703491282,
+      "loss": 1.0896,
+      "step": 12580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015713330642682744,
+      "loss": 1.078,
+      "step": 12585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015709052698731127,
+      "loss": 1.0376,
+      "step": 12590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015704773204219848,
+      "loss": 1.0438,
+      "step": 12595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015700492160311197,
+      "loss": 0.9981,
+      "step": 12600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000156962095681679,
+      "loss": 1.1354,
+      "step": 12605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015691925428953089,
+      "loss": 1.1314,
+      "step": 12610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015687639743830324,
+      "loss": 1.1792,
+      "step": 12615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001568335251396358,
+      "loss": 1.0431,
+      "step": 12620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015679063740517256,
+      "loss": 1.0919,
+      "step": 12625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001567477342465617,
+      "loss": 1.0962,
+      "step": 12630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001567048156754556,
+      "loss": 1.0615,
+      "step": 12635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015666188170351068,
+      "loss": 1.0596,
+      "step": 12640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001566189323423878,
+      "loss": 1.0684,
+      "step": 12645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015657596760375179,
+      "loss": 1.0383,
+      "step": 12650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015653298749927178,
+      "loss": 1.0603,
+      "step": 12655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015648999204062098,
+      "loss": 1.0703,
+      "step": 12660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015644698123947682,
+      "loss": 1.062,
+      "step": 12665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015640395510752093,
+      "loss": 1.1088,
+      "step": 12670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000156360913656439,
+      "loss": 1.084,
+      "step": 12675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015631785689792098,
+      "loss": 1.0751,
+      "step": 12680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001562747848436609,
+      "loss": 1.1549,
+      "step": 12685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015623169750535708,
+      "loss": 0.9869,
+      "step": 12690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001561885948947118,
+      "loss": 1.124,
+      "step": 12695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015614547702343166,
+      "loss": 0.9965,
+      "step": 12700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001561023439032273,
+      "loss": 1.0271,
+      "step": 12705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001560591955458135,
+      "loss": 1.0701,
+      "step": 12710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015601603196290924,
+      "loss": 1.1278,
+      "step": 12715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015597285316623757,
+      "loss": 1.1064,
+      "step": 12720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001559296591675258,
+      "loss": 1.1004,
+      "step": 12725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015588644997850519,
+      "loss": 1.0756,
+      "step": 12730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015584322561091124,
+      "loss": 0.9904,
+      "step": 12735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015579998607648352,
+      "loss": 1.0268,
+      "step": 12740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015575673138696578,
+      "loss": 1.1451,
+      "step": 12745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001557134615541058,
+      "loss": 1.1465,
+      "step": 12750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015567017658965564,
+      "loss": 1.1158,
+      "step": 12755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001556268765053712,
+      "loss": 1.0284,
+      "step": 12760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015558356131301272,
+      "loss": 1.1098,
+      "step": 12765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015554023102434445,
+      "loss": 1.0548,
+      "step": 12770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015549688565113476,
+      "loss": 1.0587,
+      "step": 12775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001554535252051561,
+      "loss": 1.1288,
+      "step": 12780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000155410149698185,
+      "loss": 1.0819,
+      "step": 12785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015536675914200213,
+      "loss": 1.0428,
+      "step": 12790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001553233535483922,
+      "loss": 1.077,
+      "step": 12795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015527993292914407,
+      "loss": 1.0254,
+      "step": 12800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001552364972960506,
+      "loss": 1.0026,
+      "step": 12805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015519304666090876,
+      "loss": 1.0738,
+      "step": 12810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001551495810355196,
+      "loss": 1.1154,
+      "step": 12815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015510610043168824,
+      "loss": 1.0234,
+      "step": 12820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015506260486122385,
+      "loss": 1.078,
+      "step": 12825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015501909433593972,
+      "loss": 1.0976,
+      "step": 12830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015497556886765316,
+      "loss": 1.0619,
+      "step": 12835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015493202846818553,
+      "loss": 1.0863,
+      "step": 12840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015488847314936224,
+      "loss": 1.0133,
+      "step": 12845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001548449029230128,
+      "loss": 0.9726,
+      "step": 12850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001548013178009707,
+      "loss": 1.0892,
+      "step": 12855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015475771779507356,
+      "loss": 1.0856,
+      "step": 12860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015471410291716293,
+      "loss": 1.0923,
+      "step": 12865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015467047317908455,
+      "loss": 1.0328,
+      "step": 12870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015462682859268803,
+      "loss": 0.9618,
+      "step": 12875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015458316916982715,
+      "loss": 1.0252,
+      "step": 12880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015453949492235964,
+      "loss": 0.9738,
+      "step": 12885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001544958058621473,
+      "loss": 1.1019,
+      "step": 12890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015445210200105592,
+      "loss": 1.0605,
+      "step": 12895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001544083833509553,
+      "loss": 1.0122,
+      "step": 12900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015436464992371932,
+      "loss": 0.9601,
+      "step": 12905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001543209017312258,
+      "loss": 1.0791,
+      "step": 12910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015427713878535663,
+      "loss": 1.1111,
+      "step": 12915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015423336109799765,
+      "loss": 1.1112,
+      "step": 12920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015418956868103878,
+      "loss": 1.0873,
+      "step": 12925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015414576154637385,
+      "loss": 1.1185,
+      "step": 12930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015410193970590073,
+      "loss": 1.0208,
+      "step": 12935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015405810317152134,
+      "loss": 1.024,
+      "step": 12940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015401425195514146,
+      "loss": 1.0212,
+      "step": 12945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015397038606867103,
+      "loss": 1.0881,
+      "step": 12950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015392650552402374,
+      "loss": 1.0419,
+      "step": 12955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015388261033311753,
+      "loss": 1.04,
+      "step": 12960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015383870050787412,
+      "loss": 1.0422,
+      "step": 12965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015379477606021933,
+      "loss": 1.0576,
+      "step": 12970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015375083700208282,
+      "loss": 1.065,
+      "step": 12975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015370688334539833,
+      "loss": 1.0505,
+      "step": 12980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015366291510210354,
+      "loss": 1.1005,
+      "step": 12985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015361893228414003,
+      "loss": 1.0852,
+      "step": 12990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015357493490345342,
+      "loss": 1.2097,
+      "step": 12995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015353092297199327,
+      "loss": 1.0543,
+      "step": 13000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015348689650171305,
+      "loss": 1.0732,
+      "step": 13005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015344285550457018,
+      "loss": 1.093,
+      "step": 13010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015339879999252606,
+      "loss": 1.0503,
+      "step": 13015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001533547299775461,
+      "loss": 1.1156,
+      "step": 13020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015331064547159944,
+      "loss": 1.1332,
+      "step": 13025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015326654648665938,
+      "loss": 0.9897,
+      "step": 13030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015322243303470302,
+      "loss": 0.9512,
+      "step": 13035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015317830512771142,
+      "loss": 1.1189,
+      "step": 13040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001531341627776696,
+      "loss": 0.9897,
+      "step": 13045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015309000599656646,
+      "loss": 1.0516,
+      "step": 13050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015304583479639478,
+      "loss": 0.98,
+      "step": 13055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015300164918915138,
+      "loss": 1.0644,
+      "step": 13060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015295744918683696,
+      "loss": 1.0332,
+      "step": 13065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015291323480145597,
+      "loss": 1.0145,
+      "step": 13070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000152869006045017,
+      "loss": 0.9817,
+      "step": 13075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015282476292953237,
+      "loss": 1.0523,
+      "step": 13080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001527805054670184,
+      "loss": 1.0293,
+      "step": 13085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015273623366949523,
+      "loss": 1.1496,
+      "step": 13090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015269194754898697,
+      "loss": 1.0444,
+      "step": 13095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015264764711752157,
+      "loss": 1.0438,
+      "step": 13100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015260333238713088,
+      "loss": 1.0975,
+      "step": 13105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015255900336985062,
+      "loss": 1.0362,
+      "step": 13110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015251466007772045,
+      "loss": 1.0738,
+      "step": 13115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015247030252278377,
+      "loss": 0.9671,
+      "step": 13120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000152425930717088,
+      "loss": 1.0807,
+      "step": 13125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001523815446726844,
+      "loss": 1.0393,
+      "step": 13130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015233714440162808,
+      "loss": 1.0638,
+      "step": 13135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001522927299159779,
+      "loss": 1.0394,
+      "step": 13140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015224830122779679,
+      "loss": 1.0758,
+      "step": 13145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001522038583491514,
+      "loss": 1.059,
+      "step": 13150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015215940129211224,
+      "loss": 1.0288,
+      "step": 13155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001521149300687537,
+      "loss": 1.069,
+      "step": 13160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001520704446911541,
+      "loss": 1.0558,
+      "step": 13165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015202594517139544,
+      "loss": 1.0036,
+      "step": 13170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001519814315215637,
+      "loss": 1.076,
+      "step": 13175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015193690375374852,
+      "loss": 1.0702,
+      "step": 13180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015189236188004363,
+      "loss": 1.0632,
+      "step": 13185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015184780591254637,
+      "loss": 1.0536,
+      "step": 13190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015180323586335806,
+      "loss": 1.0451,
+      "step": 13195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015175865174458367,
+      "loss": 1.0509,
+      "step": 13200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001517140535683322,
+      "loss": 1.0305,
+      "step": 13205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015166944134671634,
+      "loss": 1.0793,
+      "step": 13210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015162481509185257,
+      "loss": 1.0426,
+      "step": 13215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015158017481586134,
+      "loss": 1.0461,
+      "step": 13220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001515355205308667,
+      "loss": 1.1281,
+      "step": 13225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001514908522489966,
+      "loss": 1.0555,
+      "step": 13230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001514461699823829,
+      "loss": 1.0629,
+      "step": 13235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015140147374316106,
+      "loss": 1.0857,
+      "step": 13240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015135676354347045,
+      "loss": 1.0637,
+      "step": 13245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015131203939545423,
+      "loss": 1.0243,
+      "step": 13250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001512673013112593,
+      "loss": 1.0476,
+      "step": 13255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015122254930303642,
+      "loss": 1.0002,
+      "step": 13260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015117778338294,
+      "loss": 1.0315,
+      "step": 13265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015113300356312843,
+      "loss": 1.0639,
+      "step": 13270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001510882098557637,
+      "loss": 1.0171,
+      "step": 13275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015104340227301162,
+      "loss": 1.0584,
+      "step": 13280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015099858082704182,
+      "loss": 1.0284,
+      "step": 13285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015095374553002758,
+      "loss": 1.0266,
+      "step": 13290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001509088963941461,
+      "loss": 1.0756,
+      "step": 13295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001508640334315782,
+      "loss": 1.09,
+      "step": 13300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015081915665450855,
+      "loss": 1.0361,
+      "step": 13305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015077426607512553,
+      "loss": 1.0428,
+      "step": 13310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001507293617056212,
+      "loss": 1.0335,
+      "step": 13315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015068444355819158,
+      "loss": 1.0661,
+      "step": 13320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015063951164503615,
+      "loss": 1.0256,
+      "step": 13325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015059456597835835,
+      "loss": 1.0328,
+      "step": 13330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015054960657036518,
+      "loss": 1.0321,
+      "step": 13335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001505046334332676,
+      "loss": 1.0711,
+      "step": 13340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015045964657928006,
+      "loss": 0.9869,
+      "step": 13345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015041464602062087,
+      "loss": 1.1421,
+      "step": 13350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015036963176951205,
+      "loss": 1.0609,
+      "step": 13355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001503246038381793,
+      "loss": 1.0057,
+      "step": 13360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015027956223885206,
+      "loss": 0.974,
+      "step": 13365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001502345069837635,
+      "loss": 1.0747,
+      "step": 13370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001501894380851504,
+      "loss": 1.004,
+      "step": 13375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015014435555525343,
+      "loss": 1.2044,
+      "step": 13380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001500992594063168,
+      "loss": 1.046,
+      "step": 13385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015005414965058847,
+      "loss": 1.0737,
+      "step": 13390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015000902630032007,
+      "loss": 1.0965,
+      "step": 13395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014996388936776701,
+      "loss": 1.0508,
+      "step": 13400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014991873886518832,
+      "loss": 0.9723,
+      "step": 13405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014987357480484673,
+      "loss": 1.1091,
+      "step": 13410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001498283971990086,
+      "loss": 1.0224,
+      "step": 13415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014978320605994405,
+      "loss": 1.0962,
+      "step": 13420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014973800139992684,
+      "loss": 1.0474,
+      "step": 13425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014969278323123444,
+      "loss": 1.0073,
+      "step": 13430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014964755156614786,
+      "loss": 1.0044,
+      "step": 13435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000149602306416952,
+      "loss": 1.0293,
+      "step": 13440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014955704779593516,
+      "loss": 1.029,
+      "step": 13445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001495117757153895,
+      "loss": 1.0528,
+      "step": 13450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014946649018761078,
+      "loss": 1.0178,
+      "step": 13455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014942119122489836,
+      "loss": 1.0568,
+      "step": 13460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014937587883955534,
+      "loss": 1.0409,
+      "step": 13465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001493305530438883,
+      "loss": 1.0719,
+      "step": 13470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001492852138502077,
+      "loss": 1.0127,
+      "step": 13475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014923986127082746,
+      "loss": 1.1046,
+      "step": 13480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001491944953180652,
+      "loss": 1.0061,
+      "step": 13485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014914911600424213,
+      "loss": 1.0597,
+      "step": 13490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014910372334168314,
+      "loss": 1.048,
+      "step": 13495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014905831734271676,
+      "loss": 0.9974,
+      "step": 13500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014901289801967505,
+      "loss": 1.0682,
+      "step": 13505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001489674653848938,
+      "loss": 1.0801,
+      "step": 13510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014892201945071234,
+      "loss": 1.0247,
+      "step": 13515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001488765602294736,
+      "loss": 1.034,
+      "step": 13520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001488310877335242,
+      "loss": 1.0562,
+      "step": 13525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001487856019752143,
+      "loss": 1.0554,
+      "step": 13530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001487401029668977,
+      "loss": 1.0207,
+      "step": 13535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014869459072093172,
+      "loss": 1.047,
+      "step": 13540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001486490652496774,
+      "loss": 1.0301,
+      "step": 13545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014860352656549924,
+      "loss": 0.9327,
+      "step": 13550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014855797468076552,
+      "loss": 1.0279,
+      "step": 13555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014851240960784783,
+      "loss": 1.0293,
+      "step": 13560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014846683135912158,
+      "loss": 1.1313,
+      "step": 13565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014842123994696565,
+      "loss": 1.0193,
+      "step": 13570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014837563538376252,
+      "loss": 1.0356,
+      "step": 13575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001483300176818982,
+      "loss": 1.0761,
+      "step": 13580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014828438685376237,
+      "loss": 0.9864,
+      "step": 13585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014823874291174812,
+      "loss": 0.9976,
+      "step": 13590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001481930858682523,
+      "loss": 1.0158,
+      "step": 13595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014814741573567514,
+      "loss": 1.0162,
+      "step": 13600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014810173252642055,
+      "loss": 1.019,
+      "step": 13605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014805603625289585,
+      "loss": 1.0016,
+      "step": 13610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001480103269275121,
+      "loss": 1.0892,
+      "step": 13615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001479646045626837,
+      "loss": 1.0501,
+      "step": 13620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014791886917082876,
+      "loss": 1.0575,
+      "step": 13625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014787312076436883,
+      "loss": 1.0397,
+      "step": 13630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014782735935572904,
+      "loss": 1.0976,
+      "step": 13635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014778158495733806,
+      "loss": 1.0846,
+      "step": 13640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000147735797581628,
+      "loss": 1.059,
+      "step": 13645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014768999724103457,
+      "loss": 1.0423,
+      "step": 13650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014764418394799705,
+      "loss": 1.0667,
+      "step": 13655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014759835771495814,
+      "loss": 1.0948,
+      "step": 13660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014755251855436407,
+      "loss": 1.0605,
+      "step": 13665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014750666647866465,
+      "loss": 1.0129,
+      "step": 13670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014746080150031311,
+      "loss": 1.0129,
+      "step": 13675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014741492363176627,
+      "loss": 1.0827,
+      "step": 13680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014736903288548433,
+      "loss": 1.0618,
+      "step": 13685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014732312927393112,
+      "loss": 1.1006,
+      "step": 13690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014727721280957388,
+      "loss": 0.9995,
+      "step": 13695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014723128350488344,
+      "loss": 1.1303,
+      "step": 13700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014718534137233395,
+      "loss": 1.0614,
+      "step": 13705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001471393864244032,
+      "loss": 1.1748,
+      "step": 13710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014709341867357237,
+      "loss": 1.1271,
+      "step": 13715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014704743813232619,
+      "loss": 1.0585,
+      "step": 13720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014700144481315274,
+      "loss": 1.0674,
+      "step": 13725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014695543872854375,
+      "loss": 1.0323,
+      "step": 13730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014690941989099426,
+      "loss": 1.1482,
+      "step": 13735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014686338831300283,
+      "loss": 1.1189,
+      "step": 13740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014681734400707151,
+      "loss": 1.0277,
+      "step": 13745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014677128698570576,
+      "loss": 1.0566,
+      "step": 13750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014672521726141454,
+      "loss": 1.1427,
+      "step": 13755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014667913484671022,
+      "loss": 1.1434,
+      "step": 13760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014663303975410863,
+      "loss": 1.0007,
+      "step": 13765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014658693199612902,
+      "loss": 1.1359,
+      "step": 13770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001465408115852941,
+      "loss": 1.067,
+      "step": 13775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014649467853413008,
+      "loss": 1.0032,
+      "step": 13780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001464485328551665,
+      "loss": 1.0022,
+      "step": 13785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014640237456093634,
+      "loss": 1.0932,
+      "step": 13790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001463562036639761,
+      "loss": 1.0174,
+      "step": 13795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014631002017682565,
+      "loss": 1.0359,
+      "step": 13800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014626382411202818,
+      "loss": 1.0693,
+      "step": 13805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014621761548213048,
+      "loss": 1.0296,
+      "step": 13810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014617139429968262,
+      "loss": 1.0243,
+      "step": 13815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014612516057723813,
+      "loss": 1.0603,
+      "step": 13820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001460789143273539,
+      "loss": 1.0186,
+      "step": 13825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014603265556259034,
+      "loss": 1.045,
+      "step": 13830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014598638429551107,
+      "loss": 1.0088,
+      "step": 13835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014594010053868328,
+      "loss": 1.0959,
+      "step": 13840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014589380430467746,
+      "loss": 1.095,
+      "step": 13845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014584749560606755,
+      "loss": 1.0814,
+      "step": 13850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014580117445543077,
+      "loss": 0.9897,
+      "step": 13855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014575484086534786,
+      "loss": 1.0571,
+      "step": 13860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014570849484840282,
+      "loss": 1.0434,
+      "step": 13865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014566213641718307,
+      "loss": 1.0236,
+      "step": 13870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014561576558427946,
+      "loss": 1.0775,
+      "step": 13875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001455693823622861,
+      "loss": 1.0416,
+      "step": 13880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014552298676380056,
+      "loss": 1.1144,
+      "step": 13885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001454765788014237,
+      "loss": 1.0046,
+      "step": 13890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014543015848775975,
+      "loss": 1.0213,
+      "step": 13895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014538372583541636,
+      "loss": 1.0347,
+      "step": 13900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014533728085700448,
+      "loss": 1.0445,
+      "step": 13905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014529082356513835,
+      "loss": 1.0752,
+      "step": 13910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014524435397243567,
+      "loss": 1.0793,
+      "step": 13915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001451978720915174,
+      "loss": 1.0458,
+      "step": 13920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001451513779350079,
+      "loss": 1.0832,
+      "step": 13925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001451048715155348,
+      "loss": 1.1649,
+      "step": 13930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001450583528457291,
+      "loss": 1.1211,
+      "step": 13935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001450118219382251,
+      "loss": 1.0345,
+      "step": 13940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014496527880566044,
+      "loss": 1.1002,
+      "step": 13945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001449187234606761,
+      "loss": 1.0623,
+      "step": 13950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014487215591591635,
+      "loss": 1.0251,
+      "step": 13955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014482557618402875,
+      "loss": 0.9953,
+      "step": 13960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014477898427766426,
+      "loss": 1.0456,
+      "step": 13965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000144732380209477,
+      "loss": 1.1081,
+      "step": 13970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014468576399212458,
+      "loss": 1.1075,
+      "step": 13975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014463913563826772,
+      "loss": 1.0408,
+      "step": 13980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001445924951605706,
+      "loss": 1.0718,
+      "step": 13985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014454584257170055,
+      "loss": 1.0414,
+      "step": 13990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001444991778843283,
+      "loss": 1.1168,
+      "step": 13995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014445250111112782,
+      "loss": 1.0559,
+      "step": 14000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014440581226477634,
+      "loss": 1.068,
+      "step": 14005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014435911135795442,
+      "loss": 1.092,
+      "step": 14010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014431239840334586,
+      "loss": 1.015,
+      "step": 14015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001442656734136378,
+      "loss": 1.0918,
+      "step": 14020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014421893640152047,
+      "loss": 1.003,
+      "step": 14025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001441721873796876,
+      "loss": 0.9942,
+      "step": 14030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014412542636083598,
+      "loss": 1.0676,
+      "step": 14035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001440786533576658,
+      "loss": 1.0617,
+      "step": 14040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014403186838288047,
+      "loss": 1.0472,
+      "step": 14045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014398507144918656,
+      "loss": 1.0381,
+      "step": 14050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014393826256929406,
+      "loss": 0.9979,
+      "step": 14055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000143891441755916,
+      "loss": 1.034,
+      "step": 14060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014384460902176885,
+      "loss": 1.0551,
+      "step": 14065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014379776437957218,
+      "loss": 0.9728,
+      "step": 14070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014375090784204882,
+      "loss": 1.0445,
+      "step": 14075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014370403942192485,
+      "loss": 1.08,
+      "step": 14080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001436571591319296,
+      "loss": 0.9922,
+      "step": 14085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014361026698479562,
+      "loss": 1.1206,
+      "step": 14090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014356336299325864,
+      "loss": 0.9838,
+      "step": 14095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014351644717005763,
+      "loss": 1.0318,
+      "step": 14100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014346951952793472,
+      "loss": 1.1325,
+      "step": 14105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001434225800796354,
+      "loss": 1.032,
+      "step": 14110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014337562883790818,
+      "loss": 1.0701,
+      "step": 14115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014332866581550494,
+      "loss": 1.0635,
+      "step": 14120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014328169102518056,
+      "loss": 1.0074,
+      "step": 14125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014323470447969336,
+      "loss": 1.0799,
+      "step": 14130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014318770619180468,
+      "loss": 1.0537,
+      "step": 14135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014314069617427907,
+      "loss": 1.035,
+      "step": 14140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014309367443988437,
+      "loss": 0.9921,
+      "step": 14145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014304664100139145,
+      "loss": 1.0854,
+      "step": 14150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014299959587157446,
+      "loss": 1.048,
+      "step": 14155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014295253906321072,
+      "loss": 1.0854,
+      "step": 14160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014290547058908067,
+      "loss": 1.0692,
+      "step": 14165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014285839046196798,
+      "loss": 1.1194,
+      "step": 14170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014281129869465943,
+      "loss": 1.0178,
+      "step": 14175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000142764195299945,
+      "loss": 1.0124,
+      "step": 14180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014271708029061785,
+      "loss": 1.0741,
+      "step": 14185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014266995367947418,
+      "loss": 1.0356,
+      "step": 14190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014262281547931343,
+      "loss": 1.1438,
+      "step": 14195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014257566570293829,
+      "loss": 1.0794,
+      "step": 14200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014252850436315435,
+      "loss": 1.0509,
+      "step": 14205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014248133147277056,
+      "loss": 0.9975,
+      "step": 14210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014243414704459884,
+      "loss": 1.111,
+      "step": 14215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014238695109145444,
+      "loss": 1.0887,
+      "step": 14220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014233974362615548,
+      "loss": 1.0505,
+      "step": 14225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014229252466152345,
+      "loss": 1.0235,
+      "step": 14230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014224529421038283,
+      "loss": 1.1275,
+      "step": 14235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014219805228556128,
+      "loss": 1.1162,
+      "step": 14240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001421507988998895,
+      "loss": 1.0005,
+      "step": 14245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001421035340662014,
+      "loss": 0.9781,
+      "step": 14250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014205625779733393,
+      "loss": 1.0385,
+      "step": 14255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014200897010612718,
+      "loss": 1.1089,
+      "step": 14260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014196167100542428,
+      "loss": 1.0399,
+      "step": 14265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014191436050807157,
+      "loss": 1.1444,
+      "step": 14270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001418670386269184,
+      "loss": 0.9819,
+      "step": 14275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014181970537481718,
+      "loss": 1.063,
+      "step": 14280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014177236076462358,
+      "loss": 1.086,
+      "step": 14285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014172500480919616,
+      "loss": 1.0422,
+      "step": 14290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014167763752139664,
+      "loss": 1.0741,
+      "step": 14295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014163025891408983,
+      "loss": 0.9531,
+      "step": 14300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014158286900014361,
+      "loss": 1.1285,
+      "step": 14305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014153546779242893,
+      "loss": 1.0485,
+      "step": 14310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014148805530381976,
+      "loss": 1.0489,
+      "step": 14315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001414406315471932,
+      "loss": 1.038,
+      "step": 14320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001413931965354294,
+      "loss": 0.9999,
+      "step": 14325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001413457502814115,
+      "loss": 1.0843,
+      "step": 14330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014129829279802583,
+      "loss": 1.0535,
+      "step": 14335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014125082409816158,
+      "loss": 1.0475,
+      "step": 14340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014120334419471118,
+      "loss": 0.9993,
+      "step": 14345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014115585310056994,
+      "loss": 0.9795,
+      "step": 14350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014110835082863636,
+      "loss": 1.0308,
+      "step": 14355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014106083739181182,
+      "loss": 1.0444,
+      "step": 14360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001410133128030009,
+      "loss": 1.0206,
+      "step": 14365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000140965777075111,
+      "loss": 1.0684,
+      "step": 14370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014091823022105277,
+      "loss": 1.1241,
+      "step": 14375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014087067225373973,
+      "loss": 1.0873,
+      "step": 14380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001408231031860885,
+      "loss": 1.08,
+      "step": 14385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001407755230310186,
+      "loss": 1.0919,
+      "step": 14390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014072793180145275,
+      "loss": 1.0844,
+      "step": 14395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001406803295103165,
+      "loss": 1.0701,
+      "step": 14400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001406327161705385,
+      "loss": 1.0434,
+      "step": 14405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014058509179505036,
+      "loss": 1.0751,
+      "step": 14410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014053745639678667,
+      "loss": 1.1146,
+      "step": 14415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014048980998868514,
+      "loss": 1.0226,
+      "step": 14420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014044215258368628,
+      "loss": 1.096,
+      "step": 14425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014039448419473372,
+      "loss": 1.1701,
+      "step": 14430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014034680483477404,
+      "loss": 1.1423,
+      "step": 14435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014029911451675673,
+      "loss": 1.0496,
+      "step": 14440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014025141325363447,
+      "loss": 1.0815,
+      "step": 14445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001402037010583626,
+      "loss": 0.996,
+      "step": 14450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014015597794389973,
+      "loss": 1.0518,
+      "step": 14455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001401082439232072,
+      "loss": 1.067,
+      "step": 14460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014006049900924943,
+      "loss": 1.0787,
+      "step": 14465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014001274321499382,
+      "loss": 1.0599,
+      "step": 14470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013996497655341066,
+      "loss": 1.0192,
+      "step": 14475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013991719903747323,
+      "loss": 1.0792,
+      "step": 14480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001398694106801577,
+      "loss": 0.9789,
+      "step": 14485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013982161149444329,
+      "loss": 1.0406,
+      "step": 14490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000139773801493312,
+      "loss": 1.0086,
+      "step": 14495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000139725980689749,
+      "loss": 0.9748,
+      "step": 14500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013967814909674218,
+      "loss": 0.9832,
+      "step": 14505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013963030672728245,
+      "loss": 1.0357,
+      "step": 14510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013958245359436361,
+      "loss": 1.0804,
+      "step": 14515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001395345897109825,
+      "loss": 1.105,
+      "step": 14520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013948671509013871,
+      "loss": 1.0039,
+      "step": 14525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013943882974483485,
+      "loss": 1.0231,
+      "step": 14530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013939093368807643,
+      "loss": 1.0047,
+      "step": 14535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013934302693287184,
+      "loss": 1.1002,
+      "step": 14540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013929510949223245,
+      "loss": 1.1112,
+      "step": 14545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001392471813791724,
+      "loss": 1.1057,
+      "step": 14550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013919924260670882,
+      "loss": 1.1032,
+      "step": 14555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013915129318786175,
+      "loss": 1.064,
+      "step": 14560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001391033331356541,
+      "loss": 1.054,
+      "step": 14565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013905536246311162,
+      "loss": 1.0475,
+      "step": 14570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013900738118326303,
+      "loss": 1.0812,
+      "step": 14575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013895938930913988,
+      "loss": 1.0614,
+      "step": 14580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013891138685377655,
+      "loss": 1.0801,
+      "step": 14585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001388633738302104,
+      "loss": 1.0978,
+      "step": 14590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013881535025148155,
+      "loss": 0.9829,
+      "step": 14595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013876731613063315,
+      "loss": 1.0799,
+      "step": 14600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000138719271480711,
+      "loss": 1.0072,
+      "step": 14605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001386712163147639,
+      "loss": 1.0646,
+      "step": 14610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001386231506458435,
+      "loss": 1.1524,
+      "step": 14615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013857507448700423,
+      "loss": 1.151,
+      "step": 14620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013852698785130348,
+      "loss": 1.0918,
+      "step": 14625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013847889075180135,
+      "loss": 1.0708,
+      "step": 14630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013843078320156088,
+      "loss": 1.0876,
+      "step": 14635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013838266521364796,
+      "loss": 1.0218,
+      "step": 14640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001383345368011312,
+      "loss": 1.0623,
+      "step": 14645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013828639797708214,
+      "loss": 1.0666,
+      "step": 14650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013823824875457515,
+      "loss": 1.0269,
+      "step": 14655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013819008914668738,
+      "loss": 1.0994,
+      "step": 14660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013814191916649885,
+      "loss": 0.9915,
+      "step": 14665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001380937388270923,
+      "loss": 1.0923,
+      "step": 14670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001380455481415534,
+      "loss": 1.1469,
+      "step": 14675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013799734712297059,
+      "loss": 1.0616,
+      "step": 14680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013794913578443508,
+      "loss": 1.0857,
+      "step": 14685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013790091413904088,
+      "loss": 1.0141,
+      "step": 14690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013785268219988487,
+      "loss": 0.9453,
+      "step": 14695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001378044399800667,
+      "loss": 1.0303,
+      "step": 14700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013775618749268875,
+      "loss": 1.0637,
+      "step": 14705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013770792475085625,
+      "loss": 1.0883,
+      "step": 14710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001376596517676772,
+      "loss": 1.1404,
+      "step": 14715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013761136855626243,
+      "loss": 1.074,
+      "step": 14720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013756307512972538,
+      "loss": 1.081,
+      "step": 14725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001375147715011825,
+      "loss": 1.0106,
+      "step": 14730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013746645768375286,
+      "loss": 0.9788,
+      "step": 14735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013741813369055832,
+      "loss": 1.043,
+      "step": 14740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013736979953472354,
+      "loss": 1.0413,
+      "step": 14745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013732145522937582,
+      "loss": 1.067,
+      "step": 14750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013727310078764547,
+      "loss": 1.0588,
+      "step": 14755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013722473622266524,
+      "loss": 1.0372,
+      "step": 14760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001371763615475709,
+      "loss": 1.0794,
+      "step": 14765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001371279767755008,
+      "loss": 0.9561,
+      "step": 14770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013707958191959608,
+      "loss": 1.0172,
+      "step": 14775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013703117699300066,
+      "loss": 1.0306,
+      "step": 14780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013698276200886108,
+      "loss": 1.023,
+      "step": 14785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001369343369803268,
+      "loss": 1.1297,
+      "step": 14790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013688590192054978,
+      "loss": 1.0682,
+      "step": 14795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013683745684268486,
+      "loss": 1.0522,
+      "step": 14800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013678900175988963,
+      "loss": 1.1,
+      "step": 14805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013674053668532422,
+      "loss": 1.0242,
+      "step": 14810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013669206163215164,
+      "loss": 1.0332,
+      "step": 14815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013664357661353758,
+      "loss": 1.0912,
+      "step": 14820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013659508164265037,
+      "loss": 0.9765,
+      "step": 14825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013654657673266104,
+      "loss": 1.035,
+      "step": 14830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001364980618967435,
+      "loss": 1.0322,
+      "step": 14835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013644953714807403,
+      "loss": 1.0447,
+      "step": 14840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013640100249983196,
+      "loss": 1.1066,
+      "step": 14845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013635245796519903,
+      "loss": 1.0897,
+      "step": 14850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013630390355735985,
+      "loss": 1.061,
+      "step": 14855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013625533928950153,
+      "loss": 1.0751,
+      "step": 14860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001362067651748141,
+      "loss": 0.988,
+      "step": 14865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013615818122649,
+      "loss": 1.1014,
+      "step": 14870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013610958745772456,
+      "loss": 1.0257,
+      "step": 14875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013606098388171562,
+      "loss": 1.0674,
+      "step": 14880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001360123705116638,
+      "loss": 1.1612,
+      "step": 14885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013596374736077232,
+      "loss": 1.079,
+      "step": 14890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013591511444224704,
+      "loss": 1.0389,
+      "step": 14895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001358664717692965,
+      "loss": 1.0477,
+      "step": 14900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013581781935513188,
+      "loss": 1.0284,
+      "step": 14905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013576915721296709,
+      "loss": 1.0312,
+      "step": 14910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013572048535601851,
+      "loss": 1.0662,
+      "step": 14915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001356718037975053,
+      "loss": 1.1158,
+      "step": 14920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013562311255064922,
+      "loss": 1.072,
+      "step": 14925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001355744116286746,
+      "loss": 1.0199,
+      "step": 14930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001355257010448085,
+      "loss": 1.0483,
+      "step": 14935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013547698081228053,
+      "loss": 1.0035,
+      "step": 14940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013542825094432293,
+      "loss": 1.0901,
+      "step": 14945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001353795114541706,
+      "loss": 1.0666,
+      "step": 14950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013533076235506102,
+      "loss": 1.1802,
+      "step": 14955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013528200366023425,
+      "loss": 1.0289,
+      "step": 14960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000135233235382933,
+      "loss": 1.1238,
+      "step": 14965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001351844575364026,
+      "loss": 1.0667,
+      "step": 14970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013513567013389088,
+      "loss": 1.0561,
+      "step": 14975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013508687318864845,
+      "loss": 0.9961,
+      "step": 14980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013503806671392828,
+      "loss": 0.9969,
+      "step": 14985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013498925072298615,
+      "loss": 1.0206,
+      "step": 14990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013494042522908022,
+      "loss": 0.9305,
+      "step": 14995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013489159024547142,
+      "loss": 1.1151,
+      "step": 15000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001348427457854231,
+      "loss": 1.0533,
+      "step": 15005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013479389186220127,
+      "loss": 1.1152,
+      "step": 15010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013474502848907454,
+      "loss": 1.0926,
+      "step": 15015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000134696155679314,
+      "loss": 1.0811,
+      "step": 15020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013464727344619331,
+      "loss": 0.994,
+      "step": 15025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001345983818029888,
+      "loss": 1.0541,
+      "step": 15030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001345494807629792,
+      "loss": 1.008,
+      "step": 15035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001345005703394459,
+      "loss": 1.0462,
+      "step": 15040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013445165054567286,
+      "loss": 1.0486,
+      "step": 15045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013440272139494643,
+      "loss": 1.0209,
+      "step": 15050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001343537829005557,
+      "loss": 1.1139,
+      "step": 15055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013430483507579212,
+      "loss": 0.9679,
+      "step": 15060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013425587793394983,
+      "loss": 0.9779,
+      "step": 15065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013420691148832535,
+      "loss": 1.0195,
+      "step": 15070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013415793575221786,
+      "loss": 1.0203,
+      "step": 15075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013410895073892901,
+      "loss": 1.1401,
+      "step": 15080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013405995646176294,
+      "loss": 1.062,
+      "step": 15085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001340109529340263,
+      "loss": 1.0685,
+      "step": 15090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001339619401690284,
+      "loss": 1.032,
+      "step": 15095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013391291818008082,
+      "loss": 1.0704,
+      "step": 15100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013386388698049783,
+      "loss": 1.0899,
+      "step": 15105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013381484658359612,
+      "loss": 1.0541,
+      "step": 15110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013376579700269492,
+      "loss": 1.1297,
+      "step": 15115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013371673825111587,
+      "loss": 0.9925,
+      "step": 15120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013366767034218326,
+      "loss": 1.0554,
+      "step": 15125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001336185932892237,
+      "loss": 1.025,
+      "step": 15130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013356950710556638,
+      "loss": 1.0518,
+      "step": 15135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013352041180454292,
+      "loss": 1.0574,
+      "step": 15140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001334713073994875,
+      "loss": 1.0028,
+      "step": 15145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001334221939037366,
+      "loss": 1.0092,
+      "step": 15150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013337307133062942,
+      "loss": 1.0167,
+      "step": 15155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013332393969350738,
+      "loss": 1.0498,
+      "step": 15160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013327479900571452,
+      "loss": 1.0821,
+      "step": 15165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013322564928059728,
+      "loss": 1.0405,
+      "step": 15170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013317649053150456,
+      "loss": 1.0196,
+      "step": 15175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001331273227717877,
+      "loss": 1.0123,
+      "step": 15180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001330781460148005,
+      "loss": 1.0579,
+      "step": 15185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013302896027389923,
+      "loss": 1.114,
+      "step": 15190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013297976556244257,
+      "loss": 1.0077,
+      "step": 15195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013293056189379158,
+      "loss": 1.0462,
+      "step": 15200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013288134928130988,
+      "loss": 1.0836,
+      "step": 15205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013283212773836342,
+      "loss": 1.0526,
+      "step": 15210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013278289727832062,
+      "loss": 1.008,
+      "step": 15215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001327336579145523,
+      "loss": 1.0593,
+      "step": 15220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001326844096604317,
+      "loss": 1.0639,
+      "step": 15225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001326351525293345,
+      "loss": 1.0483,
+      "step": 15230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013258588653463878,
+      "loss": 1.0358,
+      "step": 15235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000132536611689725,
+      "loss": 1.137,
+      "step": 15240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013248732800797604,
+      "loss": 1.0667,
+      "step": 15245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013243803550277715,
+      "loss": 1.1036,
+      "step": 15250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013238873418751611,
+      "loss": 1.0985,
+      "step": 15255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013233942407558287,
+      "loss": 1.0511,
+      "step": 15260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013229010518037002,
+      "loss": 1.0785,
+      "step": 15265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013224077751527232,
+      "loss": 1.0117,
+      "step": 15270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000132191441093687,
+      "loss": 1.0678,
+      "step": 15275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001321420959290137,
+      "loss": 1.0776,
+      "step": 15280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001320927420346544,
+      "loss": 1.1453,
+      "step": 15285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013204337942401346,
+      "loss": 0.9941,
+      "step": 15290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013199400811049757,
+      "loss": 1.0364,
+      "step": 15295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001319446281075158,
+      "loss": 1.0369,
+      "step": 15300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013189523942847966,
+      "loss": 1.091,
+      "step": 15305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001318458420868029,
+      "loss": 1.0526,
+      "step": 15310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013179643609590168,
+      "loss": 1.042,
+      "step": 15315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013174702146919454,
+      "loss": 1.1163,
+      "step": 15320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013169759822010226,
+      "loss": 1.0297,
+      "step": 15325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013164816636204808,
+      "loss": 1.0572,
+      "step": 15330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001315987259084575,
+      "loss": 1.0496,
+      "step": 15335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001315492768727584,
+      "loss": 1.0029,
+      "step": 15340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013149981926838098,
+      "loss": 1.0399,
+      "step": 15345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013145035310875773,
+      "loss": 1.0575,
+      "step": 15350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013140087840732355,
+      "loss": 1.1036,
+      "step": 15355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013135139517751553,
+      "loss": 1.0369,
+      "step": 15360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013130190343277318,
+      "loss": 1.0183,
+      "step": 15365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013125240318653835,
+      "loss": 1.0165,
+      "step": 15370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013120289445225501,
+      "loss": 1.0417,
+      "step": 15375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013115337724336974,
+      "loss": 1.0339,
+      "step": 15380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001311038515733311,
+      "loss": 1.0261,
+      "step": 15385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013105431745559019,
+      "loss": 1.0969,
+      "step": 15390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013100477490360024,
+      "loss": 1.1461,
+      "step": 15395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001309552239308169,
+      "loss": 1.0499,
+      "step": 15400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000130905664550698,
+      "loss": 1.0664,
+      "step": 15405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013085609677670375,
+      "loss": 1.0011,
+      "step": 15410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013080652062229657,
+      "loss": 1.0376,
+      "step": 15415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013075693610094117,
+      "loss": 1.1185,
+      "step": 15420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013070734322610455,
+      "loss": 1.1076,
+      "step": 15425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013065774201125597,
+      "loss": 1.0464,
+      "step": 15430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013060813246986697,
+      "loss": 1.0318,
+      "step": 15435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001305585146154113,
+      "loss": 1.0371,
+      "step": 15440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013050888846136502,
+      "loss": 1.0216,
+      "step": 15445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013045925402120645,
+      "loss": 1.0305,
+      "step": 15450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013040961130841608,
+      "loss": 1.1001,
+      "step": 15455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013035996033647677,
+      "loss": 1.0506,
+      "step": 15460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001303103011188735,
+      "loss": 1.0245,
+      "step": 15465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013026063366909364,
+      "loss": 1.0199,
+      "step": 15470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001302109580006266,
+      "loss": 1.0855,
+      "step": 15475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001301612741269642,
+      "loss": 1.0364,
+      "step": 15480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013011158206160034,
+      "loss": 1.0515,
+      "step": 15485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001300618818180313,
+      "loss": 1.0078,
+      "step": 15490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013001217340975543,
+      "loss": 1.1065,
+      "step": 15495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012996245685027345,
+      "loss": 1.0,
+      "step": 15500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012991273215308811,
+      "loss": 1.1157,
+      "step": 15505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012986299933170453,
+      "loss": 1.0566,
+      "step": 15510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012981325839963,
+      "loss": 1.087,
+      "step": 15515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012976350937037395,
+      "loss": 1.0439,
+      "step": 15520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001297137522574481,
+      "loss": 1.0016,
+      "step": 15525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001296639870743662,
+      "loss": 1.0378,
+      "step": 15530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012961421383464445,
+      "loss": 1.1103,
+      "step": 15535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000129564432551801,
+      "loss": 1.0795,
+      "step": 15540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012951464323935634,
+      "loss": 1.0755,
+      "step": 15545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012946484591083303,
+      "loss": 1.1002,
+      "step": 15550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001294150405797559,
+      "loss": 1.1738,
+      "step": 15555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001293652272596519,
+      "loss": 1.1307,
+      "step": 15560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012931540596405013,
+      "loss": 1.0777,
+      "step": 15565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012926557670648193,
+      "loss": 1.031,
+      "step": 15570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012921573950048075,
+      "loss": 1.0327,
+      "step": 15575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012916589435958223,
+      "loss": 0.9593,
+      "step": 15580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012911604129732407,
+      "loss": 1.1054,
+      "step": 15585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012906618032724627,
+      "loss": 1.0416,
+      "step": 15590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012901631146289085,
+      "loss": 1.0927,
+      "step": 15595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012896643471780206,
+      "loss": 1.0218,
+      "step": 15600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001289165501055263,
+      "loss": 1.0354,
+      "step": 15605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012886665763961193,
+      "loss": 1.1281,
+      "step": 15610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012881675733360972,
+      "loss": 1.0319,
+      "step": 15615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012876684920107233,
+      "loss": 1.057,
+      "step": 15620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012871693325555472,
+      "loss": 1.1276,
+      "step": 15625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012866700951061382,
+      "loss": 1.1126,
+      "step": 15630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001286170779798088,
+      "loss": 1.0219,
+      "step": 15635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001285671386767009,
+      "loss": 1.0463,
+      "step": 15640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012851719161485344,
+      "loss": 1.0797,
+      "step": 15645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001284672368078319,
+      "loss": 1.0441,
+      "step": 15650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001284172742692038,
+      "loss": 1.0381,
+      "step": 15655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012836730401253886,
+      "loss": 1.1222,
+      "step": 15660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012831732605140877,
+      "loss": 0.9904,
+      "step": 15665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001282673403993874,
+      "loss": 1.0483,
+      "step": 15670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012821734707005073,
+      "loss": 1.1289,
+      "step": 15675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012816734607697672,
+      "loss": 1.1209,
+      "step": 15680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012811733743374552,
+      "loss": 1.1041,
+      "step": 15685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001280673211539393,
+      "loss": 0.9789,
+      "step": 15690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012801729725114226,
+      "loss": 1.1433,
+      "step": 15695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001279672657389408,
+      "loss": 1.0536,
+      "step": 15700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012791722663092327,
+      "loss": 1.0813,
+      "step": 15705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012786717994068013,
+      "loss": 1.041,
+      "step": 15710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012781712568180393,
+      "loss": 1.0024,
+      "step": 15715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012776706386788918,
+      "loss": 1.0545,
+      "step": 15720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012771699451253254,
+      "loss": 1.0062,
+      "step": 15725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012766691762933267,
+      "loss": 1.0329,
+      "step": 15730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001276168332318903,
+      "loss": 1.0642,
+      "step": 15735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012756674133380816,
+      "loss": 0.9884,
+      "step": 15740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012751664194869105,
+      "loss": 0.9666,
+      "step": 15745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001274665350901458,
+      "loss": 1.0517,
+      "step": 15750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012741642077178123,
+      "loss": 1.0895,
+      "step": 15755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001273662990072083,
+      "loss": 0.9901,
+      "step": 15760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012731616981003985,
+      "loss": 1.1071,
+      "step": 15765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012726603319389084,
+      "loss": 1.0259,
+      "step": 15770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012721588917237816,
+      "loss": 1.021,
+      "step": 15775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012716573775912085,
+      "loss": 1.068,
+      "step": 15780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012711557896773974,
+      "loss": 1.0371,
+      "step": 15785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001270654128118579,
+      "loss": 1.0592,
+      "step": 15790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012701523930510022,
+      "loss": 1.0905,
+      "step": 15795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001269650584610937,
+      "loss": 0.9654,
+      "step": 15800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012691487029346728,
+      "loss": 1.0484,
+      "step": 15805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012686467481585188,
+      "loss": 1.0274,
+      "step": 15810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001268144720418804,
+      "loss": 1.1199,
+      "step": 15815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012676426198518785,
+      "loss": 1.0497,
+      "step": 15820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000126714044659411,
+      "loss": 1.0591,
+      "step": 15825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012666382007818876,
+      "loss": 1.114,
+      "step": 15830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012661358825516198,
+      "loss": 0.9804,
+      "step": 15835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001265633492039734,
+      "loss": 1.0641,
+      "step": 15840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012651310293826783,
+      "loss": 1.1478,
+      "step": 15845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012646284947169195,
+      "loss": 1.1467,
+      "step": 15850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012641258881789446,
+      "loss": 1.0927,
+      "step": 15855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000126362320990526,
+      "loss": 1.098,
+      "step": 15860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012631204600323907,
+      "loss": 1.0826,
+      "step": 15865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012626176386968828,
+      "loss": 1.0308,
+      "step": 15870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012621147460353,
+      "loss": 1.0685,
+      "step": 15875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012616117821842273,
+      "loss": 1.1317,
+      "step": 15880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001261108747280267,
+      "loss": 1.0326,
+      "step": 15885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012606056414600424,
+      "loss": 1.0435,
+      "step": 15890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001260102464860195,
+      "loss": 1.0244,
+      "step": 15895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001259599217617386,
+      "loss": 1.0027,
+      "step": 15900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001259095899868296,
+      "loss": 1.0935,
+      "step": 15905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012585925117496237,
+      "loss": 1.0335,
+      "step": 15910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012580890533980883,
+      "loss": 1.0029,
+      "step": 15915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012575855249504267,
+      "loss": 1.026,
+      "step": 15920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012570819265433968,
+      "loss": 1.1028,
+      "step": 15925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012565782583137723,
+      "loss": 0.9975,
+      "step": 15930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000125607452039835,
+      "loss": 1.0539,
+      "step": 15935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012555707129339417,
+      "loss": 1.0188,
+      "step": 15940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012550668360573807,
+      "loss": 1.0348,
+      "step": 15945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001254562889905518,
+      "loss": 1.0054,
+      "step": 15950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001254058874615224,
+      "loss": 1.1159,
+      "step": 15955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012535547903233873,
+      "loss": 0.9767,
+      "step": 15960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012530506371669154,
+      "loss": 1.0589,
+      "step": 15965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012525464152827347,
+      "loss": 0.9356,
+      "step": 15970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012520421248077908,
+      "loss": 1.0232,
+      "step": 15975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012515377658790463,
+      "loss": 0.9649,
+      "step": 15980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001251033338633484,
+      "loss": 0.9755,
+      "step": 15985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012505288432081044,
+      "loss": 1.0673,
+      "step": 15990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001250024279739927,
+      "loss": 0.9671,
+      "step": 15995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012495196483659892,
+      "loss": 1.0316,
+      "step": 16000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001249014949223348,
+      "loss": 0.9738,
+      "step": 16005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012485101824490772,
+      "loss": 1.0598,
+      "step": 16010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000124800534818027,
+      "loss": 1.011,
+      "step": 16015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012475004465540375,
+      "loss": 1.0397,
+      "step": 16020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000124699547770751,
+      "loss": 1.0628,
+      "step": 16025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012464904417778345,
+      "loss": 1.0137,
+      "step": 16030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012459853389021775,
+      "loss": 0.9918,
+      "step": 16035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012454801692177232,
+      "loss": 1.0899,
+      "step": 16040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012449749328616742,
+      "loss": 0.996,
+      "step": 16045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012444696299712506,
+      "loss": 1.0728,
+      "step": 16050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012439642606836915,
+      "loss": 1.0236,
+      "step": 16055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001243458825136253,
+      "loss": 1.0046,
+      "step": 16060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000124295332346621,
+      "loss": 1.0115,
+      "step": 16065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012424477558108546,
+      "loss": 1.0178,
+      "step": 16070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012419421223074982,
+      "loss": 1.0456,
+      "step": 16075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012414364230934686,
+      "loss": 1.0005,
+      "step": 16080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012409306583061118,
+      "loss": 1.1397,
+      "step": 16085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001240424828082792,
+      "loss": 1.0506,
+      "step": 16090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012399189325608916,
+      "loss": 1.0607,
+      "step": 16095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012394129718778092,
+      "loss": 1.0455,
+      "step": 16100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012389069461709627,
+      "loss": 1.0779,
+      "step": 16105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012384008555777867,
+      "loss": 1.0092,
+      "step": 16110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012378947002357338,
+      "loss": 1.1115,
+      "step": 16115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012373884802822737,
+      "loss": 1.0782,
+      "step": 16120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001236882195854895,
+      "loss": 0.9904,
+      "step": 16125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012363758470911019,
+      "loss": 1.0706,
+      "step": 16130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012358694341284176,
+      "loss": 1.0316,
+      "step": 16135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001235362957104382,
+      "loss": 1.0686,
+      "step": 16140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012348564161565522,
+      "loss": 1.0992,
+      "step": 16145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012343498114225038,
+      "loss": 1.0069,
+      "step": 16150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012338431430398283,
+      "loss": 1.057,
+      "step": 16155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001233336411146136,
+      "loss": 1.1273,
+      "step": 16160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001232829615879052,
+      "loss": 1.0728,
+      "step": 16165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001232322757376222,
+      "loss": 1.0924,
+      "step": 16170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012318158357753058,
+      "loss": 1.0298,
+      "step": 16175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012313088512139824,
+      "loss": 1.0546,
+      "step": 16180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012308018038299464,
+      "loss": 1.0093,
+      "step": 16185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001230294693760911,
+      "loss": 0.9931,
+      "step": 16190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001229787521144605,
+      "loss": 0.9841,
+      "step": 16195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001229280286118775,
+      "loss": 1.1518,
+      "step": 16200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001228772988821184,
+      "loss": 1.0384,
+      "step": 16205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001228265629389613,
+      "loss": 0.979,
+      "step": 16210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012277582079618586,
+      "loss": 0.9979,
+      "step": 16215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001227250724675735,
+      "loss": 1.0261,
+      "step": 16220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012267431796690722,
+      "loss": 1.0265,
+      "step": 16225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001226235573079719,
+      "loss": 1.0429,
+      "step": 16230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012257279050455387,
+      "loss": 1.0686,
+      "step": 16235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012252201757044127,
+      "loss": 1.0279,
+      "step": 16240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001224712385194238,
+      "loss": 1.0125,
+      "step": 16245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000122420453365293,
+      "loss": 1.0541,
+      "step": 16250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001223696621218418,
+      "loss": 1.0824,
+      "step": 16255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012231886480286507,
+      "loss": 1.0409,
+      "step": 16260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012226806142215908,
+      "loss": 1.0078,
+      "step": 16265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012221725199352195,
+      "loss": 0.9656,
+      "step": 16270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012216643653075327,
+      "loss": 1.0255,
+      "step": 16275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012211561504765444,
+      "loss": 1.0577,
+      "step": 16280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012206478755802828,
+      "loss": 1.025,
+      "step": 16285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012201395407567948,
+      "loss": 1.0809,
+      "step": 16290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001219631146144142,
+      "loss": 0.9843,
+      "step": 16295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001219122691880403,
+      "loss": 1.0386,
+      "step": 16300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012186141781036717,
+      "loss": 0.9607,
+      "step": 16305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012181056049520591,
+      "loss": 1.0641,
+      "step": 16310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012175969725636924,
+      "loss": 1.0501,
+      "step": 16315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012170882810767132,
+      "loss": 1.0764,
+      "step": 16320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001216579530629282,
+      "loss": 1.0614,
+      "step": 16325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012160707213595722,
+      "loss": 1.0667,
+      "step": 16330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012155618534057757,
+      "loss": 1.0269,
+      "step": 16335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001215052926906099,
+      "loss": 1.0035,
+      "step": 16340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012145439419987652,
+      "loss": 1.0758,
+      "step": 16345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012140348988220123,
+      "loss": 1.1086,
+      "step": 16350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001213525797514095,
+      "loss": 1.0853,
+      "step": 16355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012130166382132837,
+      "loss": 1.0806,
+      "step": 16360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012125074210578642,
+      "loss": 1.0597,
+      "step": 16365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012119981461861383,
+      "loss": 1.048,
+      "step": 16370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012114888137364228,
+      "loss": 0.9718,
+      "step": 16375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012109794238470515,
+      "loss": 0.9895,
+      "step": 16380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012104699766563727,
+      "loss": 1.0457,
+      "step": 16385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012099604723027502,
+      "loss": 1.1585,
+      "step": 16390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012094509109245644,
+      "loss": 0.9743,
+      "step": 16395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012089412926602095,
+      "loss": 1.0225,
+      "step": 16400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012084316176480973,
+      "loss": 1.166,
+      "step": 16405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012079218860266526,
+      "loss": 1.0567,
+      "step": 16410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001207412097934318,
+      "loss": 1.0273,
+      "step": 16415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012069022535095491,
+      "loss": 1.055,
+      "step": 16420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001206392352890819,
+      "loss": 1.116,
+      "step": 16425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012058823962166139,
+      "loss": 0.991,
+      "step": 16430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012053723836254373,
+      "loss": 1.0946,
+      "step": 16435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012048623152558065,
+      "loss": 1.0626,
+      "step": 16440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012043521912462545,
+      "loss": 1.0081,
+      "step": 16445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001203842011735329,
+      "loss": 1.0255,
+      "step": 16450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012033317768615935,
+      "loss": 1.1157,
+      "step": 16455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012028214867636259,
+      "loss": 1.0739,
+      "step": 16460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012023111415800191,
+      "loss": 1.1111,
+      "step": 16465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001201800741449381,
+      "loss": 1.0721,
+      "step": 16470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012012902865103353,
+      "loss": 0.9954,
+      "step": 16475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012007797769015192,
+      "loss": 1.054,
+      "step": 16480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012002692127615856,
+      "loss": 1.0193,
+      "step": 16485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001199758594229202,
+      "loss": 1.0071,
+      "step": 16490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011992479214430508,
+      "loss": 1.0317,
+      "step": 16495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011987371945418288,
+      "loss": 1.0935,
+      "step": 16500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011982264136642483,
+      "loss": 0.9796,
+      "step": 16505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011977155789490347,
+      "loss": 1.0642,
+      "step": 16510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011972046905349299,
+      "loss": 0.9863,
+      "step": 16515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011966937485606891,
+      "loss": 1.0263,
+      "step": 16520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011961827531650824,
+      "loss": 1.1578,
+      "step": 16525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011956717044868945,
+      "loss": 1.0804,
+      "step": 16530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011951606026649245,
+      "loss": 1.0899,
+      "step": 16535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011946494478379863,
+      "loss": 1.0095,
+      "step": 16540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001194138240144907,
+      "loss": 1.1139,
+      "step": 16545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011936269797245301,
+      "loss": 1.0368,
+      "step": 16550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001193115666715711,
+      "loss": 1.0337,
+      "step": 16555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011926043012573216,
+      "loss": 1.0379,
+      "step": 16560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011920928834882464,
+      "loss": 1.023,
+      "step": 16565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001191581413547385,
+      "loss": 1.1335,
+      "step": 16570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011910698915736507,
+      "loss": 1.0723,
+      "step": 16575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011905583177059716,
+      "loss": 1.073,
+      "step": 16580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011900466920832892,
+      "loss": 1.0348,
+      "step": 16585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011895350148445594,
+      "loss": 1.0707,
+      "step": 16590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001189023286128752,
+      "loss": 1.0724,
+      "step": 16595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011885115060748507,
+      "loss": 1.0944,
+      "step": 16600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011879996748218532,
+      "loss": 1.0977,
+      "step": 16605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011874877925087715,
+      "loss": 1.068,
+      "step": 16610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011869758592746309,
+      "loss": 1.1718,
+      "step": 16615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011864638752584707,
+      "loss": 0.9944,
+      "step": 16620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011859518405993442,
+      "loss": 1.0414,
+      "step": 16625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011854397554363183,
+      "loss": 0.9956,
+      "step": 16630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011849276199084736,
+      "loss": 1.0472,
+      "step": 16635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011844154341549041,
+      "loss": 1.0578,
+      "step": 16640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001183903198314718,
+      "loss": 0.9445,
+      "step": 16645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011833909125270372,
+      "loss": 1.049,
+      "step": 16650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001182878576930996,
+      "loss": 1.0062,
+      "step": 16655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001182366191665744,
+      "loss": 1.035,
+      "step": 16660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011818537568704424,
+      "loss": 1.0667,
+      "step": 16665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011813412726842676,
+      "loss": 1.0438,
+      "step": 16670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001180828739246408,
+      "loss": 1.0216,
+      "step": 16675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011803161566960661,
+      "loss": 0.9874,
+      "step": 16680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011798035251724577,
+      "loss": 0.9954,
+      "step": 16685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011792908448148119,
+      "loss": 1.0287,
+      "step": 16690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011787781157623709,
+      "loss": 1.047,
+      "step": 16695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000117826533815439,
+      "loss": 1.0276,
+      "step": 16700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011777525121301383,
+      "loss": 1.0766,
+      "step": 16705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011772396378288973,
+      "loss": 1.041,
+      "step": 16710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001176726715389962,
+      "loss": 1.0318,
+      "step": 16715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011762137449526404,
+      "loss": 1.1067,
+      "step": 16720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011757007266562536,
+      "loss": 1.0843,
+      "step": 16725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001175187660640136,
+      "loss": 1.127,
+      "step": 16730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001174674547043634,
+      "loss": 1.0288,
+      "step": 16735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011741613860061078,
+      "loss": 1.0833,
+      "step": 16740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011736481776669306,
+      "loss": 1.0307,
+      "step": 16745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011731349221654875,
+      "loss": 1.0487,
+      "step": 16750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011726216196411773,
+      "loss": 1.0504,
+      "step": 16755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001172108270233411,
+      "loss": 1.0458,
+      "step": 16760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001171594874081613,
+      "loss": 1.1016,
+      "step": 16765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011710814313252197,
+      "loss": 1.1063,
+      "step": 16770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011705679421036804,
+      "loss": 1.0885,
+      "step": 16775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011700544065564569,
+      "loss": 1.0588,
+      "step": 16780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011695408248230245,
+      "loss": 0.9937,
+      "step": 16785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011690271970428692,
+      "loss": 1.0229,
+      "step": 16790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011685135233554917,
+      "loss": 1.0234,
+      "step": 16795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001167999803900403,
+      "loss": 0.9762,
+      "step": 16800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011674860388171287,
+      "loss": 1.018,
+      "step": 16805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011669722282452044,
+      "loss": 1.0489,
+      "step": 16810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011664583723241808,
+      "loss": 0.9754,
+      "step": 16815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001165944471193618,
+      "loss": 1.0288,
+      "step": 16820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001165430524993091,
+      "loss": 0.9729,
+      "step": 16825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011649165338621851,
+      "loss": 1.0008,
+      "step": 16830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011644024979404991,
+      "loss": 1.059,
+      "step": 16835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001163888417367643,
+      "loss": 1.03,
+      "step": 16840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011633742922832395,
+      "loss": 1.0356,
+      "step": 16845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011628601228269236,
+      "loss": 1.1086,
+      "step": 16850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011623459091383418,
+      "loss": 1.0117,
+      "step": 16855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011618316513571525,
+      "loss": 0.9786,
+      "step": 16860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011613173496230268,
+      "loss": 1.0918,
+      "step": 16865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001160803004075647,
+      "loss": 1.0918,
+      "step": 16870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001160288614854708,
+      "loss": 1.0374,
+      "step": 16875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001159774182099916,
+      "loss": 1.0492,
+      "step": 16880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011592597059509893,
+      "loss": 1.031,
+      "step": 16885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011587451865476575,
+      "loss": 1.0997,
+      "step": 16890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011582306240296632,
+      "loss": 1.0536,
+      "step": 16895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001157716018536759,
+      "loss": 0.9987,
+      "step": 16900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011572013702087107,
+      "loss": 1.0397,
+      "step": 16905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011566866791852944,
+      "loss": 1.1162,
+      "step": 16910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011561719456062994,
+      "loss": 1.1106,
+      "step": 16915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011556571696115244,
+      "loss": 1.1285,
+      "step": 16920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001155142351340782,
+      "loss": 1.0485,
+      "step": 16925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011546274909338945,
+      "loss": 1.0473,
+      "step": 16930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001154112588530696,
+      "loss": 1.0767,
+      "step": 16935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011535976442710334,
+      "loss": 1.016,
+      "step": 16940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011530826582947624,
+      "loss": 1.0699,
+      "step": 16945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011525676307417528,
+      "loss": 1.0717,
+      "step": 16950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011520525617518831,
+      "loss": 1.0533,
+      "step": 16955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011515374514650457,
+      "loss": 1.0271,
+      "step": 16960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011510223000211415,
+      "loss": 0.954,
+      "step": 16965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011505071075600846,
+      "loss": 1.0581,
+      "step": 16970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011499918742217996,
+      "loss": 0.9361,
+      "step": 16975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001149476600146222,
+      "loss": 1.0547,
+      "step": 16980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011489612854732988,
+      "loss": 1.0908,
+      "step": 16985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011484459303429871,
+      "loss": 1.0489,
+      "step": 16990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011479305348952563,
+      "loss": 1.0477,
+      "step": 16995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011474150992700856,
+      "loss": 1.1402,
+      "step": 17000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011468996236074658,
+      "loss": 1.0611,
+      "step": 17005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011463841080473984,
+      "loss": 1.0811,
+      "step": 17010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001145868552729896,
+      "loss": 1.0542,
+      "step": 17015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001145352957794981,
+      "loss": 1.0246,
+      "step": 17020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011448373233826878,
+      "loss": 1.0224,
+      "step": 17025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011443216496330608,
+      "loss": 1.0713,
+      "step": 17030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011438059366861551,
+      "loss": 1.0387,
+      "step": 17035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011432901846820374,
+      "loss": 0.9866,
+      "step": 17040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011427743937607829,
+      "loss": 1.0394,
+      "step": 17045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011422585640624801,
+      "loss": 1.0594,
+      "step": 17050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011417426957272255,
+      "loss": 1.0903,
+      "step": 17055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001141226788895128,
+      "loss": 1.1227,
+      "step": 17060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011407108437063057,
+      "loss": 1.0171,
+      "step": 17065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011401948603008877,
+      "loss": 1.0606,
+      "step": 17070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011396788388190133,
+      "loss": 0.9801,
+      "step": 17075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011391627794008326,
+      "loss": 1.0562,
+      "step": 17080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011386466821865052,
+      "loss": 1.0734,
+      "step": 17085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011381305473162016,
+      "loss": 1.2374,
+      "step": 17090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011376143749301023,
+      "loss": 1.1451,
+      "step": 17095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011370981651683978,
+      "loss": 1.0992,
+      "step": 17100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011365819181712891,
+      "loss": 0.998,
+      "step": 17105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011360656340789873,
+      "loss": 1.0326,
+      "step": 17110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011355493130317129,
+      "loss": 1.089,
+      "step": 17115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011350329551696978,
+      "loss": 1.0561,
+      "step": 17120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011345165606331825,
+      "loss": 1.0282,
+      "step": 17125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011340001295624181,
+      "loss": 1.0662,
+      "step": 17130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011334836620976654,
+      "loss": 1.1069,
+      "step": 17135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011329671583791959,
+      "loss": 1.0205,
+      "step": 17140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011324506185472894,
+      "loss": 1.0117,
+      "step": 17145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011319340427422372,
+      "loss": 1.1055,
+      "step": 17150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011314174311043391,
+      "loss": 1.0506,
+      "step": 17155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011309007837739049,
+      "loss": 1.0017,
+      "step": 17160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001130384100891255,
+      "loss": 1.0514,
+      "step": 17165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011298673825967183,
+      "loss": 1.0381,
+      "step": 17170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011293506290306335,
+      "loss": 1.0109,
+      "step": 17175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011288338403333494,
+      "loss": 1.0245,
+      "step": 17180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011283170166452246,
+      "loss": 1.0513,
+      "step": 17185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011278001581066258,
+      "loss": 1.1285,
+      "step": 17190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001127283264857931,
+      "loss": 1.1582,
+      "step": 17195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011267663370395256,
+      "loss": 1.1311,
+      "step": 17200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011262493747918064,
+      "loss": 1.0675,
+      "step": 17205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011257323782551779,
+      "loss": 1.0747,
+      "step": 17210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011252153475700551,
+      "loss": 1.0397,
+      "step": 17215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001124698282876862,
+      "loss": 1.0503,
+      "step": 17220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001124181184316031,
+      "loss": 1.0436,
+      "step": 17225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011236640520280049,
+      "loss": 1.0455,
+      "step": 17230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011231468861532346,
+      "loss": 1.0997,
+      "step": 17235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011226296868321813,
+      "loss": 1.0674,
+      "step": 17240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011221124542053139,
+      "loss": 1.1412,
+      "step": 17245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011215951884131115,
+      "loss": 1.0457,
+      "step": 17250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011210778895960616,
+      "loss": 1.0397,
+      "step": 17255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011205605578946609,
+      "loss": 1.103,
+      "step": 17260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011200431934494152,
+      "loss": 1.0319,
+      "step": 17265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011195257964008383,
+      "loss": 0.9845,
+      "step": 17270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001119008366889454,
+      "loss": 0.98,
+      "step": 17275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011184909050557944,
+      "loss": 0.9878,
+      "step": 17280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011179734110404003,
+      "loss": 1.1495,
+      "step": 17285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011174558849838213,
+      "loss": 1.0501,
+      "step": 17290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011169383270266163,
+      "loss": 1.0553,
+      "step": 17295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011164207373093512,
+      "loss": 0.9852,
+      "step": 17300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001115903115972603,
+      "loss": 1.0407,
+      "step": 17305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011153854631569544,
+      "loss": 1.1059,
+      "step": 17310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011148677790029997,
+      "loss": 1.0298,
+      "step": 17315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001114350063651339,
+      "loss": 0.9959,
+      "step": 17320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011138323172425829,
+      "loss": 0.9879,
+      "step": 17325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011133145399173492,
+      "loss": 1.061,
+      "step": 17330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011127967318162644,
+      "loss": 1.0454,
+      "step": 17335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001112278893079964,
+      "loss": 1.0903,
+      "step": 17340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011117610238490903,
+      "loss": 1.0882,
+      "step": 17345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001111243124264296,
+      "loss": 1.0967,
+      "step": 17350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011107251944662398,
+      "loss": 1.072,
+      "step": 17355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011102072345955903,
+      "loss": 1.0874,
+      "step": 17360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011096892447930238,
+      "loss": 1.0951,
+      "step": 17365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011091712251992244,
+      "loss": 1.0882,
+      "step": 17370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011086531759548843,
+      "loss": 1.0147,
+      "step": 17375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011081350972007043,
+      "loss": 1.0421,
+      "step": 17380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011076169890773929,
+      "loss": 1.0815,
+      "step": 17385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011070988517256659,
+      "loss": 1.0586,
+      "step": 17390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011065806852862484,
+      "loss": 1.0462,
+      "step": 17395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011060624898998724,
+      "loss": 0.9747,
+      "step": 17400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011055442657072781,
+      "loss": 1.1752,
+      "step": 17405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011050260128492131,
+      "loss": 1.0365,
+      "step": 17410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011045077314664336,
+      "loss": 0.9845,
+      "step": 17415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001103989421699703,
+      "loss": 1.0461,
+      "step": 17420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011034710836897921,
+      "loss": 1.1143,
+      "step": 17425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011029527175774807,
+      "loss": 1.0354,
+      "step": 17430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011024343235035541,
+      "loss": 1.0174,
+      "step": 17435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011019159016088077,
+      "loss": 1.03,
+      "step": 17440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011013974520340418,
+      "loss": 1.0119,
+      "step": 17445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011008789749200668,
+      "loss": 1.1158,
+      "step": 17450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011003604704076984,
+      "loss": 1.0306,
+      "step": 17455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010998419386377612,
+      "loss": 1.0364,
+      "step": 17460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010993233797510866,
+      "loss": 0.9333,
+      "step": 17465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010988047938885135,
+      "loss": 1.0868,
+      "step": 17470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010982861811908882,
+      "loss": 0.9888,
+      "step": 17475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010977675417990638,
+      "loss": 1.0597,
+      "step": 17480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010972488758539015,
+      "loss": 1.0489,
+      "step": 17485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010967301834962689,
+      "loss": 0.9978,
+      "step": 17490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001096211464867041,
+      "loss": 0.9844,
+      "step": 17495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010956927201071005,
+      "loss": 0.9725,
+      "step": 17500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010951739493573365,
+      "loss": 1.0761,
+      "step": 17505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010946551527586452,
+      "loss": 1.0293,
+      "step": 17510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010941363304519305,
+      "loss": 1.0689,
+      "step": 17515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010936174825781024,
+      "loss": 0.9875,
+      "step": 17520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010930986092780781,
+      "loss": 1.0431,
+      "step": 17525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010925797106927828,
+      "loss": 1.0609,
+      "step": 17530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010920607869631465,
+      "loss": 1.0282,
+      "step": 17535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010915418382301079,
+      "loss": 1.0159,
+      "step": 17540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010910228646346115,
+      "loss": 1.0636,
+      "step": 17545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010905038663176086,
+      "loss": 1.1105,
+      "step": 17550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010899848434200577,
+      "loss": 1.0585,
+      "step": 17555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010894657960829235,
+      "loss": 1.0458,
+      "step": 17560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010889467244471777,
+      "loss": 1.0937,
+      "step": 17565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010884276286537978,
+      "loss": 0.9878,
+      "step": 17570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010879085088437698,
+      "loss": 1.0137,
+      "step": 17575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010873893651580833,
+      "loss": 1.0564,
+      "step": 17580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010868701977377376,
+      "loss": 1.092,
+      "step": 17585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010863510067237355,
+      "loss": 1.077,
+      "step": 17590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010858317922570885,
+      "loss": 1.0662,
+      "step": 17595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010853125544788125,
+      "loss": 1.0016,
+      "step": 17600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010847932935299319,
+      "loss": 1.0857,
+      "step": 17605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010842740095514754,
+      "loss": 1.084,
+      "step": 17610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010837547026844794,
+      "loss": 1.0184,
+      "step": 17615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010832353730699854,
+      "loss": 1.0341,
+      "step": 17620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010827160208490418,
+      "loss": 1.03,
+      "step": 17625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001082196646162703,
+      "loss": 1.0349,
+      "step": 17630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010816772491520293,
+      "loss": 1.0325,
+      "step": 17635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010811578299580874,
+      "loss": 1.0324,
+      "step": 17640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010806383887219496,
+      "loss": 1.0353,
+      "step": 17645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010801189255846944,
+      "loss": 1.0845,
+      "step": 17650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010795994406874064,
+      "loss": 0.9926,
+      "step": 17655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010790799341711756,
+      "loss": 1.0083,
+      "step": 17660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010785604061770991,
+      "loss": 0.9906,
+      "step": 17665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010780408568462778,
+      "loss": 1.0359,
+      "step": 17670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010775212863198208,
+      "loss": 1.1053,
+      "step": 17675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010770016947388407,
+      "loss": 1.0255,
+      "step": 17680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010764820822444574,
+      "loss": 1.0949,
+      "step": 17685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010759624489777955,
+      "loss": 1.0844,
+      "step": 17690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010754427950799861,
+      "loss": 1.0976,
+      "step": 17695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010749231206921653,
+      "loss": 1.1104,
+      "step": 17700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001074403425955475,
+      "loss": 0.9957,
+      "step": 17705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010738837110110622,
+      "loss": 1.062,
+      "step": 17710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000107336397600008,
+      "loss": 0.9972,
+      "step": 17715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010728442210636868,
+      "loss": 1.0418,
+      "step": 17720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010723244463430464,
+      "loss": 0.9245,
+      "step": 17725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010718046519793276,
+      "loss": 0.9977,
+      "step": 17730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010712848381137048,
+      "loss": 1.0144,
+      "step": 17735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010707650048873579,
+      "loss": 1.0433,
+      "step": 17740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010702451524414718,
+      "loss": 1.0439,
+      "step": 17745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010697252809172366,
+      "loss": 1.1379,
+      "step": 17750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010692053904558479,
+      "loss": 1.0929,
+      "step": 17755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010686854811985058,
+      "loss": 1.0676,
+      "step": 17760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010681655532864165,
+      "loss": 1.0808,
+      "step": 17765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010676456068607901,
+      "loss": 0.9935,
+      "step": 17770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010671256420628428,
+      "loss": 1.0908,
+      "step": 17775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001066605659033795,
+      "loss": 1.0484,
+      "step": 17780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010660856579148724,
+      "loss": 1.063,
+      "step": 17785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010655656388473056,
+      "loss": 1.0618,
+      "step": 17790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010650456019723301,
+      "loss": 1.0161,
+      "step": 17795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010645255474311862,
+      "loss": 1.0504,
+      "step": 17800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010640054753651187,
+      "loss": 1.06,
+      "step": 17805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010634853859153779,
+      "loss": 1.0437,
+      "step": 17810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010629652792232178,
+      "loss": 1.0186,
+      "step": 17815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010624451554298984,
+      "loss": 0.9854,
+      "step": 17820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010619250146766826,
+      "loss": 1.0962,
+      "step": 17825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010614048571048401,
+      "loss": 1.0874,
+      "step": 17830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010608846828556427,
+      "loss": 1.0166,
+      "step": 17835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010603644920703691,
+      "loss": 1.0522,
+      "step": 17840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010598442848903,
+      "loss": 1.0537,
+      "step": 17845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010593240614567237,
+      "loss": 1.0014,
+      "step": 17850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010588038219109298,
+      "loss": 1.0654,
+      "step": 17855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010582835663942141,
+      "loss": 1.0212,
+      "step": 17860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010577632950478764,
+      "loss": 1.0352,
+      "step": 17865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010572430080132205,
+      "loss": 1.0201,
+      "step": 17870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010567227054315545,
+      "loss": 1.0028,
+      "step": 17875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010562023874441911,
+      "loss": 1.0203,
+      "step": 17880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010556820541924466,
+      "loss": 0.9729,
+      "step": 17885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010551617058176422,
+      "loss": 1.0253,
+      "step": 17890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010546413424611023,
+      "loss": 1.0138,
+      "step": 17895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010541209642641563,
+      "loss": 1.0885,
+      "step": 17900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001053600571368137,
+      "loss": 0.9752,
+      "step": 17905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010530801639143812,
+      "loss": 1.0413,
+      "step": 17910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010525597420442299,
+      "loss": 1.137,
+      "step": 17915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010520393058990285,
+      "loss": 1.0034,
+      "step": 17920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010515188556201247,
+      "loss": 1.0115,
+      "step": 17925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010509983913488722,
+      "loss": 1.037,
+      "step": 17930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001050477913226626,
+      "loss": 1.0491,
+      "step": 17935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010499574213947479,
+      "loss": 1.0316,
+      "step": 17940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010494369159946001,
+      "loss": 1.0817,
+      "step": 17945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010489163971675514,
+      "loss": 1.0928,
+      "step": 17950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010483958650549721,
+      "loss": 0.9814,
+      "step": 17955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010478753197982372,
+      "loss": 1.0371,
+      "step": 17960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010473547615387256,
+      "loss": 1.0381,
+      "step": 17965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010468341904178182,
+      "loss": 1.1164,
+      "step": 17970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010463136065769016,
+      "loss": 1.0019,
+      "step": 17975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010457930101573636,
+      "loss": 1.0667,
+      "step": 17980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010452724013005973,
+      "loss": 1.0587,
+      "step": 17985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010447517801479971,
+      "loss": 1.1089,
+      "step": 17990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010442311468409631,
+      "loss": 0.984,
+      "step": 17995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010437105015208972,
+      "loss": 1.0928,
+      "step": 18000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001043189844329205,
+      "loss": 1.0368,
+      "step": 18005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010426691754072952,
+      "loss": 1.074,
+      "step": 18010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010421484948965797,
+      "loss": 1.0114,
+      "step": 18015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010416278029384737,
+      "loss": 1.0457,
+      "step": 18020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001041107099674395,
+      "loss": 1.0856,
+      "step": 18025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010405863852457655,
+      "loss": 1.0463,
+      "step": 18030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001040065659794009,
+      "loss": 1.0739,
+      "step": 18035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010395449234605532,
+      "loss": 1.0107,
+      "step": 18040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001039024176386828,
+      "loss": 1.0824,
+      "step": 18045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010385034187142667,
+      "loss": 1.0984,
+      "step": 18050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010379826505843054,
+      "loss": 0.9847,
+      "step": 18055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010374618721383827,
+      "loss": 1.0309,
+      "step": 18060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010369410835179408,
+      "loss": 0.9811,
+      "step": 18065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010364202848644235,
+      "loss": 1.0825,
+      "step": 18070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010358994763192788,
+      "loss": 1.1538,
+      "step": 18075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010353786580239559,
+      "loss": 1.052,
+      "step": 18080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010348578301199077,
+      "loss": 1.0487,
+      "step": 18085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010343369927485888,
+      "loss": 1.0251,
+      "step": 18090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010338161460514577,
+      "loss": 1.0128,
+      "step": 18095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010332952901699738,
+      "loss": 1.0485,
+      "step": 18100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010327744252456003,
+      "loss": 1.0709,
+      "step": 18105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010322535514198021,
+      "loss": 1.0185,
+      "step": 18110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010317326688340472,
+      "loss": 1.0852,
+      "step": 18115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010312117776298054,
+      "loss": 1.0435,
+      "step": 18120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010306908779485487,
+      "loss": 1.0555,
+      "step": 18125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001030169969931752,
+      "loss": 1.0726,
+      "step": 18130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010296490537208923,
+      "loss": 1.0965,
+      "step": 18135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010291281294574481,
+      "loss": 0.975,
+      "step": 18140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010286071972829016,
+      "loss": 1.0603,
+      "step": 18145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010280862573387355,
+      "loss": 1.0459,
+      "step": 18150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010275653097664357,
+      "loss": 1.0516,
+      "step": 18155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010270443547074899,
+      "loss": 1.0926,
+      "step": 18160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010265233923033874,
+      "loss": 1.0855,
+      "step": 18165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010260024226956202,
+      "loss": 1.0325,
+      "step": 18170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010254814460256817,
+      "loss": 1.041,
+      "step": 18175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010249604624350672,
+      "loss": 1.0165,
+      "step": 18180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010244394720652748,
+      "loss": 0.9926,
+      "step": 18185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001023918475057803,
+      "loss": 1.1287,
+      "step": 18190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010233974715541535,
+      "loss": 1.0507,
+      "step": 18195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010228764616958287,
+      "loss": 1.0089,
+      "step": 18200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001022355445624333,
+      "loss": 1.066,
+      "step": 18205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010218344234811732,
+      "loss": 0.9528,
+      "step": 18210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010213133954078564,
+      "loss": 1.079,
+      "step": 18215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001020792361545893,
+      "loss": 1.0801,
+      "step": 18220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001020271322036793,
+      "loss": 0.9493,
+      "step": 18225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010197502770220701,
+      "loss": 1.0335,
+      "step": 18230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010192292266432377,
+      "loss": 1.0173,
+      "step": 18235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010187081710418116,
+      "loss": 1.0136,
+      "step": 18240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010181871103593084,
+      "loss": 0.9877,
+      "step": 18245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010176660447372468,
+      "loss": 1.0047,
+      "step": 18250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010171449743171466,
+      "loss": 1.0244,
+      "step": 18255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010166238992405283,
+      "loss": 1.0421,
+      "step": 18260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010161028196489147,
+      "loss": 1.01,
+      "step": 18265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010155817356838288,
+      "loss": 1.0295,
+      "step": 18270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010150606474867956,
+      "loss": 1.1261,
+      "step": 18275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010145395551993408,
+      "loss": 1.0279,
+      "step": 18280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010140184589629913,
+      "loss": 0.9927,
+      "step": 18285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010134973589192751,
+      "loss": 1.0194,
+      "step": 18290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010129762552097213,
+      "loss": 1.0692,
+      "step": 18295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010124551479758597,
+      "loss": 1.0795,
+      "step": 18300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010119340373592214,
+      "loss": 1.0817,
+      "step": 18305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001011412923501339,
+      "loss": 1.043,
+      "step": 18310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001010891806543744,
+      "loss": 1.093,
+      "step": 18315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010103706866279712,
+      "loss": 1.0912,
+      "step": 18320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010098495638955538,
+      "loss": 1.0872,
+      "step": 18325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010093284384880283,
+      "loss": 1.1468,
+      "step": 18330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010088073105469297,
+      "loss": 0.9912,
+      "step": 18335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010082861802137955,
+      "loss": 1.0617,
+      "step": 18340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010077650476301619,
+      "loss": 1.0886,
+      "step": 18345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010072439129375672,
+      "loss": 1.0772,
+      "step": 18350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010067227762775501,
+      "loss": 1.0387,
+      "step": 18355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010062016377916495,
+      "loss": 0.9528,
+      "step": 18360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001005680497621405,
+      "loss": 1.0717,
+      "step": 18365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010051593559083557,
+      "loss": 1.1101,
+      "step": 18370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001004638212794043,
+      "loss": 1.0997,
+      "step": 18375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010041170684200067,
+      "loss": 1.039,
+      "step": 18380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010035959229277889,
+      "loss": 1.1067,
+      "step": 18385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010030747764589301,
+      "loss": 1.0728,
+      "step": 18390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010025536291549724,
+      "loss": 1.0266,
+      "step": 18395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010020324811574575,
+      "loss": 1.1265,
+      "step": 18400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010015113326079275,
+      "loss": 1.065,
+      "step": 18405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010009901836479245,
+      "loss": 1.0613,
+      "step": 18410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010004690344189912,
+      "loss": 1.0724,
+      "step": 18415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.999478850626694e-05,
+      "loss": 1.0653,
+      "step": 18420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.99426735720502e-05,
+      "loss": 1.0401,
+      "step": 18425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.989055865340313e-05,
+      "loss": 1.0151,
+      "step": 18430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.983844376447994e-05,
+      "loss": 1.0327,
+      "step": 18435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.978632891943493e-05,
+      "loss": 1.0252,
+      "step": 18440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.973421413242225e-05,
+      "loss": 1.1135,
+      "step": 18445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.968209941759612e-05,
+      "loss": 1.079,
+      "step": 18450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.962998478911071e-05,
+      "loss": 1.0002,
+      "step": 18455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.95778702611202e-05,
+      "loss": 1.0053,
+      "step": 18460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.952575584777877e-05,
+      "loss": 1.0481,
+      "step": 18465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.947364156324044e-05,
+      "loss": 1.0076,
+      "step": 18470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.942152742165928e-05,
+      "loss": 1.0447,
+      "step": 18475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.936941343718937e-05,
+      "loss": 1.044,
+      "step": 18480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.931729962398468e-05,
+      "loss": 1.0084,
+      "step": 18485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.926518599619911e-05,
+      "loss": 0.9854,
+      "step": 18490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.921307256798655e-05,
+      "loss": 1.0553,
+      "step": 18495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.916095935350088e-05,
+      "loss": 1.0122,
+      "step": 18500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.910884636689585e-05,
+      "loss": 1.0318,
+      "step": 18505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.90567336223252e-05,
+      "loss": 1.0592,
+      "step": 18510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.900462113394246e-05,
+      "loss": 1.0216,
+      "step": 18515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.89525089159014e-05,
+      "loss": 0.9709,
+      "step": 18520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890039698235539e-05,
+      "loss": 1.0353,
+      "step": 18525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.884828534745787e-05,
+      "loss": 1.0568,
+      "step": 18530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.87961740253622e-05,
+      "loss": 0.986,
+      "step": 18535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.87440630302217e-05,
+      "loss": 1.0718,
+      "step": 18540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.869195237618946e-05,
+      "loss": 0.9821,
+      "step": 18545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.863984207741859e-05,
+      "loss": 1.0511,
+      "step": 18550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.858773214806206e-05,
+      "loss": 1.0493,
+      "step": 18555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.853562260227279e-05,
+      "loss": 1.0913,
+      "step": 18560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.848351345420354e-05,
+      "loss": 1.1277,
+      "step": 18565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.843140471800696e-05,
+      "loss": 1.0604,
+      "step": 18570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.837929640783563e-05,
+      "loss": 1.0629,
+      "step": 18575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.832718853784196e-05,
+      "loss": 1.0326,
+      "step": 18580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.827508112217835e-05,
+      "loss": 1.0234,
+      "step": 18585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.822297417499693e-05,
+      "loss": 1.0812,
+      "step": 18590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.81708677104498e-05,
+      "loss": 1.0066,
+      "step": 18595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.811876174268888e-05,
+      "loss": 0.9738,
+      "step": 18600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.806665628586599e-05,
+      "loss": 0.9336,
+      "step": 18605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.801455135413284e-05,
+      "loss": 1.0476,
+      "step": 18610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.79624469616409e-05,
+      "loss": 0.9799,
+      "step": 18615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.791034312254153e-05,
+      "loss": 1.1205,
+      "step": 18620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.785823985098603e-05,
+      "loss": 1.0532,
+      "step": 18625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.780613716112545e-05,
+      "loss": 1.1407,
+      "step": 18630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.77540350671107e-05,
+      "loss": 1.0658,
+      "step": 18635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.770193358309248e-05,
+      "loss": 1.1047,
+      "step": 18640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.764983272322147e-05,
+      "loss": 1.0169,
+      "step": 18645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.759773250164805e-05,
+      "loss": 1.1041,
+      "step": 18650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.754563293252249e-05,
+      "loss": 1.1431,
+      "step": 18655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.749353402999476e-05,
+      "loss": 1.0489,
+      "step": 18660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.744143580821489e-05,
+      "loss": 1.0622,
+      "step": 18665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.738933828133251e-05,
+      "loss": 1.0615,
+      "step": 18670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.733724146349714e-05,
+      "loss": 1.0008,
+      "step": 18675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.728514536885808e-05,
+      "loss": 1.0968,
+      "step": 18680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.723305001156452e-05,
+      "loss": 0.9672,
+      "step": 18685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.718095540576531e-05,
+      "loss": 1.0255,
+      "step": 18690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.712886156560923e-05,
+      "loss": 1.0624,
+      "step": 18695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.707676850524473e-05,
+      "loss": 0.975,
+      "step": 18700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.702467623882017e-05,
+      "loss": 1.0515,
+      "step": 18705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.697258478048366e-05,
+      "loss": 1.016,
+      "step": 18710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.692049414438299e-05,
+      "loss": 1.0567,
+      "step": 18715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.68684043446658e-05,
+      "loss": 1.0485,
+      "step": 18720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.681631539547958e-05,
+      "loss": 1.1189,
+      "step": 18725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.676422731097149e-05,
+      "loss": 1.0126,
+      "step": 18730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.671214010528845e-05,
+      "loss": 1.1028,
+      "step": 18735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.666005379257715e-05,
+      "loss": 1.0622,
+      "step": 18740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.660796838698413e-05,
+      "loss": 1.1035,
+      "step": 18745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.655588390265558e-05,
+      "loss": 1.0244,
+      "step": 18750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.650380035373749e-05,
+      "loss": 1.0085,
+      "step": 18755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.645171775437548e-05,
+      "loss": 1.0342,
+      "step": 18760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.639963611871517e-05,
+      "loss": 1.0005,
+      "step": 18765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.634755546090164e-05,
+      "loss": 1.0919,
+      "step": 18770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.62954757950799e-05,
+      "loss": 1.0313,
+      "step": 18775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.624339713539454e-05,
+      "loss": 0.9648,
+      "step": 18780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.619131949598997e-05,
+      "loss": 1.0076,
+      "step": 18785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.613924289101033e-05,
+      "loss": 1.0311,
+      "step": 18790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.608716733459945e-05,
+      "loss": 1.0399,
+      "step": 18795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.603509284090089e-05,
+      "loss": 1.0373,
+      "step": 18800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.598301942405781e-05,
+      "loss": 1.0153,
+      "step": 18805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.59309470982133e-05,
+      "loss": 0.9998,
+      "step": 18810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.587887587750998e-05,
+      "loss": 1.0996,
+      "step": 18815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.582680577609022e-05,
+      "loss": 1.0624,
+      "step": 18820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.577473680809607e-05,
+      "loss": 1.0477,
+      "step": 18825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.572266898766934e-05,
+      "loss": 1.0601,
+      "step": 18830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.567060232895143e-05,
+      "loss": 1.062,
+      "step": 18835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.561853684608348e-05,
+      "loss": 1.0563,
+      "step": 18840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.556647255320627e-05,
+      "loss": 1.0382,
+      "step": 18845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.551440946446034e-05,
+      "loss": 1.0021,
+      "step": 18850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.546234759398586e-05,
+      "loss": 1.0535,
+      "step": 18855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.541028695592263e-05,
+      "loss": 1.0387,
+      "step": 18860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.535822756441011e-05,
+      "loss": 1.1016,
+      "step": 18865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.53061694335875e-05,
+      "loss": 1.0378,
+      "step": 18870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.525411257759367e-05,
+      "loss": 1.0364,
+      "step": 18875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.520205701056698e-05,
+      "loss": 1.05,
+      "step": 18880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.515000274664558e-05,
+      "loss": 1.1044,
+      "step": 18885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.509794979996727e-05,
+      "loss": 0.9341,
+      "step": 18890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.504589818466946e-05,
+      "loss": 1.0315,
+      "step": 18895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.499384791488918e-05,
+      "loss": 1.0559,
+      "step": 18900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.494179900476303e-05,
+      "loss": 1.0413,
+      "step": 18905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.488975146842746e-05,
+      "loss": 1.0278,
+      "step": 18910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.483770532001833e-05,
+      "loss": 1.062,
+      "step": 18915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.478566057367124e-05,
+      "loss": 1.0192,
+      "step": 18920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.473361724352126e-05,
+      "loss": 1.0762,
+      "step": 18925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.468157534370333e-05,
+      "loss": 1.0652,
+      "step": 18930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.462953488835177e-05,
+      "loss": 1.0606,
+      "step": 18935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.457749589160061e-05,
+      "loss": 1.0173,
+      "step": 18940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.452545836758347e-05,
+      "loss": 0.9727,
+      "step": 18945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.447342233043357e-05,
+      "loss": 1.0488,
+      "step": 18950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.442138779428376e-05,
+      "loss": 1.1852,
+      "step": 18955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.436935477326638e-05,
+      "loss": 1.0184,
+      "step": 18960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.431732328151344e-05,
+      "loss": 1.0133,
+      "step": 18965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.426529333315657e-05,
+      "loss": 1.065,
+      "step": 18970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.421326494232691e-05,
+      "loss": 1.0759,
+      "step": 18975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.416123812315515e-05,
+      "loss": 1.0336,
+      "step": 18980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.410921288977165e-05,
+      "loss": 0.9559,
+      "step": 18985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.405718925630625e-05,
+      "loss": 1.0605,
+      "step": 18990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.400516723688842e-05,
+      "loss": 1.0237,
+      "step": 18995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.395314684564721e-05,
+      "loss": 1.0342,
+      "step": 19000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.390112809671112e-05,
+      "loss": 1.0295,
+      "step": 19005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.384911100420826e-05,
+      "loss": 1.062,
+      "step": 19010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.379709558226634e-05,
+      "loss": 1.076,
+      "step": 19015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.374508184501261e-05,
+      "loss": 1.0548,
+      "step": 19020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.369306980657374e-05,
+      "loss": 1.0735,
+      "step": 19025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.364105948107604e-05,
+      "loss": 0.9875,
+      "step": 19030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.35890508826454e-05,
+      "loss": 1.0767,
+      "step": 19035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.353704402540715e-05,
+      "loss": 1.0084,
+      "step": 19040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.34850389234862e-05,
+      "loss": 1.0401,
+      "step": 19045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.343303559100688e-05,
+      "loss": 1.1108,
+      "step": 19050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.338103404209325e-05,
+      "loss": 1.0995,
+      "step": 19055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.332903429086865e-05,
+      "loss": 1.0767,
+      "step": 19060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.327703635145609e-05,
+      "loss": 1.0869,
+      "step": 19065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.3225040237978e-05,
+      "loss": 1.0413,
+      "step": 19070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.317304596455642e-05,
+      "loss": 1.086,
+      "step": 19075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.312105354531275e-05,
+      "loss": 1.0192,
+      "step": 19080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.3069062994368e-05,
+      "loss": 0.9762,
+      "step": 19085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.301707432584257e-05,
+      "loss": 0.9809,
+      "step": 19090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.296508755385649e-05,
+      "loss": 1.0015,
+      "step": 19095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.291310269252919e-05,
+      "loss": 1.1162,
+      "step": 19100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.286111975597951e-05,
+      "loss": 1.0403,
+      "step": 19105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.280913875832587e-05,
+      "loss": 0.9406,
+      "step": 19110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.275715971368618e-05,
+      "loss": 0.9841,
+      "step": 19115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.270518263617776e-05,
+      "loss": 1.0812,
+      "step": 19120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.265320753991738e-05,
+      "loss": 1.1473,
+      "step": 19125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.260123443902129e-05,
+      "loss": 1.0274,
+      "step": 19130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.254926334760526e-05,
+      "loss": 1.1219,
+      "step": 19135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.249729427978445e-05,
+      "loss": 0.9909,
+      "step": 19140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.24453272496735e-05,
+      "loss": 1.058,
+      "step": 19145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.23933622713864e-05,
+      "loss": 1.045,
+      "step": 19150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.234139935903678e-05,
+      "loss": 1.0022,
+      "step": 19155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.228943852673754e-05,
+      "loss": 1.1066,
+      "step": 19160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.223747978860108e-05,
+      "loss": 1.1173,
+      "step": 19165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.218552315873913e-05,
+      "loss": 1.068,
+      "step": 19170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.21335686512631e-05,
+      "loss": 0.9987,
+      "step": 19175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.208161628028355e-05,
+      "loss": 0.9943,
+      "step": 19180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.202966605991059e-05,
+      "loss": 1.0014,
+      "step": 19185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.197771800425377e-05,
+      "loss": 1.1158,
+      "step": 19190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.19257721274219e-05,
+      "loss": 1.0035,
+      "step": 19195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.187382844352345e-05,
+      "loss": 1.0211,
+      "step": 19200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.182188696666605e-05,
+      "loss": 1.0379,
+      "step": 19205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.176994771095687e-05,
+      "loss": 1.0895,
+      "step": 19210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.17180106905024e-05,
+      "loss": 1.0134,
+      "step": 19215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.166607591940864e-05,
+      "loss": 1.0929,
+      "step": 19220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.161414341178084e-05,
+      "loss": 1.0536,
+      "step": 19225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.156221318172371e-05,
+      "loss": 1.0845,
+      "step": 19230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.15102852433413e-05,
+      "loss": 1.0895,
+      "step": 19235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.145835961073711e-05,
+      "loss": 1.0186,
+      "step": 19240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.140643629801398e-05,
+      "loss": 0.9856,
+      "step": 19245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.135451531927403e-05,
+      "loss": 1.0144,
+      "step": 19250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.130259668861889e-05,
+      "loss": 0.9828,
+      "step": 19255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.125068042014946e-05,
+      "loss": 1.0741,
+      "step": 19260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.119876652796608e-05,
+      "loss": 1.1327,
+      "step": 19265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.11468550261683e-05,
+      "loss": 1.0403,
+      "step": 19270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.109494592885515e-05,
+      "loss": 1.0563,
+      "step": 19275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.104303925012499e-05,
+      "loss": 1.0044,
+      "step": 19280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.099113500407547e-05,
+      "loss": 1.0216,
+      "step": 19285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.093923320480366e-05,
+      "loss": 1.009,
+      "step": 19290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.088733386640582e-05,
+      "loss": 1.0848,
+      "step": 19295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.083543700297776e-05,
+      "loss": 1.0012,
+      "step": 19300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.07835426286144e-05,
+      "loss": 1.1211,
+      "step": 19305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.073165075741014e-05,
+      "loss": 1.0528,
+      "step": 19310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.067976140345854e-05,
+      "loss": 1.0084,
+      "step": 19315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.062787458085271e-05,
+      "loss": 0.985,
+      "step": 19320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.057599030368486e-05,
+      "loss": 1.0455,
+      "step": 19325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.052410858604658e-05,
+      "loss": 1.0717,
+      "step": 19330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.047222944202878e-05,
+      "loss": 1.0461,
+      "step": 19335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.042035288572171e-05,
+      "loss": 0.9991,
+      "step": 19340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.036847893121482e-05,
+      "loss": 0.9998,
+      "step": 19345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.031660759259693e-05,
+      "loss": 1.0555,
+      "step": 19350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.026473888395607e-05,
+      "loss": 1.0372,
+      "step": 19355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.021287281937967e-05,
+      "loss": 1.0222,
+      "step": 19360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.016100941295436e-05,
+      "loss": 1.0414,
+      "step": 19365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.010914867876608e-05,
+      "loss": 1.0488,
+      "step": 19370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.005729063089996e-05,
+      "loss": 1.0192,
+      "step": 19375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.000543528344057e-05,
+      "loss": 1.0481,
+      "step": 19380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.995358265047161e-05,
+      "loss": 1.0587,
+      "step": 19385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.99017327460761e-05,
+      "loss": 1.0652,
+      "step": 19390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.984988558433626e-05,
+      "loss": 1.1306,
+      "step": 19395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.979804117933361e-05,
+      "loss": 0.9957,
+      "step": 19400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.974619954514895e-05,
+      "loss": 1.0876,
+      "step": 19405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.96943606958623e-05,
+      "loss": 1.0128,
+      "step": 19410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.964252464555288e-05,
+      "loss": 1.1002,
+      "step": 19415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.95906914082992e-05,
+      "loss": 1.0111,
+      "step": 19420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.953886099817901e-05,
+      "loss": 0.999,
+      "step": 19425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.948703342926927e-05,
+      "loss": 1.0582,
+      "step": 19430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.943520871564619e-05,
+      "loss": 1.0095,
+      "step": 19435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.938338687138511e-05,
+      "loss": 1.0403,
+      "step": 19440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.933156791056081e-05,
+      "loss": 1.0413,
+      "step": 19445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.927975184724704e-05,
+      "loss": 1.1094,
+      "step": 19450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.922793869551691e-05,
+      "loss": 1.0715,
+      "step": 19455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.917612846944263e-05,
+      "loss": 1.0836,
+      "step": 19460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.912432118309582e-05,
+      "loss": 0.9936,
+      "step": 19465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.907251685054705e-05,
+      "loss": 1.0693,
+      "step": 19470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.902071548586624e-05,
+      "loss": 1.0952,
+      "step": 19475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.896891710312247e-05,
+      "loss": 1.0376,
+      "step": 19480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.891712171638402e-05,
+      "loss": 1.0159,
+      "step": 19485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.886532933971835e-05,
+      "loss": 1.0278,
+      "step": 19490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.881353998719206e-05,
+      "loss": 0.9362,
+      "step": 19495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.876175367287096e-05,
+      "loss": 1.0408,
+      "step": 19500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.870997041082008e-05,
+      "loss": 1.0393,
+      "step": 19505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.865819021510361e-05,
+      "loss": 1.021,
+      "step": 19510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.86064130997848e-05,
+      "loss": 0.9919,
+      "step": 19515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.855463907892615e-05,
+      "loss": 1.0068,
+      "step": 19520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.850286816658937e-05,
+      "loss": 1.0133,
+      "step": 19525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.845110037683523e-05,
+      "loss": 1.0431,
+      "step": 19530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.839933572372372e-05,
+      "loss": 0.9891,
+      "step": 19535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.834757422131385e-05,
+      "loss": 0.9849,
+      "step": 19540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.8295815883664e-05,
+      "loss": 1.0081,
+      "step": 19545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.824406072483149e-05,
+      "loss": 0.9854,
+      "step": 19550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.819230875887288e-05,
+      "loss": 1.0988,
+      "step": 19555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.814055999984374e-05,
+      "loss": 1.0118,
+      "step": 19560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.808881446179902e-05,
+      "loss": 1.0347,
+      "step": 19565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.803707215879252e-05,
+      "loss": 1.0522,
+      "step": 19570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.79853331048773e-05,
+      "loss": 1.0272,
+      "step": 19575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.79335973141055e-05,
+      "loss": 1.0951,
+      "step": 19580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.788186480052844e-05,
+      "loss": 1.0286,
+      "step": 19585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.783013557819647e-05,
+      "loss": 1.0635,
+      "step": 19590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.777840966115907e-05,
+      "loss": 1.0037,
+      "step": 19595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.77266870634648e-05,
+      "loss": 1.0106,
+      "step": 19600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.767496779916138e-05,
+      "loss": 1.0343,
+      "step": 19605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.76232518822956e-05,
+      "loss": 0.9862,
+      "step": 19610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.757153932691332e-05,
+      "loss": 1.1068,
+      "step": 19615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.751983014705948e-05,
+      "loss": 1.0674,
+      "step": 19620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.746812435677809e-05,
+      "loss": 0.9786,
+      "step": 19625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.741642197011235e-05,
+      "loss": 1.0812,
+      "step": 19630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.736472300110443e-05,
+      "loss": 1.0084,
+      "step": 19635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.731302746379557e-05,
+      "loss": 1.1247,
+      "step": 19640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.726133537222609e-05,
+      "loss": 1.0363,
+      "step": 19645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.720964674043546e-05,
+      "loss": 1.0824,
+      "step": 19650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.71579615824621e-05,
+      "loss": 0.9967,
+      "step": 19655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.710627991234351e-05,
+      "loss": 1.0766,
+      "step": 19660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.705460174411624e-05,
+      "loss": 1.0158,
+      "step": 19665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.700292709181599e-05,
+      "loss": 1.0214,
+      "step": 19670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.695125596947736e-05,
+      "loss": 1.0265,
+      "step": 19675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.68995883911341e-05,
+      "loss": 1.0964,
+      "step": 19680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.684792437081884e-05,
+      "loss": 1.0707,
+      "step": 19685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.679626392256352e-05,
+      "loss": 1.0274,
+      "step": 19690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.674460706039884e-05,
+      "loss": 1.023,
+      "step": 19695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.669295379835467e-05,
+      "loss": 1.0721,
+      "step": 19700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.664130415045978e-05,
+      "loss": 0.9963,
+      "step": 19705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.658965813074219e-05,
+      "loss": 1.0424,
+      "step": 19710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.653801575322867e-05,
+      "loss": 1.0207,
+      "step": 19715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.648637703194516e-05,
+      "loss": 1.0594,
+      "step": 19720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.643474198091653e-05,
+      "loss": 0.9973,
+      "step": 19725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.638311061416675e-05,
+      "loss": 0.9855,
+      "step": 19730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.63314829457187e-05,
+      "loss": 1.0401,
+      "step": 19735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.627985898959426e-05,
+      "loss": 1.0647,
+      "step": 19740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.622823875981433e-05,
+      "loss": 1.072,
+      "step": 19745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.61766222703988e-05,
+      "loss": 0.9667,
+      "step": 19750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.612500953536659e-05,
+      "loss": 1.0564,
+      "step": 19755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.607340056873545e-05,
+      "loss": 1.043,
+      "step": 19760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.602179538452222e-05,
+      "loss": 1.0367,
+      "step": 19765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.597019399674277e-05,
+      "loss": 1.0277,
+      "step": 19770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.591859641941181e-05,
+      "loss": 0.9827,
+      "step": 19775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.586700266654309e-05,
+      "loss": 1.1811,
+      "step": 19780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.581541275214926e-05,
+      "loss": 1.0014,
+      "step": 19785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.576382669024202e-05,
+      "loss": 1.0769,
+      "step": 19790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.571224449483194e-05,
+      "loss": 0.9879,
+      "step": 19795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.566066617992861e-05,
+      "loss": 1.0132,
+      "step": 19800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.560909175954048e-05,
+      "loss": 1.0454,
+      "step": 19805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.555752124767498e-05,
+      "loss": 1.0743,
+      "step": 19810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.550595465833855e-05,
+      "loss": 1.0543,
+      "step": 19815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.545439200553647e-05,
+      "loss": 1.0509,
+      "step": 19820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.540283330327302e-05,
+      "loss": 1.0275,
+      "step": 19825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.535127856555126e-05,
+      "loss": 1.0079,
+      "step": 19830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.529972780637345e-05,
+      "loss": 1.007,
+      "step": 19835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.524818103974047e-05,
+      "loss": 0.9611,
+      "step": 19840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.519663827965235e-05,
+      "loss": 1.0132,
+      "step": 19845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.514509954010779e-05,
+      "loss": 1.0575,
+      "step": 19850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.509356483510471e-05,
+      "loss": 0.9422,
+      "step": 19855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.504203417863964e-05,
+      "loss": 0.997,
+      "step": 19860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.499050758470822e-05,
+      "loss": 1.1067,
+      "step": 19865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.49389850673048e-05,
+      "loss": 1.0309,
+      "step": 19870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.488746664042282e-05,
+      "loss": 1.0246,
+      "step": 19875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.48359523180545e-05,
+      "loss": 0.9817,
+      "step": 19880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.47844421141909e-05,
+      "loss": 1.0363,
+      "step": 19885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.473293604282204e-05,
+      "loss": 1.0398,
+      "step": 19890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.468143411793683e-05,
+      "loss": 0.9841,
+      "step": 19895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.462993635352303e-05,
+      "loss": 1.0846,
+      "step": 19900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.457844276356721e-05,
+      "loss": 1.0432,
+      "step": 19905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.452695336205485e-05,
+      "loss": 1.0889,
+      "step": 19910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.447546816297037e-05,
+      "loss": 0.9778,
+      "step": 19915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.442398718029691e-05,
+      "loss": 1.098,
+      "step": 19920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.43725104280166e-05,
+      "loss": 1.0139,
+      "step": 19925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.432103792011027e-05,
+      "loss": 1.1115,
+      "step": 19930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.426956967055774e-05,
+      "loss": 1.0365,
+      "step": 19935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.421810569333761e-05,
+      "loss": 1.0237,
+      "step": 19940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.416664600242733e-05,
+      "loss": 1.025,
+      "step": 19945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.41151906118031e-05,
+      "loss": 1.0474,
+      "step": 19950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.406373953544018e-05,
+      "loss": 0.9769,
+      "step": 19955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.40122927873124e-05,
+      "loss": 1.0006,
+      "step": 19960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.396085038139254e-05,
+      "loss": 1.0814,
+      "step": 19965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.390941233165219e-05,
+      "loss": 1.0761,
+      "step": 19970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.385797865206178e-05,
+      "loss": 1.0671,
+      "step": 19975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.380654935659054e-05,
+      "loss": 1.0598,
+      "step": 19980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.375512445920644e-05,
+      "loss": 1.0675,
+      "step": 19985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.37037039738763e-05,
+      "loss": 1.0792,
+      "step": 19990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.365228791456582e-05,
+      "loss": 0.9376,
+      "step": 19995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.360087629523941e-05,
+      "loss": 1.0024,
+      "step": 20000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.354946912986027e-05,
+      "loss": 1.0184,
+      "step": 20005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.349806643239042e-05,
+      "loss": 1.0091,
+      "step": 20010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.344666821679064e-05,
+      "loss": 0.9846,
+      "step": 20015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.339527449702057e-05,
+      "loss": 1.0314,
+      "step": 20020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.334388528703856e-05,
+      "loss": 1.038,
+      "step": 20025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.329250060080171e-05,
+      "loss": 0.9777,
+      "step": 20030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.324112045226593e-05,
+      "loss": 1.0191,
+      "step": 20035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.318974485538594e-05,
+      "loss": 1.0784,
+      "step": 20040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.313837382411518e-05,
+      "loss": 1.0337,
+      "step": 20045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.308700737240581e-05,
+      "loss": 1.0218,
+      "step": 20050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.303564551420878e-05,
+      "loss": 1.0479,
+      "step": 20055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.298428826347385e-05,
+      "loss": 1.039,
+      "step": 20060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.293293563414945e-05,
+      "loss": 1.0485,
+      "step": 20065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.28815876401828e-05,
+      "loss": 0.9933,
+      "step": 20070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.28302442955198e-05,
+      "loss": 1.0571,
+      "step": 20075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.277890561410518e-05,
+      "loss": 1.0464,
+      "step": 20080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.272757160988232e-05,
+      "loss": 1.0745,
+      "step": 20085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.267624229679342e-05,
+      "loss": 1.0487,
+      "step": 20090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.262491768877923e-05,
+      "loss": 1.0406,
+      "step": 20095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.257359779977948e-05,
+      "loss": 1.0946,
+      "step": 20100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.252228264373243e-05,
+      "loss": 1.1052,
+      "step": 20105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.247097223457509e-05,
+      "loss": 1.0344,
+      "step": 20110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.241966658624316e-05,
+      "loss": 0.989,
+      "step": 20115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.236836571267117e-05,
+      "loss": 0.9706,
+      "step": 20120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.231706962779222e-05,
+      "loss": 0.9942,
+      "step": 20125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.226577834553817e-05,
+      "loss": 1.0648,
+      "step": 20130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.221449187983951e-05,
+      "loss": 0.9843,
+      "step": 20135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.216321024462553e-05,
+      "loss": 0.9998,
+      "step": 20140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.211193345382417e-05,
+      "loss": 1.0423,
+      "step": 20145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.206066152136196e-05,
+      "loss": 1.1048,
+      "step": 20150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.20093944611642e-05,
+      "loss": 1.0661,
+      "step": 20155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.195813228715492e-05,
+      "loss": 1.0182,
+      "step": 20160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.19068750132567e-05,
+      "loss": 1.0461,
+      "step": 20165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.185562265339088e-05,
+      "loss": 0.9456,
+      "step": 20170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.180437522147736e-05,
+      "loss": 1.1003,
+      "step": 20175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.175313273143484e-05,
+      "loss": 1.0519,
+      "step": 20180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.17018951971806e-05,
+      "loss": 1.0309,
+      "step": 20185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.165066263263058e-05,
+      "loss": 1.0265,
+      "step": 20190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.159943505169929e-05,
+      "loss": 1.1492,
+      "step": 20195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.154821246830013e-05,
+      "loss": 1.0117,
+      "step": 20200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.149699489634485e-05,
+      "loss": 1.0529,
+      "step": 20205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.144578234974405e-05,
+      "loss": 1.0871,
+      "step": 20210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.139457484240687e-05,
+      "loss": 1.0707,
+      "step": 20215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.1343372388241e-05,
+      "loss": 1.0291,
+      "step": 20220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.129217500115302e-05,
+      "loss": 1.0278,
+      "step": 20225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.124098269504787e-05,
+      "loss": 1.0559,
+      "step": 20230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.118979548382924e-05,
+      "loss": 1.0535,
+      "step": 20235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.113861338139934e-05,
+      "loss": 1.119,
+      "step": 20240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.108743640165917e-05,
+      "loss": 1.1151,
+      "step": 20245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.103626455850815e-05,
+      "loss": 1.0692,
+      "step": 20250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.09850978658444e-05,
+      "loss": 1.0262,
+      "step": 20255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.093393633756459e-05,
+      "loss": 1.0921,
+      "step": 20260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.088277998756411e-05,
+      "loss": 1.0789,
+      "step": 20265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.083162882973679e-05,
+      "loss": 1.0858,
+      "step": 20270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.078048287797512e-05,
+      "loss": 1.0099,
+      "step": 20275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.072934214617014e-05,
+      "loss": 0.9959,
+      "step": 20280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.067820664821155e-05,
+      "loss": 1.0503,
+      "step": 20285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.06270763979876e-05,
+      "loss": 1.0409,
+      "step": 20290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.057595140938502e-05,
+      "loss": 1.0584,
+      "step": 20295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.052483169628923e-05,
+      "loss": 1.049,
+      "step": 20300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.04737172725842e-05,
+      "loss": 1.0352,
+      "step": 20305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.042260815215237e-05,
+      "loss": 1.0053,
+      "step": 20310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.03715043488749e-05,
+      "loss": 1.1098,
+      "step": 20315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.032040587663129e-05,
+      "loss": 0.9724,
+      "step": 20320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.02693127492998e-05,
+      "loss": 1.075,
+      "step": 20325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.021822498075715e-05,
+      "loss": 0.9909,
+      "step": 20330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.01671425848786e-05,
+      "loss": 1.0475,
+      "step": 20335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.01160655755379e-05,
+      "loss": 1.0481,
+      "step": 20340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.00649939666075e-05,
+      "loss": 0.9658,
+      "step": 20345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.001392777195821e-05,
+      "loss": 1.0676,
+      "step": 20350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.996286700545944e-05,
+      "loss": 1.0189,
+      "step": 20355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.991181168097912e-05,
+      "loss": 1.0566,
+      "step": 20360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.986076181238374e-05,
+      "loss": 1.0728,
+      "step": 20365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.980971741353828e-05,
+      "loss": 1.1276,
+      "step": 20370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.975867849830618e-05,
+      "loss": 0.9845,
+      "step": 20375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.970764508054944e-05,
+      "loss": 1.0863,
+      "step": 20380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.965661717412861e-05,
+      "loss": 1.0802,
+      "step": 20385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.960559479290271e-05,
+      "loss": 1.0586,
+      "step": 20390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.955457795072918e-05,
+      "loss": 0.9488,
+      "step": 20395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.950356666146404e-05,
+      "loss": 1.0166,
+      "step": 20400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.945256093896182e-05,
+      "loss": 1.0272,
+      "step": 20405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.940156079707552e-05,
+      "loss": 1.0289,
+      "step": 20410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.935056624965658e-05,
+      "loss": 1.2075,
+      "step": 20415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.929957731055491e-05,
+      "loss": 1.1268,
+      "step": 20420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.924859399361898e-05,
+      "loss": 1.0521,
+      "step": 20425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.919761631269568e-05,
+      "loss": 1.0836,
+      "step": 20430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.914664428163044e-05,
+      "loss": 1.0285,
+      "step": 20435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.909567791426697e-05,
+      "loss": 0.9632,
+      "step": 20440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.904471722444764e-05,
+      "loss": 1.0638,
+      "step": 20445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.899376222601319e-05,
+      "loss": 1.115,
+      "step": 20450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.894281293280284e-05,
+      "loss": 1.0408,
+      "step": 20455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.889186935865425e-05,
+      "loss": 0.9961,
+      "step": 20460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.884093151740348e-05,
+      "loss": 1.0027,
+      "step": 20465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.878999942288514e-05,
+      "loss": 1.063,
+      "step": 20470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.873907308893218e-05,
+      "loss": 1.0017,
+      "step": 20475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.868815252937604e-05,
+      "loss": 1.1122,
+      "step": 20480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.863723775804651e-05,
+      "loss": 1.034,
+      "step": 20485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.858632878877199e-05,
+      "loss": 1.0552,
+      "step": 20490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.85354256353791e-05,
+      "loss": 1.0095,
+      "step": 20495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.848452831169298e-05,
+      "loss": 1.0155,
+      "step": 20500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.843363683153717e-05,
+      "loss": 1.071,
+      "step": 20505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.838275120873367e-05,
+      "loss": 0.9675,
+      "step": 20510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.833187145710282e-05,
+      "loss": 1.0667,
+      "step": 20515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.828099759046336e-05,
+      "loss": 1.0239,
+      "step": 20520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.823012962263248e-05,
+      "loss": 0.9948,
+      "step": 20525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.817926756742576e-05,
+      "loss": 1.0068,
+      "step": 20530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.812841143865719e-05,
+      "loss": 1.0498,
+      "step": 20535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.807756125013907e-05,
+      "loss": 1.0642,
+      "step": 20540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.802671701568215e-05,
+      "loss": 1.0512,
+      "step": 20545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.797587874909561e-05,
+      "loss": 1.0076,
+      "step": 20550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.79250464641869e-05,
+      "loss": 1.0092,
+      "step": 20555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.787422017476195e-05,
+      "loss": 1.0627,
+      "step": 20560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.782339989462493e-05,
+      "loss": 0.9308,
+      "step": 20565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.777258563757851e-05,
+      "loss": 0.9622,
+      "step": 20570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.772177741742369e-05,
+      "loss": 1.0578,
+      "step": 20575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.767097524795979e-05,
+      "loss": 1.0613,
+      "step": 20580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.762017914298444e-05,
+      "loss": 0.9722,
+      "step": 20585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.756938911629385e-05,
+      "loss": 0.9329,
+      "step": 20590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.751860518168229e-05,
+      "loss": 1.0925,
+      "step": 20595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.746782735294257e-05,
+      "loss": 1.064,
+      "step": 20600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.741705564386572e-05,
+      "loss": 1.0534,
+      "step": 20605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.736629006824127e-05,
+      "loss": 1.0442,
+      "step": 20610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.731553063985688e-05,
+      "loss": 1.0477,
+      "step": 20615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.726477737249869e-05,
+      "loss": 1.0563,
+      "step": 20620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.721403027995112e-05,
+      "loss": 1.014,
+      "step": 20625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.716328937599684e-05,
+      "loss": 1.0804,
+      "step": 20630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.711255467441702e-05,
+      "loss": 0.9797,
+      "step": 20635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.706182618899099e-05,
+      "loss": 1.0717,
+      "step": 20640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.701110393349641e-05,
+      "loss": 0.9689,
+      "step": 20645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.696038792170928e-05,
+      "loss": 1.0407,
+      "step": 20650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.690967816740394e-05,
+      "loss": 1.0465,
+      "step": 20655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.685897468435299e-05,
+      "loss": 1.0416,
+      "step": 20660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.680827748632728e-05,
+      "loss": 1.019,
+      "step": 20665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.675758658709599e-05,
+      "loss": 1.0309,
+      "step": 20670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.670690200042667e-05,
+      "loss": 1.071,
+      "step": 20675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.665622374008506e-05,
+      "loss": 1.1,
+      "step": 20680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.660555181983518e-05,
+      "loss": 1.0473,
+      "step": 20685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.655488625343934e-05,
+      "loss": 1.0667,
+      "step": 20690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.650422705465818e-05,
+      "loss": 1.0735,
+      "step": 20695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.645357423725057e-05,
+      "loss": 1.0172,
+      "step": 20700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.640292781497365e-05,
+      "loss": 1.059,
+      "step": 20705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.635228780158274e-05,
+      "loss": 0.9599,
+      "step": 20710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.630165421083161e-05,
+      "loss": 1.0773,
+      "step": 20715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.625102705647212e-05,
+      "loss": 0.9966,
+      "step": 20720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.620040635225444e-05,
+      "loss": 0.9644,
+      "step": 20725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.614979211192693e-05,
+      "loss": 1.0969,
+      "step": 20730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.609918434923637e-05,
+      "loss": 1.0133,
+      "step": 20735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.604858307792758e-05,
+      "loss": 1.1019,
+      "step": 20740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.599798831174371e-05,
+      "loss": 1.0323,
+      "step": 20745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.59474000644261e-05,
+      "loss": 1.0687,
+      "step": 20750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.589681834971443e-05,
+      "loss": 1.0403,
+      "step": 20755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.584624318134643e-05,
+      "loss": 1.0164,
+      "step": 20760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.579567457305821e-05,
+      "loss": 1.0435,
+      "step": 20765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.574511253858399e-05,
+      "loss": 1.0278,
+      "step": 20770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.569455709165629e-05,
+      "loss": 0.972,
+      "step": 20775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.564400824600583e-05,
+      "loss": 1.0591,
+      "step": 20780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.559346601536142e-05,
+      "loss": 1.0275,
+      "step": 20785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.55429304134502e-05,
+      "loss": 1.1189,
+      "step": 20790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.549240145399749e-05,
+      "loss": 1.0474,
+      "step": 20795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.544187915072676e-05,
+      "loss": 1.0348,
+      "step": 20800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.539136351735975e-05,
+      "loss": 1.1389,
+      "step": 20805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.534085456761625e-05,
+      "loss": 1.0451,
+      "step": 20810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.52903523152144e-05,
+      "loss": 1.095,
+      "step": 20815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.52398567738704e-05,
+      "loss": 1.0086,
+      "step": 20820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.518936795729872e-05,
+      "loss": 1.0101,
+      "step": 20825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.513888587921189e-05,
+      "loss": 1.0322,
+      "step": 20830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.508841055332069e-05,
+      "loss": 0.9902,
+      "step": 20835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.503794199333408e-05,
+      "loss": 1.0357,
+      "step": 20840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.498748021295914e-05,
+      "loss": 1.0299,
+      "step": 20845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.493702522590114e-05,
+      "loss": 0.963,
+      "step": 20850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.488657704586342e-05,
+      "loss": 1.0577,
+      "step": 20855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.48361356865476e-05,
+      "loss": 1.0572,
+      "step": 20860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.478570116165335e-05,
+      "loss": 1.035,
+      "step": 20865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.473527348487859e-05,
+      "loss": 1.0197,
+      "step": 20870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.468485266991916e-05,
+      "loss": 0.9865,
+      "step": 20875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.463443873046936e-05,
+      "loss": 1.0324,
+      "step": 20880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.458403168022133e-05,
+      "loss": 1.0307,
+      "step": 20885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.45336315328655e-05,
+      "loss": 1.0425,
+      "step": 20890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.448323830209037e-05,
+      "loss": 0.9878,
+      "step": 20895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.443285200158256e-05,
+      "loss": 0.9753,
+      "step": 20900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.43824726450269e-05,
+      "loss": 1.0525,
+      "step": 20905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.433210024610615e-05,
+      "loss": 1.0253,
+      "step": 20910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.428173481850131e-05,
+      "loss": 1.0437,
+      "step": 20915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.42313763758915e-05,
+      "loss": 1.075,
+      "step": 20920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.418102493195391e-05,
+      "loss": 1.0196,
+      "step": 20925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.413068050036379e-05,
+      "loss": 1.0875,
+      "step": 20930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.40803430947945e-05,
+      "loss": 1.065,
+      "step": 20935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.403001272891758e-05,
+      "loss": 1.1119,
+      "step": 20940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.397968941640254e-05,
+      "loss": 1.0132,
+      "step": 20945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.392937317091704e-05,
+      "loss": 1.0434,
+      "step": 20950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.387906400612678e-05,
+      "loss": 1.0286,
+      "step": 20955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.382876193569559e-05,
+      "loss": 1.0776,
+      "step": 20960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.377846697328533e-05,
+      "loss": 1.0002,
+      "step": 20965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.372817913255598e-05,
+      "loss": 1.0014,
+      "step": 20970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.367789842716546e-05,
+      "loss": 1.03,
+      "step": 20975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.362762487076995e-05,
+      "loss": 1.0707,
+      "step": 20980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.357735847702351e-05,
+      "loss": 0.9777,
+      "step": 20985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.352709925957837e-05,
+      "loss": 1.0573,
+      "step": 20990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.34768472320847e-05,
+      "loss": 1.0957,
+      "step": 20995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.342660240819087e-05,
+      "loss": 1.0524,
+      "step": 21000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.337636480154314e-05,
+      "loss": 1.1013,
+      "step": 21005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.33261344257859e-05,
+      "loss": 0.9895,
+      "step": 21010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.327591129456153e-05,
+      "loss": 1.099,
+      "step": 21015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.32256954215105e-05,
+      "loss": 1.0233,
+      "step": 21020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.317548682027127e-05,
+      "loss": 1.0572,
+      "step": 21025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.312528550448033e-05,
+      "loss": 1.0569,
+      "step": 21030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.307509148777214e-05,
+      "loss": 1.0385,
+      "step": 21035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.302490478377925e-05,
+      "loss": 1.0655,
+      "step": 21040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.297472540613227e-05,
+      "loss": 1.01,
+      "step": 21045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.292455336845967e-05,
+      "loss": 1.0154,
+      "step": 21050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.287438868438804e-05,
+      "loss": 1.0352,
+      "step": 21055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.282423136754191e-05,
+      "loss": 1.0227,
+      "step": 21060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.27740814315439e-05,
+      "loss": 1.0199,
+      "step": 21065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.272393889001454e-05,
+      "loss": 1.0171,
+      "step": 21070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.267380375657234e-05,
+      "loss": 1.0911,
+      "step": 21075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.262367604483384e-05,
+      "loss": 1.0005,
+      "step": 21080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.257355576841363e-05,
+      "loss": 0.9668,
+      "step": 21085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.252344294092413e-05,
+      "loss": 1.0422,
+      "step": 21090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.24733375759759e-05,
+      "loss": 1.0111,
+      "step": 21095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.242323968717727e-05,
+      "loss": 1.041,
+      "step": 21100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.237314928813477e-05,
+      "loss": 1.115,
+      "step": 21105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.232306639245273e-05,
+      "loss": 1.0297,
+      "step": 21110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.227299101373352e-05,
+      "loss": 1.0608,
+      "step": 21115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.222292316557739e-05,
+      "loss": 1.0178,
+      "step": 21120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.217286286158272e-05,
+      "loss": 1.1094,
+      "step": 21125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.212281011534561e-05,
+      "loss": 1.0823,
+      "step": 21130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.207276494046027e-05,
+      "loss": 1.0058,
+      "step": 21135
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.202272735051877e-05,
+      "loss": 1.0488,
+      "step": 21140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.197269735911121e-05,
+      "loss": 1.007,
+      "step": 21145
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.19226749798255e-05,
+      "loss": 0.9527,
+      "step": 21150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.18726602262476e-05,
+      "loss": 1.003,
+      "step": 21155
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.18226531119613e-05,
+      "loss": 0.9969,
+      "step": 21160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.177265365054842e-05,
+      "loss": 1.0685,
+      "step": 21165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.172266185558865e-05,
+      "loss": 1.0057,
+      "step": 21170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.167267774065956e-05,
+      "loss": 1.0718,
+      "step": 21175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.162270131933664e-05,
+      "loss": 1.0914,
+      "step": 21180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.157273260519337e-05,
+      "loss": 1.0093,
+      "step": 21185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.15227716118011e-05,
+      "loss": 1.0721,
+      "step": 21190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.147281835272906e-05,
+      "loss": 1.0187,
+      "step": 21195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.14228728415443e-05,
+      "loss": 1.0526,
+      "step": 21200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.137293509181197e-05,
+      "loss": 0.9904,
+      "step": 21205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.132300511709495e-05,
+      "loss": 0.9864,
+      "step": 21210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.127308293095408e-05,
+      "loss": 1.0133,
+      "step": 21215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.122316854694797e-05,
+      "loss": 1.0415,
+      "step": 21220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.117326197863333e-05,
+      "loss": 0.9956,
+      "step": 21225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.112336323956453e-05,
+      "loss": 0.9833,
+      "step": 21230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.107347234329393e-05,
+      "loss": 1.0292,
+      "step": 21235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.102358930337174e-05,
+      "loss": 0.9817,
+      "step": 21240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.097371413334596e-05,
+      "loss": 0.9572,
+      "step": 21245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.092384684676262e-05,
+      "loss": 1.0192,
+      "step": 21250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.087398745716544e-05,
+      "loss": 1.0864,
+      "step": 21255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.082413597809612e-05,
+      "loss": 1.0634,
+      "step": 21260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.077429242309405e-05,
+      "loss": 1.0421,
+      "step": 21265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.07244568056967e-05,
+      "loss": 0.997,
+      "step": 21270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.067462913943919e-05,
+      "loss": 0.977,
+      "step": 21275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.062480943785455e-05,
+      "loss": 1.0244,
+      "step": 21280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.057499771447364e-05,
+      "loss": 0.9681,
+      "step": 21285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.05251939828252e-05,
+      "loss": 1.0397,
+      "step": 21290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.047539825643572e-05,
+      "loss": 0.9898,
+      "step": 21295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.042561054882956e-05,
+      "loss": 0.9757,
+      "step": 21300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.037583087352886e-05,
+      "loss": 1.0079,
+      "step": 21305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.032605924405369e-05,
+      "loss": 1.0019,
+      "step": 21310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.027629567392185e-05,
+      "loss": 1.012,
+      "step": 21315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.02265401766489e-05,
+      "loss": 1.0205,
+      "step": 21320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.017679276574824e-05,
+      "loss": 1.007,
+      "step": 21325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.012705345473122e-05,
+      "loss": 0.9901,
+      "step": 21330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.00773222571068e-05,
+      "loss": 1.0259,
+      "step": 21335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.002759918638183e-05,
+      "loss": 1.0575,
+      "step": 21340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.997788425606087e-05,
+      "loss": 0.9954,
+      "step": 21345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.992817747964639e-05,
+      "loss": 0.9667,
+      "step": 21350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.987847887063857e-05,
+      "loss": 1.0434,
+      "step": 21355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.982878844253542e-05,
+      "loss": 1.0775,
+      "step": 21360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.977910620883259e-05,
+      "loss": 1.0187,
+      "step": 21365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.972943218302374e-05,
+      "loss": 1.0345,
+      "step": 21370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.967976637860009e-05,
+      "loss": 0.9984,
+      "step": 21375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.963010880905071e-05,
+      "loss": 0.9697,
+      "step": 21380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.958045948786245e-05,
+      "loss": 1.0702,
+      "step": 21385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.953081842851992e-05,
+      "loss": 0.9962,
+      "step": 21390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.948118564450543e-05,
+      "loss": 1.1215,
+      "step": 21395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.943156114929907e-05,
+      "loss": 1.0921,
+      "step": 21400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.938194495637871e-05,
+      "loss": 1.0253,
+      "step": 21405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.933233707921996e-05,
+      "loss": 1.0581,
+      "step": 21410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.928273753129613e-05,
+      "loss": 0.9602,
+      "step": 21415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.92331463260783e-05,
+      "loss": 1.0642,
+      "step": 21420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.918356347703523e-05,
+      "loss": 1.0929,
+      "step": 21425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.913398899763352e-05,
+      "loss": 1.0574,
+      "step": 21430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.908442290133743e-05,
+      "loss": 1.1696,
+      "step": 21435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.903486520160893e-05,
+      "loss": 0.9912,
+      "step": 21440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.898531591190771e-05,
+      "loss": 1.0293,
+      "step": 21445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.893577504569119e-05,
+      "loss": 0.9497,
+      "step": 21450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.888624261641453e-05,
+      "loss": 1.0432,
+      "step": 21455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.883671863753059e-05,
+      "loss": 1.0041,
+      "step": 21460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.878720312248985e-05,
+      "loss": 1.034,
+      "step": 21465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.873769608474059e-05,
+      "loss": 1.1052,
+      "step": 21470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.868819753772879e-05,
+      "loss": 1.0642,
+      "step": 21475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.863870749489805e-05,
+      "loss": 0.964,
+      "step": 21480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.858922596968974e-05,
+      "loss": 1.0292,
+      "step": 21485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.85397529755428e-05,
+      "loss": 1.0618,
+      "step": 21490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.849028852589403e-05,
+      "loss": 1.0198,
+      "step": 21495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.844083263417773e-05,
+      "loss": 1.0171,
+      "step": 21500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.839138531382603e-05,
+      "loss": 1.0647,
+      "step": 21505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.834194657826853e-05,
+      "loss": 0.9924,
+      "step": 21510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.82925164409328e-05,
+      "loss": 1.1799,
+      "step": 21515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.824309491524377e-05,
+      "loss": 1.0201,
+      "step": 21520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.81936820146242e-05,
+      "loss": 1.0877,
+      "step": 21525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.814427775249444e-05,
+      "loss": 1.0297,
+      "step": 21530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.809488214227262e-05,
+      "loss": 1.0273,
+      "step": 21535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.804549519737431e-05,
+      "loss": 1.0013,
+      "step": 21540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.79961169312129e-05,
+      "loss": 1.0006,
+      "step": 21545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.794674735719932e-05,
+      "loss": 0.9971,
+      "step": 21550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.789738648874223e-05,
+      "loss": 1.036,
+      "step": 21555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.784803433924788e-05,
+      "loss": 1.0316,
+      "step": 21560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.77986909221201e-05,
+      "loss": 1.018,
+      "step": 21565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.77493562507604e-05,
+      "loss": 1.042,
+      "step": 21570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.770003033856795e-05,
+      "loss": 1.0631,
+      "step": 21575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.765071319893951e-05,
+      "loss": 1.0074,
+      "step": 21580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.760140484526943e-05,
+      "loss": 0.9571,
+      "step": 21585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.755210529094966e-05,
+      "loss": 1.0485,
+      "step": 21590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.750281454936984e-05,
+      "loss": 1.0482,
+      "step": 21595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.745353263391718e-05,
+      "loss": 1.0444,
+      "step": 21600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.740425955797645e-05,
+      "loss": 1.016,
+      "step": 21605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.735499533493003e-05,
+      "loss": 0.9718,
+      "step": 21610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.7305739978158e-05,
+      "loss": 1.0851,
+      "step": 21615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.725649350103788e-05,
+      "loss": 0.9837,
+      "step": 21620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.72072559169449e-05,
+      "loss": 1.1297,
+      "step": 21625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.71580272392518e-05,
+      "loss": 1.0344,
+      "step": 21630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.710880748132887e-05,
+      "loss": 1.0933,
+      "step": 21635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.705959665654411e-05,
+      "loss": 1.0745,
+      "step": 21640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.7010394778263e-05,
+      "loss": 1.0246,
+      "step": 21645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.69612018598486e-05,
+      "loss": 1.0386,
+      "step": 21650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.691201791466148e-05,
+      "loss": 1.0815,
+      "step": 21655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.686284295605996e-05,
+      "loss": 0.9723,
+      "step": 21660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.681367699739971e-05,
+      "loss": 1.1018,
+      "step": 21665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.676452005203406e-05,
+      "loss": 0.9758,
+      "step": 21670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.671537213331383e-05,
+      "loss": 1.0906,
+      "step": 21675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.666623325458754e-05,
+      "loss": 1.0115,
+      "step": 21680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.661710342920102e-05,
+      "loss": 1.0424,
+      "step": 21685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.656798267049786e-05,
+      "loss": 1.0698,
+      "step": 21690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.651887099181902e-05,
+      "loss": 0.9792,
+      "step": 21695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.646976840650312e-05,
+      "loss": 1.0206,
+      "step": 21700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.642067492788626e-05,
+      "loss": 0.9743,
+      "step": 21705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.637159056930204e-05,
+      "loss": 1.0986,
+      "step": 21710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.632251534408161e-05,
+      "loss": 0.9439,
+      "step": 21715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.627344926555363e-05,
+      "loss": 1.0671,
+      "step": 21720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.622439234704433e-05,
+      "loss": 1.0098,
+      "step": 21725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.617534460187737e-05,
+      "loss": 1.0365,
+      "step": 21730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.612630604337394e-05,
+      "loss": 1.0042,
+      "step": 21735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.607727668485276e-05,
+      "loss": 1.0486,
+      "step": 21740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.602825653963006e-05,
+      "loss": 0.9115,
+      "step": 21745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.597924562101956e-05,
+      "loss": 0.977,
+      "step": 21750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.593024394233236e-05,
+      "loss": 1.025,
+      "step": 21755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.58812515168773e-05,
+      "loss": 1.0721,
+      "step": 21760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.583226835796046e-05,
+      "loss": 1.0149,
+      "step": 21765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.578329447888552e-05,
+      "loss": 1.0312,
+      "step": 21770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.573432989295363e-05,
+      "loss": 1.0394,
+      "step": 21775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.568537461346345e-05,
+      "loss": 0.974,
+      "step": 21780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.5636428653711e-05,
+      "loss": 0.9556,
+      "step": 21785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.558749202698988e-05,
+      "loss": 1.0889,
+      "step": 21790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.553856474659108e-05,
+      "loss": 1.0775,
+      "step": 21795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.548964682580312e-05,
+      "loss": 0.9611,
+      "step": 21800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.544073827791197e-05,
+      "loss": 0.9315,
+      "step": 21805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.539183911620097e-05,
+      "loss": 1.0437,
+      "step": 21810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.534294935395096e-05,
+      "loss": 1.0201,
+      "step": 21815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.52940690044403e-05,
+      "loss": 0.9683,
+      "step": 21820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.524519808094472e-05,
+      "loss": 1.0442,
+      "step": 21825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.519633659673736e-05,
+      "loss": 1.0363,
+      "step": 21830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.514748456508885e-05,
+      "loss": 1.0045,
+      "step": 21835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.509864199926722e-05,
+      "loss": 1.0309,
+      "step": 21840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.5049808912538e-05,
+      "loss": 1.0241,
+      "step": 21845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.500098531816408e-05,
+      "loss": 1.053,
+      "step": 21850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.495217122940573e-05,
+      "loss": 1.0612,
+      "step": 21855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.490336665952075e-05,
+      "loss": 1.0393,
+      "step": 21860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.485457162176427e-05,
+      "loss": 0.9692,
+      "step": 21865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.48057861293889e-05,
+      "loss": 1.0318,
+      "step": 21870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.475701019564457e-05,
+      "loss": 1.0194,
+      "step": 21875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.470824383377867e-05,
+      "loss": 0.9939,
+      "step": 21880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.465948705703601e-05,
+      "loss": 0.9914,
+      "step": 21885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.461073987865873e-05,
+      "loss": 0.9881,
+      "step": 21890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.456200231188645e-05,
+      "loss": 1.0155,
+      "step": 21895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.451327436995605e-05,
+      "loss": 1.0796,
+      "step": 21900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.446455606610196e-05,
+      "loss": 1.087,
+      "step": 21905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.441584741355587e-05,
+      "loss": 1.1039,
+      "step": 21910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.43671484255469e-05,
+      "loss": 1.0081,
+      "step": 21915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.43184591153015e-05,
+      "loss": 1.0351,
+      "step": 21920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.426977949604358e-05,
+      "loss": 0.9946,
+      "step": 21925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.422110958099432e-05,
+      "loss": 1.0562,
+      "step": 21930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.417244938337232e-05,
+      "loss": 1.0539,
+      "step": 21935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.412379891639349e-05,
+      "loss": 1.0099,
+      "step": 21940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.40751581932712e-05,
+      "loss": 1.0153,
+      "step": 21945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.40265272272161e-05,
+      "loss": 1.0834,
+      "step": 21950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.397790603143616e-05,
+      "loss": 0.9907,
+      "step": 21955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.392929461913671e-05,
+      "loss": 1.0546,
+      "step": 21960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.388069300352053e-05,
+      "loss": 1.0626,
+      "step": 21965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.38321011977876e-05,
+      "loss": 1.0607,
+      "step": 21970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.378351921513533e-05,
+      "loss": 1.0059,
+      "step": 21975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.373494706875836e-05,
+      "loss": 1.0457,
+      "step": 21980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.36863847718488e-05,
+      "loss": 1.0239,
+      "step": 21985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.363783233759594e-05,
+      "loss": 1.0057,
+      "step": 21990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.358928977918654e-05,
+      "loss": 1.081,
+      "step": 21995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.354075710980449e-05,
+      "loss": 0.9775,
+      "step": 22000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.34922343426312e-05,
+      "loss": 1.0266,
+      "step": 22005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.344372149084523e-05,
+      "loss": 1.1396,
+      "step": 22010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.339521856762254e-05,
+      "loss": 1.018,
+      "step": 22015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.334672558613634e-05,
+      "loss": 1.0971,
+      "step": 22020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.32982425595572e-05,
+      "loss": 0.9912,
+      "step": 22025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.32497695010529e-05,
+      "loss": 0.9475,
+      "step": 22030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.320130642378859e-05,
+      "loss": 0.9811,
+      "step": 22035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.315285334092668e-05,
+      "loss": 1.0516,
+      "step": 22040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.310441026562681e-05,
+      "loss": 1.0314,
+      "step": 22045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.305597721104606e-05,
+      "loss": 0.9786,
+      "step": 22050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.300755419033861e-05,
+      "loss": 1.0361,
+      "step": 22055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.295914121665602e-05,
+      "loss": 0.9937,
+      "step": 22060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.291073830314706e-05,
+      "loss": 1.0318,
+      "step": 22065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.286234546295788e-05,
+      "loss": 1.0173,
+      "step": 22070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.281396270923172e-05,
+      "loss": 1.0642,
+      "step": 22075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.276559005510922e-05,
+      "loss": 0.9787,
+      "step": 22080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.271722751372821e-05,
+      "loss": 1.0253,
+      "step": 22085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.266887509822383e-05,
+      "loss": 1.0239,
+      "step": 22090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.262053282172842e-05,
+      "loss": 0.9752,
+      "step": 22095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.257220069737156e-05,
+      "loss": 1.0337,
+      "step": 22100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.25238787382801e-05,
+      "loss": 1.0883,
+      "step": 22105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.247556695757814e-05,
+      "loss": 1.0738,
+      "step": 22110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.242726536838699e-05,
+      "loss": 1.0183,
+      "step": 22115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.237897398382522e-05,
+      "loss": 1.01,
+      "step": 22120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.233069281700856e-05,
+      "loss": 0.998,
+      "step": 22125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.228242188105004e-05,
+      "loss": 0.9595,
+      "step": 22130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.223416118905992e-05,
+      "loss": 0.9883,
+      "step": 22135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.218591075414562e-05,
+      "loss": 1.0987,
+      "step": 22140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.213767058941172e-05,
+      "loss": 1.0562,
+      "step": 22145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.208944070796022e-05,
+      "loss": 1.0778,
+      "step": 22150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.204122112289012e-05,
+      "loss": 1.0006,
+      "step": 22155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.19930118472977e-05,
+      "loss": 1.0669,
+      "step": 22160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.194481289427643e-05,
+      "loss": 1.0686,
+      "step": 22165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.189662427691704e-05,
+      "loss": 0.9937,
+      "step": 22170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.184844600830735e-05,
+      "loss": 1.0599,
+      "step": 22175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.180027810153241e-05,
+      "loss": 1.0469,
+      "step": 22180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.175212056967447e-05,
+      "loss": 0.9516,
+      "step": 22185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.1703973425813e-05,
+      "loss": 0.925,
+      "step": 22190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.165583668302457e-05,
+      "loss": 0.9317,
+      "step": 22195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.160771035438295e-05,
+      "loss": 1.0561,
+      "step": 22200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.155959445295906e-05,
+      "loss": 1.0619,
+      "step": 22205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.15114889918211e-05,
+      "loss": 1.0342,
+      "step": 22210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.146339398403431e-05,
+      "loss": 1.0479,
+      "step": 22215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.141530944266113e-05,
+      "loss": 1.0351,
+      "step": 22220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.136723538076112e-05,
+      "loss": 1.0337,
+      "step": 22225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.131917181139112e-05,
+      "loss": 1.0554,
+      "step": 22230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.127111874760499e-05,
+      "loss": 0.9963,
+      "step": 22235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.12230762024538e-05,
+      "loss": 1.1237,
+      "step": 22240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.11750441889857e-05,
+      "loss": 0.952,
+      "step": 22245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.112702272024605e-05,
+      "loss": 1.0486,
+      "step": 22250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.107901180927734e-05,
+      "loss": 0.9922,
+      "step": 22255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.103101146911916e-05,
+      "loss": 1.0016,
+      "step": 22260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0983021712808264e-05,
+      "loss": 1.0003,
+      "step": 22265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.093504255337844e-05,
+      "loss": 1.0429,
+      "step": 22270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.088707400386073e-05,
+      "loss": 1.1042,
+      "step": 22275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0839116077283234e-05,
+      "loss": 1.0487,
+      "step": 22280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.079116878667115e-05,
+      "loss": 1.0229,
+      "step": 22285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.074323214504675e-05,
+      "loss": 0.9849,
+      "step": 22290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.069530616542958e-05,
+      "loss": 1.0102,
+      "step": 22295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.064739086083607e-05,
+      "loss": 1.0163,
+      "step": 22300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0599486244279914e-05,
+      "loss": 0.968,
+      "step": 22305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.055159232877181e-05,
+      "loss": 1.0006,
+      "step": 22310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.050370912731964e-05,
+      "loss": 1.0839,
+      "step": 22315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.045583665292826e-05,
+      "loss": 1.1559,
+      "step": 22320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.040797491859972e-05,
+      "loss": 1.1212,
+      "step": 22325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.036012393733306e-05,
+      "loss": 1.0982,
+      "step": 22330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.03122837221245e-05,
+      "loss": 1.0124,
+      "step": 22335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0264454285967274e-05,
+      "loss": 1.0703,
+      "step": 22340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.021663564185167e-05,
+      "loss": 1.0294,
+      "step": 22345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0168827802765047e-05,
+      "loss": 0.9891,
+      "step": 22350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0121030781691936e-05,
+      "loss": 1.0205,
+      "step": 22355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.007324459161381e-05,
+      "loss": 1.0753,
+      "step": 22360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0025469245509205e-05,
+      "loss": 0.9947,
+      "step": 22365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.997770475635377e-05,
+      "loss": 1.0358,
+      "step": 22370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.992995113712019e-05,
+      "loss": 0.9991,
+      "step": 22375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9882208400778195e-05,
+      "loss": 1.0785,
+      "step": 22380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.983447656029456e-05,
+      "loss": 1.068,
+      "step": 22385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.978675562863301e-05,
+      "loss": 1.034,
+      "step": 22390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.973904561875452e-05,
+      "loss": 1.0238,
+      "step": 22395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.969134654361688e-05,
+      "loss": 1.0538,
+      "step": 22400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.964365841617503e-05,
+      "loss": 1.0011,
+      "step": 22405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9595981249380884e-05,
+      "loss": 1.0089,
+      "step": 22410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.954831505618346e-05,
+      "loss": 1.0016,
+      "step": 22415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9500659849528664e-05,
+      "loss": 1.0507,
+      "step": 22420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.945301564235952e-05,
+      "loss": 1.0141,
+      "step": 22425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9405382447616034e-05,
+      "loss": 1.0033,
+      "step": 22430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.935776027823523e-05,
+      "loss": 1.1135,
+      "step": 22435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.931014914715116e-05,
+      "loss": 1.0443,
+      "step": 22440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.92625490672948e-05,
+      "loss": 1.0286,
+      "step": 22445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9214960051594185e-05,
+      "loss": 1.0384,
+      "step": 22450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.91673821129743e-05,
+      "loss": 1.0772,
+      "step": 22455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9119815264357245e-05,
+      "loss": 1.0308,
+      "step": 22460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.907225951866195e-05,
+      "loss": 1.0125,
+      "step": 22465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.90247148888044e-05,
+      "loss": 1.0407,
+      "step": 22470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.897718138769755e-05,
+      "loss": 1.0572,
+      "step": 22475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.892965902825138e-05,
+      "loss": 1.0788,
+      "step": 22480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.88821478233728e-05,
+      "loss": 1.102,
+      "step": 22485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8834647785965656e-05,
+      "loss": 1.0168,
+      "step": 22490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8787158928930805e-05,
+      "loss": 1.0518,
+      "step": 22495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8739681265166094e-05,
+      "loss": 1.0672,
+      "step": 22500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.869221480756628e-05,
+      "loss": 1.0906,
+      "step": 22505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.864475956902312e-05,
+      "loss": 1.0837,
+      "step": 22510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.859731556242524e-05,
+      "loss": 1.0575,
+      "step": 22515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.854988280065832e-05,
+      "loss": 1.0595,
+      "step": 22520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8502461296604935e-05,
+      "loss": 1.016,
+      "step": 22525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8455051063144614e-05,
+      "loss": 1.0779,
+      "step": 22530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.840765211315376e-05,
+      "loss": 1.034,
+      "step": 22535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8360264459505846e-05,
+      "loss": 1.0922,
+      "step": 22540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8312888115071183e-05,
+      "loss": 1.0531,
+      "step": 22545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.826552309271702e-05,
+      "loss": 1.0164,
+      "step": 22550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.821816940530749e-05,
+      "loss": 1.0518,
+      "step": 22555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8170827065703745e-05,
+      "loss": 1.1282,
+      "step": 22560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.812349608676384e-05,
+      "loss": 0.995,
+      "step": 22565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.807617648134268e-05,
+      "loss": 1.0809,
+      "step": 22570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8028868262292055e-05,
+      "loss": 1.0238,
+      "step": 22575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.798157144246077e-05,
+      "loss": 1.0387,
+      "step": 22580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.793428603469453e-05,
+      "loss": 1.0427,
+      "step": 22585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7887012051835844e-05,
+      "loss": 1.0098,
+      "step": 22590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7839749506724086e-05,
+      "loss": 1.0782,
+      "step": 22595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7792498412195764e-05,
+      "loss": 1.0559,
+      "step": 22600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7745258781084056e-05,
+      "loss": 1.0369,
+      "step": 22605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.769803062621904e-05,
+      "loss": 0.9849,
+      "step": 22610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.765081396042773e-05,
+      "loss": 1.0181,
+      "step": 22615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.760360879653412e-05,
+      "loss": 1.0398,
+      "step": 22620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7556415147358875e-05,
+      "loss": 1.0062,
+      "step": 22625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7509233025719646e-05,
+      "loss": 1.0936,
+      "step": 22630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.746206244443093e-05,
+      "loss": 1.06,
+      "step": 22635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.741490341630418e-05,
+      "loss": 0.972,
+      "step": 22640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.736775595414756e-05,
+      "loss": 1.0549,
+      "step": 22645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.732062007076613e-05,
+      "loss": 1.046,
+      "step": 22650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.727349577896194e-05,
+      "loss": 1.0119,
+      "step": 22655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7226383091533685e-05,
+      "loss": 1.0378,
+      "step": 22660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.717928202127707e-05,
+      "loss": 1.0774,
+      "step": 22665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7132192580984615e-05,
+      "loss": 1.0147,
+      "step": 22670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7085114783445627e-05,
+      "loss": 1.0307,
+      "step": 22675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.703804864144624e-05,
+      "loss": 0.982,
+      "step": 22680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.699099416776947e-05,
+      "loss": 1.0381,
+      "step": 22685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.694395137519525e-05,
+      "loss": 1.0664,
+      "step": 22690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6896920276500165e-05,
+      "loss": 1.0649,
+      "step": 22695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6849900884457674e-05,
+      "loss": 1.0693,
+      "step": 22700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.680289321183813e-05,
+      "loss": 0.9577,
+      "step": 22705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6755897271408684e-05,
+      "loss": 1.1129,
+      "step": 22710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.670891307593327e-05,
+      "loss": 0.9631,
+      "step": 22715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.666194063817257e-05,
+      "loss": 0.9776,
+      "step": 22720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6614979970884196e-05,
+      "loss": 1.067,
+      "step": 22725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.656803108682254e-05,
+      "loss": 1.0042,
+      "step": 22730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.652109399873873e-05,
+      "loss": 1.0149,
+      "step": 22735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6474168719380625e-05,
+      "loss": 1.0012,
+      "step": 22740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.642725526149316e-05,
+      "loss": 1.0228,
+      "step": 22745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.638035363781777e-05,
+      "loss": 1.0947,
+      "step": 22750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.633346386109277e-05,
+      "loss": 1.0123,
+      "step": 22755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.628658594405327e-05,
+      "loss": 1.0747,
+      "step": 22760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.623971989943122e-05,
+      "loss": 1.0502,
+      "step": 22765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6192865739955236e-05,
+      "loss": 0.96,
+      "step": 22770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6146023478350693e-05,
+      "loss": 1.0684,
+      "step": 22775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6099193127339864e-05,
+      "loss": 1.0147,
+      "step": 22780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6052374699641755e-05,
+      "loss": 1.1096,
+      "step": 22785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6005568207972024e-05,
+      "loss": 1.1119,
+      "step": 22790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.595877366504314e-05,
+      "loss": 1.0665,
+      "step": 22795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.591199108356437e-05,
+      "loss": 1.0583,
+      "step": 22800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5865220476241784e-05,
+      "loss": 1.0542,
+      "step": 22805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5818461855777994e-05,
+      "loss": 1.0864,
+      "step": 22810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.577171523487259e-05,
+      "loss": 1.1055,
+      "step": 22815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5724980626221735e-05,
+      "loss": 1.0969,
+      "step": 22820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.567825804251843e-05,
+      "loss": 1.0596,
+      "step": 22825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.563154749645233e-05,
+      "loss": 1.0339,
+      "step": 22830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5584849000709926e-05,
+      "loss": 1.0841,
+      "step": 22835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5538162567974296e-05,
+      "loss": 1.0568,
+      "step": 22840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.54914882109254e-05,
+      "loss": 0.962,
+      "step": 22845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5444825942239765e-05,
+      "loss": 1.0881,
+      "step": 22850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.539817577459078e-05,
+      "loss": 1.0252,
+      "step": 22855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.535153772064843e-05,
+      "loss": 0.997,
+      "step": 22860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.530491179307942e-05,
+      "loss": 1.0913,
+      "step": 22865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5258298004547246e-05,
+      "loss": 1.0363,
+      "step": 22870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.521169636771209e-05,
+      "loss": 1.0639,
+      "step": 22875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5165106895230756e-05,
+      "loss": 0.988,
+      "step": 22880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.51185295997567e-05,
+      "loss": 1.0001,
+      "step": 22885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.507196449394034e-05,
+      "loss": 1.066,
+      "step": 22890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.502541159042853e-05,
+      "loss": 1.1076,
+      "step": 22895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.497887090186482e-05,
+      "loss": 1.0608,
+      "step": 22900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4932342440889564e-05,
+      "loss": 1.0028,
+      "step": 22905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4885826220139757e-05,
+      "loss": 1.1048,
+      "step": 22910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4839322252249016e-05,
+      "loss": 1.024,
+      "step": 22915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4792830549847626e-05,
+      "loss": 1.0747,
+      "step": 22920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4746351125562633e-05,
+      "loss": 1.0503,
+      "step": 22925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.469988399201773e-05,
+      "loss": 1.0338,
+      "step": 22930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.4653429161833166e-05,
+      "loss": 1.0143,
+      "step": 22935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.460698664762591e-05,
+      "loss": 1.0853,
+      "step": 22940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4560556462009615e-05,
+      "loss": 1.0827,
+      "step": 22945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.451413861759461e-05,
+      "loss": 1.0079,
+      "step": 22950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4467733126987766e-05,
+      "loss": 1.0716,
+      "step": 22955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.44213400027927e-05,
+      "loss": 1.0328,
+      "step": 22960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.437495925760958e-05,
+      "loss": 1.0953,
+      "step": 22965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.432859090403532e-05,
+      "loss": 1.0937,
+      "step": 22970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.428223495466337e-05,
+      "loss": 1.0848,
+      "step": 22975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4235891422083886e-05,
+      "loss": 1.022,
+      "step": 22980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.418956031888356e-05,
+      "loss": 1.0855,
+      "step": 22985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.414324165764584e-05,
+      "loss": 1.0069,
+      "step": 22990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4096935450950634e-05,
+      "loss": 1.0144,
+      "step": 22995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.405064171137465e-05,
+      "loss": 1.0191,
+      "step": 23000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.400436045149102e-05,
+      "loss": 0.9528,
+      "step": 23005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.395809168386962e-05,
+      "loss": 1.0537,
+      "step": 23010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.391183542107693e-05,
+      "loss": 1.0501,
+      "step": 23015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.386559167567596e-05,
+      "loss": 1.0477,
+      "step": 23020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.381936046022633e-05,
+      "loss": 1.0276,
+      "step": 23025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.37731417872843e-05,
+      "loss": 1.0735,
+      "step": 23030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.372693566940277e-05,
+      "loss": 0.9587,
+      "step": 23035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.368074211913112e-05,
+      "loss": 1.0012,
+      "step": 23040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3634561149015304e-05,
+      "loss": 1.0758,
+      "step": 23045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3588392771598e-05,
+      "loss": 1.0512,
+      "step": 23050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3542236999418385e-05,
+      "loss": 1.014,
+      "step": 23055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3496093845012196e-05,
+      "loss": 1.0385,
+      "step": 23060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.344996332091171e-05,
+      "loss": 1.0693,
+      "step": 23065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.340384543964587e-05,
+      "loss": 1.0734,
+      "step": 23070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.335774021374018e-05,
+      "loss": 1.0545,
+      "step": 23075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.331164765571661e-05,
+      "loss": 1.0067,
+      "step": 23080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3265567778093725e-05,
+      "loss": 1.0438,
+      "step": 23085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3219500593386696e-05,
+      "loss": 0.9867,
+      "step": 23090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3173446114107265e-05,
+      "loss": 0.9986,
+      "step": 23095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3127404352763574e-05,
+      "loss": 1.0792,
+      "step": 23100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3081375321860514e-05,
+      "loss": 1.0735,
+      "step": 23105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.303535903389932e-05,
+      "loss": 1.0221,
+      "step": 23110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.298935550137796e-05,
+      "loss": 1.0265,
+      "step": 23115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.294336473679075e-05,
+      "loss": 0.9942,
+      "step": 23120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2897386752628694e-05,
+      "loss": 0.9674,
+      "step": 23125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.285142156137921e-05,
+      "loss": 1.0765,
+      "step": 23130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2805469175526345e-05,
+      "loss": 1.046,
+      "step": 23135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.275952960755055e-05,
+      "loss": 0.9572,
+      "step": 23140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.271360286992893e-05,
+      "loss": 1.0608,
+      "step": 23145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.266768897513494e-05,
+      "loss": 0.9907,
+      "step": 23150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.26217879356387e-05,
+      "loss": 0.9444,
+      "step": 23155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2575899763906824e-05,
+      "loss": 1.0413,
+      "step": 23160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.253002447240233e-05,
+      "loss": 1.0723,
+      "step": 23165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2484162073584765e-05,
+      "loss": 1.026,
+      "step": 23170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.243831257991023e-05,
+      "loss": 1.058,
+      "step": 23175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.239247600383136e-05,
+      "loss": 0.9806,
+      "step": 23180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2346652357797146e-05,
+      "loss": 1.1136,
+      "step": 23185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.230084165425312e-05,
+      "loss": 1.1165,
+      "step": 23190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2255043905641343e-05,
+      "loss": 1.1087,
+      "step": 23195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.220925912440038e-05,
+      "loss": 1.1278,
+      "step": 23200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2163487322965166e-05,
+      "loss": 1.0854,
+      "step": 23205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.211772851376715e-05,
+      "loss": 0.9943,
+      "step": 23210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2071982709234324e-05,
+      "loss": 1.0254,
+      "step": 23215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.202624992179109e-05,
+      "loss": 1.0516,
+      "step": 23220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.198053016385833e-05,
+      "loss": 1.037,
+      "step": 23225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.193482344785327e-05,
+      "loss": 1.0886,
+      "step": 23230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.188912978618985e-05,
+      "loss": 0.9994,
+      "step": 23235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.184344919127827e-05,
+      "loss": 1.0089,
+      "step": 23240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.179778167552517e-05,
+      "loss": 1.0749,
+      "step": 23245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.175212725133374e-05,
+      "loss": 0.9711,
+      "step": 23250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1706485931103586e-05,
+      "loss": 1.0018,
+      "step": 23255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.166085772723073e-05,
+      "loss": 1.0036,
+      "step": 23260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1615242652107574e-05,
+      "loss": 1.0598,
+      "step": 23265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1569640718123136e-05,
+      "loss": 0.9796,
+      "step": 23270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.152405193766263e-05,
+      "loss": 0.9443,
+      "step": 23275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1478476323107914e-05,
+      "loss": 1.1372,
+      "step": 23280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.143291388683709e-05,
+      "loss": 1.0268,
+      "step": 23285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.138736464122484e-05,
+      "loss": 1.0774,
+      "step": 23290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.134182859864214e-05,
+      "loss": 1.0254,
+      "step": 23295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.129630577145641e-05,
+      "loss": 1.0682,
+      "step": 23300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.125079617203158e-05,
+      "loss": 1.1262,
+      "step": 23305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1205299812727855e-05,
+      "loss": 1.0814,
+      "step": 23310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1159816705901844e-05,
+      "loss": 0.9872,
+      "step": 23315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.111434686390666e-05,
+      "loss": 1.0258,
+      "step": 23320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.10688902990918e-05,
+      "loss": 1.0399,
+      "step": 23325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.102344702380306e-05,
+      "loss": 1.0754,
+      "step": 23330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0978017050382664e-05,
+      "loss": 0.9977,
+      "step": 23335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.093260039116927e-05,
+      "loss": 0.9582,
+      "step": 23340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.088719705849791e-05,
+      "loss": 0.9749,
+      "step": 23345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0841807064699986e-05,
+      "loss": 1.0317,
+      "step": 23350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.079643042210319e-05,
+      "loss": 0.9753,
+      "step": 23355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.075106714303173e-05,
+      "loss": 0.9702,
+      "step": 23360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0705717239806125e-05,
+      "loss": 1.0298,
+      "step": 23365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.066038072474326e-05,
+      "loss": 1.0973,
+      "step": 23370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.061505761015629e-05,
+      "loss": 1.0243,
+      "step": 23375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.056974790835496e-05,
+      "loss": 1.027,
+      "step": 23380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0524451631645174e-05,
+      "loss": 1.1367,
+      "step": 23385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.047916879232921e-05,
+      "loss": 1.0595,
+      "step": 23390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.043389940270578e-05,
+      "loss": 1.1675,
+      "step": 23395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0388643475069906e-05,
+      "loss": 1.0172,
+      "step": 23400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.034340102171293e-05,
+      "loss": 1.0002,
+      "step": 23405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0298172054922534e-05,
+      "loss": 1.0191,
+      "step": 23410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.025295658698276e-05,
+      "loss": 1.0589,
+      "step": 23415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.020775463017403e-05,
+      "loss": 1.1018,
+      "step": 23420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.016256619677301e-05,
+      "loss": 1.0038,
+      "step": 23425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0117391299052684e-05,
+      "loss": 1.0303,
+      "step": 23430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.007222994928243e-05,
+      "loss": 1.075,
+      "step": 23435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.002708215972799e-05,
+      "loss": 1.0731,
+      "step": 23440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.998194794265124e-05,
+      "loss": 1.0187,
+      "step": 23445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9936827310310595e-05,
+      "loss": 0.9958,
+      "step": 23450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9891720274960566e-05,
+      "loss": 1.0743,
+      "step": 23455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.984662684885215e-05,
+      "loss": 0.9918,
+      "step": 23460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.980154704423251e-05,
+      "loss": 1.1038,
+      "step": 23465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9756480873345235e-05,
+      "loss": 0.9632,
+      "step": 23470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.971142834843011e-05,
+      "loss": 1.02,
+      "step": 23475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.966638948172322e-05,
+      "loss": 1.0382,
+      "step": 23480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9621364285456995e-05,
+      "loss": 1.0099,
+      "step": 23485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.957635277186018e-05,
+      "loss": 0.9854,
+      "step": 23490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9531354953157704e-05,
+      "loss": 1.0535,
+      "step": 23495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9486370841570803e-05,
+      "loss": 0.9094,
+      "step": 23500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9441400449317033e-05,
+      "loss": 1.0067,
+      "step": 23505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.939644378861024e-05,
+      "loss": 0.9878,
+      "step": 23510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9351500871660474e-05,
+      "loss": 1.0436,
+      "step": 23515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9306571710674e-05,
+      "loss": 1.0238,
+      "step": 23520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9261656317853575e-05,
+      "loss": 1.0529,
+      "step": 23525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.921675470539801e-05,
+      "loss": 0.9522,
+      "step": 23530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9171866885502375e-05,
+      "loss": 1.0172,
+      "step": 23535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.912699287035811e-05,
+      "loss": 1.0055,
+      "step": 23540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.908213267215287e-05,
+      "loss": 0.9824,
+      "step": 23545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.90372863030705e-05,
+      "loss": 1.0748,
+      "step": 23550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.89924537752911e-05,
+      "loss": 1.1017,
+      "step": 23555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.894763510099106e-05,
+      "loss": 1.021,
+      "step": 23560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8902830292343024e-05,
+      "loss": 0.999,
+      "step": 23565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8858039361515795e-05,
+      "loss": 0.9548,
+      "step": 23570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8813262320674394e-05,
+      "loss": 1.0854,
+      "step": 23575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8768499181980146e-05,
+      "loss": 1.1565,
+      "step": 23580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8723749957590625e-05,
+      "loss": 0.934,
+      "step": 23585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.867901465965949e-05,
+      "loss": 1.0272,
+      "step": 23590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.863429330033677e-05,
+      "loss": 1.0316,
+      "step": 23595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.858958589176857e-05,
+      "loss": 0.9774,
+      "step": 23600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8544892446097343e-05,
+      "loss": 1.0794,
+      "step": 23605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.85002129754616e-05,
+      "loss": 0.9816,
+      "step": 23610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.845554749199622e-05,
+      "loss": 0.9492,
+      "step": 23615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.841089600783213e-05,
+      "loss": 1.0576,
+      "step": 23620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8366258535096574e-05,
+      "loss": 0.9775,
+      "step": 23625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.832163508591289e-05,
+      "loss": 1.1189,
+      "step": 23630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.827702567240072e-05,
+      "loss": 1.0222,
+      "step": 23635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.823243030667576e-05,
+      "loss": 0.9981,
+      "step": 23640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8187849000850035e-05,
+      "loss": 1.0213,
+      "step": 23645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.814328176703161e-05,
+      "loss": 1.0586,
+      "step": 23650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8098728617324873e-05,
+      "loss": 1.0136,
+      "step": 23655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8054189563830234e-05,
+      "loss": 1.0523,
+      "step": 23660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.800966461864438e-05,
+      "loss": 1.161,
+      "step": 23665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.796515379386019e-05,
+      "loss": 1.0076,
+      "step": 23670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.79206571015666e-05,
+      "loss": 1.0145,
+      "step": 23675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.787617455384874e-05,
+      "loss": 0.972,
+      "step": 23680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7831706162787957e-05,
+      "loss": 1.0061,
+      "step": 23685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7787251940461754e-05,
+      "loss": 1.0057,
+      "step": 23690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7742811898943703e-05,
+      "loss": 0.9991,
+      "step": 23695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7698386050303544e-05,
+      "loss": 1.0247,
+      "step": 23700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7653974406607226e-05,
+      "loss": 1.039,
+      "step": 23705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7609576979916845e-05,
+      "loss": 1.0237,
+      "step": 23710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.756519378229054e-05,
+      "loss": 1.0395,
+      "step": 23715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.752082482578262e-05,
+      "loss": 1.0263,
+      "step": 23720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7476470122443576e-05,
+      "loss": 1.0659,
+      "step": 23725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.743212968432003e-05,
+      "loss": 1.0864,
+      "step": 23730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.738780352345462e-05,
+      "loss": 1.0443,
+      "step": 23735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.734349165188626e-05,
+      "loss": 0.9634,
+      "step": 23740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7299194081649856e-05,
+      "loss": 1.0846,
+      "step": 23745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.725491082477652e-05,
+      "loss": 1.0824,
+      "step": 23750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7210641893293384e-05,
+      "loss": 1.1069,
+      "step": 23755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7166387299223815e-05,
+      "loss": 1.0246,
+      "step": 23760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7122147054587115e-05,
+      "loss": 1.0064,
+      "step": 23765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7077921171398895e-05,
+      "loss": 1.0085,
+      "step": 23770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7033709661670664e-05,
+      "loss": 1.0802,
+      "step": 23775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.69895125374102e-05,
+      "loss": 1.0185,
+      "step": 23780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.694532981062122e-05,
+      "loss": 1.0976,
+      "step": 23785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.690116149330367e-05,
+      "loss": 0.9947,
+      "step": 23790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6857007597453464e-05,
+      "loss": 1.0323,
+      "step": 23795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.68128681350627e-05,
+      "loss": 1.0424,
+      "step": 23800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6768743118119465e-05,
+      "loss": 1.0363,
+      "step": 23805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6724632558607974e-05,
+      "loss": 1.0163,
+      "step": 23810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.668053646850856e-05,
+      "loss": 1.0354,
+      "step": 23815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.663645485979754e-05,
+      "loss": 1.0464,
+      "step": 23820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.65923877444473e-05,
+      "loss": 0.9962,
+      "step": 23825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6548335134426336e-05,
+      "loss": 1.0191,
+      "step": 23830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.650429704169924e-05,
+      "loss": 0.9682,
+      "step": 23835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.646027347822657e-05,
+      "loss": 1.103,
+      "step": 23840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.641626445596495e-05,
+      "loss": 1.0213,
+      "step": 23845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6372269986867114e-05,
+      "loss": 1.0144,
+      "step": 23850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6328290082881856e-05,
+      "loss": 0.9686,
+      "step": 23855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.628432475595393e-05,
+      "loss": 0.9751,
+      "step": 23860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.62403740180241e-05,
+      "loss": 1.0068,
+      "step": 23865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6196437881029385e-05,
+      "loss": 1.0581,
+      "step": 23870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.615251635690262e-05,
+      "loss": 0.985,
+      "step": 23875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6108609457572706e-05,
+      "loss": 1.0123,
+      "step": 23880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6064717194964676e-05,
+      "loss": 1.0942,
+      "step": 23885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.602083958099947e-05,
+      "loss": 1.0989,
+      "step": 23890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.597697662759417e-05,
+      "loss": 1.0915,
+      "step": 23895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.593312834666171e-05,
+      "loss": 1.0765,
+      "step": 23900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.588929475011125e-05,
+      "loss": 1.0501,
+      "step": 23905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.584547584984774e-05,
+      "loss": 1.0934,
+      "step": 23910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5801671657772325e-05,
+      "loss": 0.9727,
+      "step": 23915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.575788218578203e-05,
+      "loss": 0.9537,
+      "step": 23920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.571410744576998e-05,
+      "loss": 1.0179,
+      "step": 23925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.567034744962517e-05,
+      "loss": 1.0323,
+      "step": 23930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.562660220923277e-05,
+      "loss": 1.0962,
+      "step": 23935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.558287173647374e-05,
+      "loss": 1.0285,
+      "step": 23940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5539156043225215e-05,
+      "loss": 0.9922,
+      "step": 23945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5495455141360165e-05,
+      "loss": 1.0047,
+      "step": 23950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.545176904274764e-05,
+      "loss": 0.9985,
+      "step": 23955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.540809775925267e-05,
+      "loss": 1.0361,
+      "step": 23960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.536444130273618e-05,
+      "loss": 1.0133,
+      "step": 23965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.532079968505511e-05,
+      "loss": 1.021,
+      "step": 23970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.527717291806242e-05,
+      "loss": 1.0596,
+      "step": 23975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5233561013606995e-05,
+      "loss": 1.0078,
+      "step": 23980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.518996398353367e-05,
+      "loss": 1.1767,
+      "step": 23985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5146381839683215e-05,
+      "loss": 1.0408,
+      "step": 23990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.510281459389242e-05,
+      "loss": 1.0506,
+      "step": 23995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.505926225799405e-05,
+      "loss": 1.1078,
+      "step": 24000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.501572484381673e-05,
+      "loss": 1.0494,
+      "step": 24005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.497220236318502e-05,
+      "loss": 0.9725,
+      "step": 24010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4928694827919614e-05,
+      "loss": 1.0271,
+      "step": 24015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4885202249836936e-05,
+      "loss": 1.0323,
+      "step": 24020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4841724640749396e-05,
+      "loss": 1.0988,
+      "step": 24025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4798262012465395e-05,
+      "loss": 1.0998,
+      "step": 24030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4754814376789304e-05,
+      "loss": 1.011,
+      "step": 24035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4711381745521275e-05,
+      "loss": 1.0341,
+      "step": 24040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4667964130457455e-05,
+      "loss": 1.1316,
+      "step": 24045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4624561543389965e-05,
+      "loss": 1.082,
+      "step": 24050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4581173996106815e-05,
+      "loss": 1.0976,
+      "step": 24055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4537801500391905e-05,
+      "loss": 1.0122,
+      "step": 24060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.449444406802502e-05,
+      "loss": 1.0056,
+      "step": 24065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4451101710781926e-05,
+      "loss": 1.0614,
+      "step": 24070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4407774440434304e-05,
+      "loss": 1.0769,
+      "step": 24075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.436446226874961e-05,
+      "loss": 1.0284,
+      "step": 24080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.432116520749137e-05,
+      "loss": 1.0566,
+      "step": 24085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.427788326841887e-05,
+      "loss": 1.0947,
+      "step": 24090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.423461646328733e-05,
+      "loss": 1.0225,
+      "step": 24095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.419136480384788e-05,
+      "loss": 0.9583,
+      "step": 24100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.414812830184758e-05,
+      "loss": 1.0862,
+      "step": 24105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.410490696902928e-05,
+      "loss": 1.0767,
+      "step": 24110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4061700817131714e-05,
+      "loss": 1.0255,
+      "step": 24115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4018509857889555e-05,
+      "loss": 0.9876,
+      "step": 24120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.397533410303337e-05,
+      "loss": 1.0428,
+      "step": 24125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3932173564289506e-05,
+      "loss": 1.0191,
+      "step": 24130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.388902825338019e-05,
+      "loss": 1.0826,
+      "step": 24135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3845898182023593e-05,
+      "loss": 0.9883,
+      "step": 24140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.380278336193371e-05,
+      "loss": 1.0221,
+      "step": 24145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.375968380482036e-05,
+      "loss": 1.0634,
+      "step": 24150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.371659952238917e-05,
+      "loss": 1.0419,
+      "step": 24155
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.367353052634181e-05,
+      "loss": 1.0532,
+      "step": 24160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.363047682837563e-05,
+      "loss": 0.9826,
+      "step": 24165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.358743844018384e-05,
+      "loss": 1.0173,
+      "step": 24170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.354441537345553e-05,
+      "loss": 1.0544,
+      "step": 24175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.350140763987567e-05,
+      "loss": 0.9892,
+      "step": 24180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.345841525112497e-05,
+      "loss": 1.1121,
+      "step": 24185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.341543821888001e-05,
+      "loss": 0.9874,
+      "step": 24190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3372476554813234e-05,
+      "loss": 1.0357,
+      "step": 24195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.332953027059291e-05,
+      "loss": 1.0759,
+      "step": 24200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.328659937788307e-05,
+      "loss": 1.0526,
+      "step": 24205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.324368388834359e-05,
+      "loss": 1.0329,
+      "step": 24210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3200783813630184e-05,
+      "loss": 1.0259,
+      "step": 24215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.315789916539441e-05,
+      "loss": 1.0497,
+      "step": 24220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.311502995528354e-05,
+      "loss": 0.9784,
+      "step": 24225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.307217619494075e-05,
+      "loss": 1.0273,
+      "step": 24230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.302933789600493e-05,
+      "loss": 1.0561,
+      "step": 24235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.298651507011089e-05,
+      "loss": 1.0762,
+      "step": 24240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2943707728889096e-05,
+      "loss": 1.0775,
+      "step": 24245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.290091588396593e-05,
+      "loss": 1.0461,
+      "step": 24250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.285813954696346e-05,
+      "loss": 1.0256,
+      "step": 24255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.281537872949967e-05,
+      "loss": 1.0484,
+      "step": 24260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.277263344318817e-05,
+      "loss": 1.0277,
+      "step": 24265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.272990369963853e-05,
+      "loss": 1.0292,
+      "step": 24270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.268718951045592e-05,
+      "loss": 1.0142,
+      "step": 24275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.264449088724144e-05,
+      "loss": 1.0345,
+      "step": 24280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2601807841591826e-05,
+      "loss": 0.981,
+      "step": 24285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.255914038509973e-05,
+      "loss": 1.0421,
+      "step": 24290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2516488529353436e-05,
+      "loss": 0.9925,
+      "step": 24295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.247385228593699e-05,
+      "loss": 1.1191,
+      "step": 24300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.243123166643039e-05,
+      "loss": 1.0789,
+      "step": 24305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2388626682409194e-05,
+      "loss": 0.9518,
+      "step": 24310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2346037345444734e-05,
+      "loss": 1.0821,
+      "step": 24315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2303463667104146e-05,
+      "loss": 1.0193,
+      "step": 24320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2260905658950366e-05,
+      "loss": 0.9792,
+      "step": 24325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2218363332541944e-05,
+      "loss": 1.101,
+      "step": 24330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.217583669943322e-05,
+      "loss": 0.9779,
+      "step": 24335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.21333257711743e-05,
+      "loss": 1.0956,
+      "step": 24340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.209083055931108e-05,
+      "loss": 1.0498,
+      "step": 24345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.204835107538504e-05,
+      "loss": 0.9724,
+      "step": 24350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2005887330933457e-05,
+      "loss": 1.0702,
+      "step": 24355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1963439337489374e-05,
+      "loss": 0.9962,
+      "step": 24360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.192100710658155e-05,
+      "loss": 0.9585,
+      "step": 24365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1878590649734385e-05,
+      "loss": 1.0712,
+      "step": 24370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.183618997846811e-05,
+      "loss": 0.985,
+      "step": 24375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.179380510429852e-05,
+      "loss": 1.0523,
+      "step": 24380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.175143603873729e-05,
+      "loss": 0.985,
+      "step": 24385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.170908279329164e-05,
+      "loss": 1.0437,
+      "step": 24390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.166674537946466e-05,
+      "loss": 0.9906,
+      "step": 24395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1624423808754956e-05,
+      "loss": 1.0445,
+      "step": 24400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1582118092656994e-05,
+      "loss": 1.079,
+      "step": 24405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1539828242660826e-05,
+      "loss": 1.0078,
+      "step": 24410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.149755427025227e-05,
+      "loss": 1.1474,
+      "step": 24415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.145529618691274e-05,
+      "loss": 1.0133,
+      "step": 24420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1413054004119466e-05,
+      "loss": 1.0607,
+      "step": 24425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.137082773334521e-05,
+      "loss": 1.0482,
+      "step": 24430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1328617386058545e-05,
+      "loss": 1.0178,
+      "step": 24435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.12864229737236e-05,
+      "loss": 1.1005,
+      "step": 24440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.124424450780029e-05,
+      "loss": 1.0106,
+      "step": 24445
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.120208199974415e-05,
+      "loss": 1.0358,
+      "step": 24450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.115993546100635e-05,
+      "loss": 1.0837,
+      "step": 24455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1117804903033716e-05,
+      "loss": 1.0712,
+      "step": 24460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1075690337268804e-05,
+      "loss": 1.0587,
+      "step": 24465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.103359177514983e-05,
+      "loss": 1.0568,
+      "step": 24470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.099150922811058e-05,
+      "loss": 1.0275,
+      "step": 24475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.09494427075805e-05,
+      "loss": 1.0289,
+      "step": 24480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.090739222498475e-05,
+      "loss": 1.0327,
+      "step": 24485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.086535779174415e-05,
+      "loss": 1.0277,
+      "step": 24490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0823339419275054e-05,
+      "loss": 0.9374,
+      "step": 24495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.078133711898949e-05,
+      "loss": 1.0074,
+      "step": 24500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.073935090229518e-05,
+      "loss": 1.0913,
+      "step": 24505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0697380780595464e-05,
+      "loss": 1.1023,
+      "step": 24510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.065542676528923e-05,
+      "loss": 0.9609,
+      "step": 24515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.06134888677711e-05,
+      "loss": 0.9609,
+      "step": 24520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.057156709943122e-05,
+      "loss": 1.0638,
+      "step": 24525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0529661471655456e-05,
+      "loss": 1.0388,
+      "step": 24530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.048777199582518e-05,
+      "loss": 1.0745,
+      "step": 24535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.04458986833175e-05,
+      "loss": 0.9541,
+      "step": 24540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.040404154550498e-05,
+      "loss": 1.0863,
+      "step": 24545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.036220059375596e-05,
+      "loss": 1.0682,
+      "step": 24550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0320375839434255e-05,
+      "loss": 1.035,
+      "step": 24555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.027856729389936e-05,
+      "loss": 0.9574,
+      "step": 24560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.02367749685063e-05,
+      "loss": 0.9858,
+      "step": 24565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0194998874605786e-05,
+      "loss": 1.0886,
+      "step": 24570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0153239023544e-05,
+      "loss": 1.0576,
+      "step": 24575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.011149542666284e-05,
+      "loss": 1.0024,
+      "step": 24580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0069768095299656e-05,
+      "loss": 1.0798,
+      "step": 24585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.00280570407875e-05,
+      "loss": 1.0521,
+      "step": 24590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.998636227445499e-05,
+      "loss": 0.9686,
+      "step": 24595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9944683807626234e-05,
+      "loss": 1.0113,
+      "step": 24600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9903021651620954e-05,
+      "loss": 1.0971,
+      "step": 24605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.986137581775446e-05,
+      "loss": 1.0351,
+      "step": 24610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.981974631733768e-05,
+      "loss": 1.0056,
+      "step": 24615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.977813316167699e-05,
+      "loss": 1.0296,
+      "step": 24620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.973653636207437e-05,
+      "loss": 1.0432,
+      "step": 24625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9694955929827405e-05,
+      "loss": 1.0553,
+      "step": 24630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9653391876229215e-05,
+      "loss": 0.9968,
+      "step": 24635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.961184421256845e-05,
+      "loss": 1.0706,
+      "step": 24640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.957031295012924e-05,
+      "loss": 0.9676,
+      "step": 24645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.952879810019147e-05,
+      "loss": 1.0219,
+      "step": 24650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.948729967403035e-05,
+      "loss": 1.0704,
+      "step": 24655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9445817682916716e-05,
+      "loss": 0.959,
+      "step": 24660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.940435213811695e-05,
+      "loss": 1.0821,
+      "step": 24665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9362903050892996e-05,
+      "loss": 1.0856,
+      "step": 24670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.932147043250225e-05,
+      "loss": 1.0722,
+      "step": 24675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9280054294197634e-05,
+      "loss": 1.0352,
+      "step": 24680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.923865464722768e-05,
+      "loss": 1.0356,
+      "step": 24685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.919727150283642e-05,
+      "loss": 1.0199,
+      "step": 24690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.915590487226335e-05,
+      "loss": 1.0598,
+      "step": 24695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.911455476674346e-05,
+      "loss": 1.0317,
+      "step": 24700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.907322119750738e-05,
+      "loss": 1.0269,
+      "step": 24705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.903190417578111e-05,
+      "loss": 1.0629,
+      "step": 24710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.899060371278627e-05,
+      "loss": 1.0572,
+      "step": 24715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.894931981973985e-05,
+      "loss": 1.0297,
+      "step": 24720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.890805250785451e-05,
+      "loss": 0.9756,
+      "step": 24725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8866801788338234e-05,
+      "loss": 0.9862,
+      "step": 24730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.882556767239463e-05,
+      "loss": 1.0501,
+      "step": 24735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.878435017122274e-05,
+      "loss": 1.0846,
+      "step": 24740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.87431492960171e-05,
+      "loss": 0.9786,
+      "step": 24745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.870196505796769e-05,
+      "loss": 1.0771,
+      "step": 24750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8660797468260047e-05,
+      "loss": 0.9828,
+      "step": 24755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8619646538075184e-05,
+      "loss": 0.9843,
+      "step": 24760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8578512278589505e-05,
+      "loss": 1.0299,
+      "step": 24765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.853739470097493e-05,
+      "loss": 1.0329,
+      "step": 24770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.849629381639886e-05,
+      "loss": 1.069,
+      "step": 24775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8455209636024235e-05,
+      "loss": 0.9475,
+      "step": 24780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.84141421710093e-05,
+      "loss": 0.9991,
+      "step": 24785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.837309143250781e-05,
+      "loss": 1.0701,
+      "step": 24790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.833205743166913e-05,
+      "loss": 0.9952,
+      "step": 24795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.829104017963789e-05,
+      "loss": 1.0295,
+      "step": 24800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8250039687554195e-05,
+      "loss": 0.9371,
+      "step": 24805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.82090559665537e-05,
+      "loss": 1.0486,
+      "step": 24810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.816808902776746e-05,
+      "loss": 0.9272,
+      "step": 24815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.812713888232193e-05,
+      "loss": 1.0242,
+      "step": 24820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8086205541339e-05,
+      "loss": 1.08,
+      "step": 24825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.804528901593606e-05,
+      "loss": 1.0033,
+      "step": 24830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.800438931722594e-05,
+      "loss": 0.9959,
+      "step": 24835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7963506456316835e-05,
+      "loss": 1.0322,
+      "step": 24840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.792264044431235e-05,
+      "loss": 1.0896,
+      "step": 24845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.788179129231159e-05,
+      "loss": 1.0203,
+      "step": 24850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.784095901140908e-05,
+      "loss": 1.0591,
+      "step": 24855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7800143612694704e-05,
+      "loss": 0.9945,
+      "step": 24860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7759345107253764e-05,
+      "loss": 0.9986,
+      "step": 24865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7718563506167e-05,
+      "loss": 1.0429,
+      "step": 24870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767779882051063e-05,
+      "loss": 1.0375,
+      "step": 24875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7637051061356124e-05,
+      "loss": 0.9869,
+      "step": 24880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.759632023977049e-05,
+      "loss": 1.05,
+      "step": 24885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7555606366816034e-05,
+      "loss": 1.0418,
+      "step": 24890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7514909453550575e-05,
+      "loss": 1.0846,
+      "step": 24895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.747422951102718e-05,
+      "loss": 1.0166,
+      "step": 24900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7433566550294464e-05,
+      "loss": 1.0291,
+      "step": 24905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.739292058239632e-05,
+      "loss": 1.0282,
+      "step": 24910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7352291618372006e-05,
+      "loss": 1.0929,
+      "step": 24915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.731167966925626e-05,
+      "loss": 1.0216,
+      "step": 24920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.727108474607919e-05,
+      "loss": 1.0365,
+      "step": 24925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7230506859866186e-05,
+      "loss": 1.003,
+      "step": 24930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.718994602163803e-05,
+      "loss": 0.9694,
+      "step": 24935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7149402242411023e-05,
+      "loss": 0.9899,
+      "step": 24940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.710887553319666e-05,
+      "loss": 1.0177,
+      "step": 24945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.706836590500183e-05,
+      "loss": 1.0651,
+      "step": 24950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.702787336882883e-05,
+      "loss": 1.0599,
+      "step": 24955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.698739793567536e-05,
+      "loss": 1.0101,
+      "step": 24960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.694693961653436e-05,
+      "loss": 1.0426,
+      "step": 24965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.690649842239413e-05,
+      "loss": 1.1313,
+      "step": 24970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.686607436423842e-05,
+      "loss": 0.9959,
+      "step": 24975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.68256674530463e-05,
+      "loss": 1.0177,
+      "step": 24980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.678527769979211e-05,
+      "loss": 1.112,
+      "step": 24985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.674490511544553e-05,
+      "loss": 1.002,
+      "step": 24990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.670454971097167e-05,
+      "loss": 1.0708,
+      "step": 24995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.666421149733095e-05,
+      "loss": 1.0295,
+      "step": 25000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.662389048547906e-05,
+      "loss": 1.0685,
+      "step": 25005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.658358668636702e-05,
+      "loss": 1.0458,
+      "step": 25010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.654330011094124e-05,
+      "loss": 1.0473,
+      "step": 25015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.650303077014345e-05,
+      "loss": 0.9706,
+      "step": 25020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6462778674910613e-05,
+      "loss": 1.0385,
+      "step": 25025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.642254383617511e-05,
+      "loss": 1.1033,
+      "step": 25030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.638232626486455e-05,
+      "loss": 1.0382,
+      "step": 25035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.634212597190194e-05,
+      "loss": 1.1013,
+      "step": 25040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.630194296820548e-05,
+      "loss": 1.0228,
+      "step": 25045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.62617772646888e-05,
+      "loss": 0.9911,
+      "step": 25050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.622162887226073e-05,
+      "loss": 1.0289,
+      "step": 25055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.618149780182547e-05,
+      "loss": 1.0379,
+      "step": 25060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6141384064282444e-05,
+      "loss": 1.084,
+      "step": 25065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.610128767052647e-05,
+      "loss": 1.0367,
+      "step": 25070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.606120863144753e-05,
+      "loss": 1.0422,
+      "step": 25075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6021146957930985e-05,
+      "loss": 1.12,
+      "step": 25080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.598110266085748e-05,
+      "loss": 1.0735,
+      "step": 25085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5941075751102894e-05,
+      "loss": 0.9916,
+      "step": 25090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5901066239538364e-05,
+      "loss": 0.9859,
+      "step": 25095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.586107413703036e-05,
+      "loss": 1.0192,
+      "step": 25100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5821099454440656e-05,
+      "loss": 1.0239,
+      "step": 25105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5781142202626195e-05,
+      "loss": 1.0951,
+      "step": 25110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.574120239243921e-05,
+      "loss": 0.9859,
+      "step": 25115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.570128003472725e-05,
+      "loss": 1.0418,
+      "step": 25120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.566137514033312e-05,
+      "loss": 1.0314,
+      "step": 25125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.562148772009484e-05,
+      "loss": 0.9898,
+      "step": 25130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.558161778484566e-05,
+      "loss": 1.0646,
+      "step": 25135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.554176534541414e-05,
+      "loss": 0.9335,
+      "step": 25140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5501930412624106e-05,
+      "loss": 1.0956,
+      "step": 25145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5462112997294586e-05,
+      "loss": 1.0649,
+      "step": 25150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.542231311023979e-05,
+      "loss": 1.0705,
+      "step": 25155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.538253076226928e-05,
+      "loss": 1.0392,
+      "step": 25160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5342765964187864e-05,
+      "loss": 1.0846,
+      "step": 25165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5303018726795434e-05,
+      "loss": 0.9589,
+      "step": 25170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5263289060887284e-05,
+      "loss": 1.1038,
+      "step": 25175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5223576977253794e-05,
+      "loss": 0.978,
+      "step": 25180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.518388248668069e-05,
+      "loss": 1.0487,
+      "step": 25185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.514420559994881e-05,
+      "loss": 1.0262,
+      "step": 25190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.510454632783432e-05,
+      "loss": 0.9691,
+      "step": 25195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.506490468110848e-05,
+      "loss": 1.0164,
+      "step": 25200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.502528067053791e-05,
+      "loss": 1.044,
+      "step": 25205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4985674306884274e-05,
+      "loss": 0.9652,
+      "step": 25210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.49460856009046e-05,
+      "loss": 1.0593,
+      "step": 25215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.490651456335099e-05,
+      "loss": 1.1057,
+      "step": 25220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.486696120497083e-05,
+      "loss": 1.04,
+      "step": 25225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.482742553650672e-05,
+      "loss": 1.0138,
+      "step": 25230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4787907568696374e-05,
+      "loss": 1.0212,
+      "step": 25235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.474840731227272e-05,
+      "loss": 0.9836,
+      "step": 25240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4708924777963924e-05,
+      "loss": 1.0747,
+      "step": 25245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.466945997649333e-05,
+      "loss": 1.0274,
+      "step": 25250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.463001291857943e-05,
+      "loss": 1.0325,
+      "step": 25255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.459058361493588e-05,
+      "loss": 1.0744,
+      "step": 25260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.455117207627157e-05,
+      "loss": 0.9569,
+      "step": 25265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4511778313290576e-05,
+      "loss": 1.0063,
+      "step": 25270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.447240233669209e-05,
+      "loss": 1.0097,
+      "step": 25275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.443304415717041e-05,
+      "loss": 1.0799,
+      "step": 25280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.439370378541524e-05,
+      "loss": 1.0297,
+      "step": 25285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.435438123211123e-05,
+      "loss": 1.032,
+      "step": 25290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.43150765079382e-05,
+      "loss": 1.1305,
+      "step": 25295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4275789623571234e-05,
+      "loss": 0.9546,
+      "step": 25300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4236520589680544e-05,
+      "loss": 1.0394,
+      "step": 25305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.419726941693142e-05,
+      "loss": 1.0623,
+      "step": 25310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.415803611598435e-05,
+      "loss": 0.9996,
+      "step": 25315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4118820697495004e-05,
+      "loss": 1.0376,
+      "step": 25320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.407962317211411e-05,
+      "loss": 1.1037,
+      "step": 25325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4040443550487645e-05,
+      "loss": 0.9663,
+      "step": 25330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4001281843256594e-05,
+      "loss": 1.0691,
+      "step": 25335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.396213806105721e-05,
+      "loss": 1.1158,
+      "step": 25340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3923012214520763e-05,
+      "loss": 1.0353,
+      "step": 25345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.388390431427376e-05,
+      "loss": 0.9863,
+      "step": 25350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3844814370937706e-05,
+      "loss": 1.0788,
+      "step": 25355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.380574239512937e-05,
+      "loss": 1.0027,
+      "step": 25360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.37666883974605e-05,
+      "loss": 1.0338,
+      "step": 25365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3727652388538066e-05,
+      "loss": 1.0699,
+      "step": 25370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3688634378964154e-05,
+      "loss": 1.028,
+      "step": 25375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3649634379335896e-05,
+      "loss": 1.0691,
+      "step": 25380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.361065240024551e-05,
+      "loss": 1.1715,
+      "step": 25385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.357168845228043e-05,
+      "loss": 1.0491,
+      "step": 25390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.353274254602315e-05,
+      "loss": 1.0672,
+      "step": 25395
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3493814692051206e-05,
+      "loss": 1.0116,
+      "step": 25400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3454904900937265e-05,
+      "loss": 1.0837,
+      "step": 25405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3416013183249116e-05,
+      "loss": 1.0185,
+      "step": 25410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3377139549549666e-05,
+      "loss": 1.0415,
+      "step": 25415
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3338284010396816e-05,
+      "loss": 0.9857,
+      "step": 25420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.329944657634353e-05,
+      "loss": 1.0154,
+      "step": 25425
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.326062725793807e-05,
+      "loss": 1.0374,
+      "step": 25430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.322182606572357e-05,
+      "loss": 1.0429,
+      "step": 25435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.318304301023826e-05,
+      "loss": 1.0897,
+      "step": 25440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3144278102015524e-05,
+      "loss": 1.0368,
+      "step": 25445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.310553135158382e-05,
+      "loss": 1.017,
+      "step": 25450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.306680276946661e-05,
+      "loss": 0.9524,
+      "step": 25455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3028092366182394e-05,
+      "loss": 1.0269,
+      "step": 25460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2989400152244834e-05,
+      "loss": 1.0666,
+      "step": 25465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2950726138162635e-05,
+      "loss": 0.9549,
+      "step": 25470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2912070334439505e-05,
+      "loss": 0.9711,
+      "step": 25475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2873432751574194e-05,
+      "loss": 1.0817,
+      "step": 25480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2834813400060574e-05,
+      "loss": 0.9732,
+      "step": 25485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.279621229038756e-05,
+      "loss": 1.0078,
+      "step": 25490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.275762943303906e-05,
+      "loss": 0.9929,
+      "step": 25495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2719064838494006e-05,
+      "loss": 1.0068,
+      "step": 25500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2680518517226455e-05,
+      "loss": 0.9716,
+      "step": 25505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.26419904797055e-05,
+      "loss": 1.0621,
+      "step": 25510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.260348073639513e-05,
+      "loss": 1.1248,
+      "step": 25515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2564989297754565e-05,
+      "loss": 1.0597,
+      "step": 25520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.25265161742379e-05,
+      "loss": 1.0261,
+      "step": 25525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2488061376294277e-05,
+      "loss": 1.0731,
+      "step": 25530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2449624914367914e-05,
+      "loss": 1.0612,
+      "step": 25535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.241120679889807e-05,
+      "loss": 1.0419,
+      "step": 25540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.237280704031895e-05,
+      "loss": 1.0817,
+      "step": 25545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.233442564905976e-05,
+      "loss": 1.0345,
+      "step": 25550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.22960626355448e-05,
+      "loss": 1.0572,
+      "step": 25555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.225771801019334e-05,
+      "loss": 1.0847,
+      "step": 25560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.221939178341965e-05,
+      "loss": 1.0792,
+      "step": 25565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.218108396563293e-05,
+      "loss": 1.0027,
+      "step": 25570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.214279456723758e-05,
+      "loss": 1.012,
+      "step": 25575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.210452359863282e-05,
+      "loss": 1.0112,
+      "step": 25580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2066271070212874e-05,
+      "loss": 1.1183,
+      "step": 25585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.202803699236703e-05,
+      "loss": 1.0012,
+      "step": 25590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.198982137547959e-05,
+      "loss": 1.0009,
+      "step": 25595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1951624229929736e-05,
+      "loss": 1.0293,
+      "step": 25600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.191344556609165e-05,
+      "loss": 1.0824,
+      "step": 25605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.187528539433458e-05,
+      "loss": 1.0356,
+      "step": 25610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.183714372502272e-05,
+      "loss": 1.0295,
+      "step": 25615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.179902056851519e-05,
+      "loss": 1.0563,
+      "step": 25620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.176091593516608e-05,
+      "loss": 1.0399,
+      "step": 25625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.172282983532453e-05,
+      "loss": 1.0585,
+      "step": 25630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.16847622793346e-05,
+      "loss": 1.0386,
+      "step": 25635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.16467132775353e-05,
+      "loss": 1.0185,
+      "step": 25640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.160868284026057e-05,
+      "loss": 0.9609,
+      "step": 25645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.157067097783939e-05,
+      "loss": 1.089,
+      "step": 25650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.153267770059568e-05,
+      "loss": 1.0197,
+      "step": 25655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.149470301884823e-05,
+      "loss": 1.0694,
+      "step": 25660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1456746942910896e-05,
+      "loss": 1.0853,
+      "step": 25665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.141880948309236e-05,
+      "loss": 1.0008,
+      "step": 25670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.138089064969637e-05,
+      "loss": 1.0344,
+      "step": 25675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.134299045302149e-05,
+      "loss": 0.9639,
+      "step": 25680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.130510890336136e-05,
+      "loss": 0.9792,
+      "step": 25685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1267246011004405e-05,
+      "loss": 1.1146,
+      "step": 25690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1229401786234134e-05,
+      "loss": 1.1178,
+      "step": 25695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.119157623932884e-05,
+      "loss": 1.1204,
+      "step": 25700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.11537693805619e-05,
+      "loss": 1.0428,
+      "step": 25705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.111598122020144e-05,
+      "loss": 0.9447,
+      "step": 25710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.107821176851065e-05,
+      "loss": 1.0541,
+      "step": 25715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1040461035747605e-05,
+      "loss": 1.0856,
+      "step": 25720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1002729032165256e-05,
+      "loss": 1.0516,
+      "step": 25725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0965015768011455e-05,
+      "loss": 1.0017,
+      "step": 25730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.092732125352902e-05,
+      "loss": 1.106,
+      "step": 25735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.08896454989557e-05,
+      "loss": 1.1169,
+      "step": 25740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.085198851452407e-05,
+      "loss": 1.079,
+      "step": 25745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0814350310461615e-05,
+      "loss": 1.1127,
+      "step": 25750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0776730896990756e-05,
+      "loss": 1.0324,
+      "step": 25755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0739130284328856e-05,
+      "loss": 1.0837,
+      "step": 25760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.070154848268807e-05,
+      "loss": 1.0753,
+      "step": 25765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.066398550227547e-05,
+      "loss": 1.0661,
+      "step": 25770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.062644135329306e-05,
+      "loss": 1.0329,
+      "step": 25775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.058891604593773e-05,
+      "loss": 1.0285,
+      "step": 25780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0551409590401214e-05,
+      "loss": 1.0029,
+      "step": 25785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0513921996870087e-05,
+      "loss": 0.97,
+      "step": 25790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0476453275525906e-05,
+      "loss": 1.0401,
+      "step": 25795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0439003436545076e-05,
+      "loss": 1.0599,
+      "step": 25800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0401572490098783e-05,
+      "loss": 1.0251,
+      "step": 25805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0364160446353217e-05,
+      "loss": 0.9775,
+      "step": 25810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0326767315469285e-05,
+      "loss": 1.0372,
+      "step": 25815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0289393107602914e-05,
+      "loss": 1.0068,
+      "step": 25820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0252037832904745e-05,
+      "loss": 0.9607,
+      "step": 25825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0214701501520415e-05,
+      "loss": 1.0407,
+      "step": 25830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.017738412359027e-05,
+      "loss": 1.0031,
+      "step": 25835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0140085709249667e-05,
+      "loss": 1.0148,
+      "step": 25840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0102806268628658e-05,
+      "loss": 0.9733,
+      "step": 25845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.006554581185228e-05,
+      "loss": 1.0016,
+      "step": 25850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.002830434904028e-05,
+      "loss": 0.9631,
+      "step": 25855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.999108189030735e-05,
+      "loss": 1.1022,
+      "step": 25860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9953878445763016e-05,
+      "loss": 1.0955,
+      "step": 25865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9916694025511594e-05,
+      "loss": 1.0236,
+      "step": 25870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9879528639652188e-05,
+      "loss": 0.9832,
+      "step": 25875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.984238229827885e-05,
+      "loss": 0.9476,
+      "step": 25880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9805255011480415e-05,
+      "loss": 1.0375,
+      "step": 25885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9768146789340513e-05,
+      "loss": 1.0643,
+      "step": 25890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9731057641937566e-05,
+      "loss": 0.9928,
+      "step": 25895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9693987579344917e-05,
+      "loss": 0.9597,
+      "step": 25900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9656936611630683e-05,
+      "loss": 1.0707,
+      "step": 25905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9619904748857765e-05,
+      "loss": 1.0085,
+      "step": 25910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9582892001083805e-05,
+      "loss": 1.0242,
+      "step": 25915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9545898378361502e-05,
+      "loss": 1.0192,
+      "step": 25920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9508923890738127e-05,
+      "loss": 0.9726,
+      "step": 25925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.947196854825581e-05,
+      "loss": 1.0651,
+      "step": 25930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9435032360951486e-05,
+      "loss": 1.0181,
+      "step": 25935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9398115338856923e-05,
+      "loss": 1.0653,
+      "step": 25940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9361217491998694e-05,
+      "loss": 1.0232,
+      "step": 25945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9324338830398068e-05,
+      "loss": 1.0497,
+      "step": 25950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9287479364071237e-05,
+      "loss": 1.0093,
+      "step": 25955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.9250639103029033e-05,
+      "loss": 1.0346,
+      "step": 25960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.921381805727722e-05,
+      "loss": 0.9252,
+      "step": 25965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.917701623681621e-05,
+      "loss": 0.9798,
+      "step": 25970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.9140233651641314e-05,
+      "loss": 1.1014,
+      "step": 25975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.9103470311742485e-05,
+      "loss": 1.0062,
+      "step": 25980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.9066726227104612e-05,
+      "loss": 1.087,
+      "step": 25985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.9030001407707174e-05,
+      "loss": 0.9848,
+      "step": 25990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8993295863524593e-05,
+      "loss": 1.0126,
+      "step": 25995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8956609604525908e-05,
+      "loss": 1.0782,
+      "step": 26000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8919942640675002e-05,
+      "loss": 1.0928,
+      "step": 26005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8883294981930542e-05,
+      "loss": 1.0186,
+      "step": 26010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8846666638245878e-05,
+      "loss": 0.9454,
+      "step": 26015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.881005761956911e-05,
+      "loss": 0.9777,
+      "step": 26020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8773467935843167e-05,
+      "loss": 1.04,
+      "step": 26025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.873689759700571e-05,
+      "loss": 1.0774,
+      "step": 26030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8700346612989083e-05,
+      "loss": 0.9891,
+      "step": 26035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8663814993720385e-05,
+      "loss": 1.0537,
+      "step": 26040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.862730274912152e-05,
+      "loss": 1.1639,
+      "step": 26045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8590809889109104e-05,
+      "loss": 1.0412,
+      "step": 26050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8554336423594476e-05,
+      "loss": 1.0299,
+      "step": 26055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8517882362483617e-05,
+      "loss": 1.0334,
+      "step": 26060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8481447715677466e-05,
+      "loss": 1.0254,
+      "step": 26065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.844503249307149e-05,
+      "loss": 0.9921,
+      "step": 26070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8408636704555923e-05,
+      "loss": 1.0319,
+      "step": 26075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.837226036001571e-05,
+      "loss": 0.9438,
+      "step": 26080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.833590346933065e-05,
+      "loss": 1.0092,
+      "step": 26085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.829956604237508e-05,
+      "loss": 0.9951,
+      "step": 26090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8263248089018113e-05,
+      "loss": 1.0773,
+      "step": 26095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.822694961912361e-05,
+      "loss": 1.0475,
+      "step": 26100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8190670642550143e-05,
+      "loss": 0.9991,
+      "step": 26105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.815441116915093e-05,
+      "loss": 1.1085,
+      "step": 26110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.81181712087739e-05,
+      "loss": 1.054,
+      "step": 26115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8081950771261723e-05,
+      "loss": 1.0013,
+      "step": 26120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8045749866451776e-05,
+      "loss": 1.0207,
+      "step": 26125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8009568504176097e-05,
+      "loss": 0.9987,
+      "step": 26130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7973406694261362e-05,
+      "loss": 1.0233,
+      "step": 26135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7937264446529087e-05,
+      "loss": 1.1119,
+      "step": 26140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7901141770795315e-05,
+      "loss": 0.9424,
+      "step": 26145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7865038676870868e-05,
+      "loss": 1.0441,
+      "step": 26150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7828955174561257e-05,
+      "loss": 1.0475,
+      "step": 26155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7792891273666623e-05,
+      "loss": 1.121,
+      "step": 26160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7756846983981753e-05,
+      "loss": 1.0811,
+      "step": 26165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.77208223152962e-05,
+      "loss": 1.0469,
+      "step": 26170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7684817277394172e-05,
+      "loss": 1.0711,
+      "step": 26175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7648831880054483e-05,
+      "loss": 1.0412,
+      "step": 26180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.761286613305062e-05,
+      "loss": 1.0333,
+      "step": 26185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7576920046150778e-05,
+      "loss": 1.0395,
+      "step": 26190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7540993629117852e-05,
+      "loss": 0.9771,
+      "step": 26195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7505086891709277e-05,
+      "loss": 1.0648,
+      "step": 26200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.746919984367715e-05,
+      "loss": 1.0203,
+      "step": 26205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7433332494768415e-05,
+      "loss": 1.0132,
+      "step": 26210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7397484854724433e-05,
+      "loss": 1.0347,
+      "step": 26215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.736165693328133e-05,
+      "loss": 1.0185,
+      "step": 26220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.732584874016977e-05,
+      "loss": 1.0573,
+      "step": 26225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.729006028511527e-05,
+      "loss": 0.9561,
+      "step": 26230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.72542915778378e-05,
+      "loss": 1.0253,
+      "step": 26235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.721854262805198e-05,
+      "loss": 0.9673,
+      "step": 26240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7182813445467138e-05,
+      "loss": 0.9305,
+      "step": 26245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.714710403978723e-05,
+      "loss": 0.9437,
+      "step": 26250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.711141442071078e-05,
+      "loss": 1.0434,
+      "step": 26255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.707574459793095e-05,
+      "loss": 0.9906,
+      "step": 26260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7040094581135565e-05,
+      "loss": 1.084,
+      "step": 26265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.700446438000709e-05,
+      "loss": 0.986,
+      "step": 26270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.696885400422252e-05,
+      "loss": 1.0865,
+      "step": 26275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6933263463453505e-05,
+      "loss": 0.9887,
+      "step": 26280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6897692767366334e-05,
+      "loss": 1.0351,
+      "step": 26285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.686214192562193e-05,
+      "loss": 1.0374,
+      "step": 26290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.682661094787571e-05,
+      "loss": 1.113,
+      "step": 26295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6791099843777846e-05,
+      "loss": 1.1425,
+      "step": 26300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.675560862297296e-05,
+      "loss": 0.9771,
+      "step": 26305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.672013729510041e-05,
+      "loss": 1.0047,
+      "step": 26310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6684685869794035e-05,
+      "loss": 0.9878,
+      "step": 26315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6649254356682386e-05,
+      "loss": 1.0397,
+      "step": 26320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6613842765388486e-05,
+      "loss": 1.0402,
+      "step": 26325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6578451105530055e-05,
+      "loss": 1.0617,
+      "step": 26330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6543079386719295e-05,
+      "loss": 1.005,
+      "step": 26335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6507727618563117e-05,
+      "loss": 1.0364,
+      "step": 26340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6472395810662888e-05,
+      "loss": 1.053,
+      "step": 26345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6437083972614572e-05,
+      "loss": 1.0284,
+      "step": 26350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.640179211400886e-05,
+      "loss": 1.091,
+      "step": 26355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6366520244430836e-05,
+      "loss": 1.0511,
+      "step": 26360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6331268373460207e-05,
+      "loss": 1.0229,
+      "step": 26365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6296036510671284e-05,
+      "loss": 1.0336,
+      "step": 26370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6260824665632944e-05,
+      "loss": 1.0379,
+      "step": 26375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.62256328479086e-05,
+      "loss": 1.034,
+      "step": 26380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6190461067056172e-05,
+      "loss": 1.0471,
+      "step": 26385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6155309332628254e-05,
+      "loss": 0.9789,
+      "step": 26390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6120177654171963e-05,
+      "loss": 1.0769,
+      "step": 26395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.608506604122891e-05,
+      "loss": 1.0738,
+      "step": 26400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6049974503335273e-05,
+      "loss": 1.0762,
+      "step": 26405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.601490305002181e-05,
+      "loss": 1.0625,
+      "step": 26410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.597985169081385e-05,
+      "loss": 0.9961,
+      "step": 26415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5944820435231198e-05,
+      "loss": 0.922,
+      "step": 26420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.590980929278819e-05,
+      "loss": 0.9967,
+      "step": 26425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5874818272993773e-05,
+      "loss": 1.0476,
+      "step": 26430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5839847385351413e-05,
+      "loss": 1.0677,
+      "step": 26435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5804896639359032e-05,
+      "loss": 1.0986,
+      "step": 26440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.57699660445092e-05,
+      "loss": 1.0052,
+      "step": 26445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5735055610288872e-05,
+      "loss": 1.0569,
+      "step": 26450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5700165346179685e-05,
+      "loss": 0.9948,
+      "step": 26455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5665295261657652e-05,
+      "loss": 0.9262,
+      "step": 26460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.563044536619342e-05,
+      "loss": 1.0423,
+      "step": 26465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.559561566925206e-05,
+      "loss": 1.081,
+      "step": 26470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.556080618029326e-05,
+      "loss": 1.0541,
+      "step": 26475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.55260169087711e-05,
+      "loss": 1.0365,
+      "step": 26480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.549124786413428e-05,
+      "loss": 1.0402,
+      "step": 26485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5456499055825912e-05,
+      "loss": 0.9999,
+      "step": 26490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.542177049328367e-05,
+      "loss": 1.0206,
+      "step": 26495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5387062185939757e-05,
+      "loss": 1.0495,
+      "step": 26500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5352374143220803e-05,
+      "loss": 1.0061,
+      "step": 26505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5317706374547933e-05,
+      "loss": 1.052,
+      "step": 26510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.528305888933683e-05,
+      "loss": 1.08,
+      "step": 26515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5248431696997654e-05,
+      "loss": 1.0569,
+      "step": 26520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5213824806935015e-05,
+      "loss": 1.0738,
+      "step": 26525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5179238228547997e-05,
+      "loss": 1.0215,
+      "step": 26530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5144671971230228e-05,
+      "loss": 0.9556,
+      "step": 26535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.511012604436982e-05,
+      "loss": 1.0332,
+      "step": 26540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.50756004573493e-05,
+      "loss": 1.0865,
+      "step": 26545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5041095219545674e-05,
+      "loss": 1.0271,
+      "step": 26550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5006610340330472e-05,
+      "loss": 1.0613,
+      "step": 26555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4972145829069714e-05,
+      "loss": 0.9664,
+      "step": 26560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4937701695123815e-05,
+      "loss": 0.9145,
+      "step": 26565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.490327794784765e-05,
+      "loss": 1.0345,
+      "step": 26570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4868874596590618e-05,
+      "loss": 1.0098,
+      "step": 26575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.48344916506966e-05,
+      "loss": 0.9982,
+      "step": 26580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4800129119503822e-05,
+      "loss": 1.0655,
+      "step": 26585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.47657870123451e-05,
+      "loss": 1.0143,
+      "step": 26590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4731465338547556e-05,
+      "loss": 1.046,
+      "step": 26595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4697164107432912e-05,
+      "loss": 0.9702,
+      "step": 26600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4662883328317222e-05,
+      "loss": 1.0466,
+      "step": 26605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4628623010511075e-05,
+      "loss": 1.1417,
+      "step": 26610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.459438316331941e-05,
+      "loss": 1.0417,
+      "step": 26615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.45601637960417e-05,
+      "loss": 0.9875,
+      "step": 26620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4525964917971767e-05,
+      "loss": 0.9851,
+      "step": 26625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4491786538397967e-05,
+      "loss": 1.0625,
+      "step": 26630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.445762866660297e-05,
+      "loss": 0.9772,
+      "step": 26635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4423491311863965e-05,
+      "loss": 1.003,
+      "step": 26640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.438937448345259e-05,
+      "loss": 1.0837,
+      "step": 26645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.435527819063482e-05,
+      "loss": 1.0253,
+      "step": 26650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4321202442671066e-05,
+      "loss": 0.9579,
+      "step": 26655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.428714724881622e-05,
+      "loss": 1.1074,
+      "step": 26660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4253112618319584e-05,
+      "loss": 1.0136,
+      "step": 26665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4219098560424826e-05,
+      "loss": 1.0544,
+      "step": 26670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4185105084370018e-05,
+      "loss": 1.0551,
+      "step": 26675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4151132199387704e-05,
+      "loss": 1.051,
+      "step": 26680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4117179914704835e-05,
+      "loss": 1.0234,
+      "step": 26685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.408324823954272e-05,
+      "loss": 1.137,
+      "step": 26690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.404933718311704e-05,
+      "loss": 1.0517,
+      "step": 26695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4015446754637973e-05,
+      "loss": 1.0093,
+      "step": 26700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3981576963310048e-05,
+      "loss": 0.9862,
+      "step": 26705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.394772781833219e-05,
+      "loss": 0.9466,
+      "step": 26710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3913899328897647e-05,
+      "loss": 0.9921,
+      "step": 26715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3880091504194225e-05,
+      "loss": 1.0177,
+      "step": 26720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.384630435340397e-05,
+      "loss": 1.0909,
+      "step": 26725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3812537885703336e-05,
+      "loss": 1.0527,
+      "step": 26730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3778792110263203e-05,
+      "loss": 1.0238,
+      "step": 26735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3745067036248836e-05,
+      "loss": 1.0966,
+      "step": 26740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3711362672819826e-05,
+      "loss": 1.0737,
+      "step": 26745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3677679029130152e-05,
+      "loss": 1.0462,
+      "step": 26750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3644016114328214e-05,
+      "loss": 0.9681,
+      "step": 26755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3610373937556717e-05,
+      "loss": 1.056,
+      "step": 26760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3576752507952793e-05,
+      "loss": 1.0341,
+      "step": 26765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.354315183464787e-05,
+      "loss": 1.0972,
+      "step": 26770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3509571926767825e-05,
+      "loss": 0.9849,
+      "step": 26775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.34760127934328e-05,
+      "loss": 1.0213,
+      "step": 26780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3442474443757366e-05,
+      "loss": 1.0322,
+      "step": 26785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3408956886850474e-05,
+      "loss": 1.0333,
+      "step": 26790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.337546013181533e-05,
+      "loss": 1.0361,
+      "step": 26795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3341984187749512e-05,
+      "loss": 0.9704,
+      "step": 26800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.330852906374502e-05,
+      "loss": 1.0536,
+      "step": 26805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.327509476888817e-05,
+      "loss": 1.05,
+      "step": 26810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.324168131225959e-05,
+      "loss": 1.0976,
+      "step": 26815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.320828870293421e-05,
+      "loss": 1.0781,
+      "step": 26820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3174916949981407e-05,
+      "loss": 1.0256,
+      "step": 26825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3141566062464847e-05,
+      "loss": 0.9687,
+      "step": 26830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.31082360494425e-05,
+      "loss": 0.9723,
+      "step": 26835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3074926919966654e-05,
+      "loss": 1.0468,
+      "step": 26840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3041638683083988e-05,
+      "loss": 1.0212,
+      "step": 26845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3008371347835512e-05,
+      "loss": 1.066,
+      "step": 26850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2975124923256485e-05,
+      "loss": 1.0378,
+      "step": 26855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2941899418376466e-05,
+      "loss": 1.0009,
+      "step": 26860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2908694842219513e-05,
+      "loss": 1.028,
+      "step": 26865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2875511203803813e-05,
+      "loss": 1.0512,
+      "step": 26870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2842348512141908e-05,
+      "loss": 1.0158,
+      "step": 26875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.280920677624071e-05,
+      "loss": 1.0609,
+      "step": 26880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2776086005101417e-05,
+      "loss": 1.0954,
+      "step": 26885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2742986207719506e-05,
+      "loss": 1.0528,
+      "step": 26890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2709907393084738e-05,
+      "loss": 1.0503,
+      "step": 26895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2676849570181225e-05,
+      "loss": 1.0591,
+      "step": 26900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2643812747987414e-05,
+      "loss": 1.0174,
+      "step": 26905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2610796935475943e-05,
+      "loss": 1.0398,
+      "step": 26910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.257780214161378e-05,
+      "loss": 1.0512,
+      "step": 26915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.254482837536224e-05,
+      "loss": 0.9905,
+      "step": 26920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2511875645676894e-05,
+      "loss": 1.0876,
+      "step": 26925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.247894396150755e-05,
+      "loss": 1.0574,
+      "step": 26930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.244603333179839e-05,
+      "loss": 1.0374,
+      "step": 26935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2413143765487786e-05,
+      "loss": 0.9323,
+      "step": 26940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2380275271508488e-05,
+      "loss": 1.1288,
+      "step": 26945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2347427858787416e-05,
+      "loss": 1.0254,
+      "step": 26950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.231460153624586e-05,
+      "loss": 0.9757,
+      "step": 26955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.228179631279933e-05,
+      "loss": 0.9398,
+      "step": 26960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.224901219735758e-05,
+      "loss": 1.0836,
+      "step": 26965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.221624919882469e-05,
+      "loss": 1.017,
+      "step": 26970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2183507326099005e-05,
+      "loss": 0.9517,
+      "step": 26975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.21507865880731e-05,
+      "loss": 1.1093,
+      "step": 26980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2118086993633746e-05,
+      "loss": 0.9549,
+      "step": 26985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2085408551662156e-05,
+      "loss": 0.9998,
+      "step": 26990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2052751271033632e-05,
+      "loss": 1.1442,
+      "step": 26995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2020115160617783e-05,
+      "loss": 0.9667,
+      "step": 27000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1987500229278425e-05,
+      "loss": 1.0595,
+      "step": 27005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.195490648587375e-05,
+      "loss": 1.0077,
+      "step": 27010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1922333939256067e-05,
+      "loss": 1.0227,
+      "step": 27015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1889782598271957e-05,
+      "loss": 1.0444,
+      "step": 27020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1857252471762258e-05,
+      "loss": 1.0428,
+      "step": 27025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.182474356856209e-05,
+      "loss": 0.9872,
+      "step": 27030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1792255897500734e-05,
+      "loss": 1.0965,
+      "step": 27035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1759789467401692e-05,
+      "loss": 1.0762,
+      "step": 27040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1727344287082774e-05,
+      "loss": 1.001,
+      "step": 27045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.169492036535602e-05,
+      "loss": 0.9293,
+      "step": 27050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.166251771102761e-05,
+      "loss": 0.9666,
+      "step": 27055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.163013633289799e-05,
+      "loss": 1.0457,
+      "step": 27060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1597776239761847e-05,
+      "loss": 1.0084,
+      "step": 27065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.156543744040811e-05,
+      "loss": 0.9755,
+      "step": 27070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1533119943619827e-05,
+      "loss": 1.0413,
+      "step": 27075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1500823758174392e-05,
+      "loss": 1.0121,
+      "step": 27080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1468548892843265e-05,
+      "loss": 1.0324,
+      "step": 27085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1436295356392267e-05,
+      "loss": 1.0253,
+      "step": 27090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1404063157581277e-05,
+      "loss": 0.971,
+      "step": 27095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.137185230516453e-05,
+      "loss": 0.9728,
+      "step": 27100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1339662807890302e-05,
+      "loss": 1.0464,
+      "step": 27105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.130749467450124e-05,
+      "loss": 0.9913,
+      "step": 27110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1275347913734022e-05,
+      "loss": 1.0261,
+      "step": 27115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1243222534319673e-05,
+      "loss": 1.0351,
+      "step": 27120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.121111854498328e-05,
+      "loss": 1.0623,
+      "step": 27125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1179035954444193e-05,
+      "loss": 1.0173,
+      "step": 27130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1146974771415973e-05,
+      "loss": 0.9972,
+      "step": 27135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1114935004606307e-05,
+      "loss": 0.9912,
+      "step": 27140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1082916662717055e-05,
+      "loss": 0.9162,
+      "step": 27145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1050919754444332e-05,
+      "loss": 0.9906,
+      "step": 27150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.101894428847839e-05,
+      "loss": 1.0139,
+      "step": 27155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.098699027350367e-05,
+      "loss": 1.0463,
+      "step": 27160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0955057718198722e-05,
+      "loss": 0.9878,
+      "step": 27165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0923146631236358e-05,
+      "loss": 1.0481,
+      "step": 27170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.089125702128355e-05,
+      "loss": 1.0226,
+      "step": 27175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0859388897001375e-05,
+      "loss": 1.1479,
+      "step": 27180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0827542267045085e-05,
+      "loss": 1.0502,
+      "step": 27185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0795717140064163e-05,
+      "loss": 0.9748,
+      "step": 27190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0763913524702205e-05,
+      "loss": 1.0257,
+      "step": 27195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0732131429596956e-05,
+      "loss": 1.061,
+      "step": 27200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0700370863380304e-05,
+      "loss": 1.0184,
+      "step": 27205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.066863183467833e-05,
+      "loss": 1.059,
+      "step": 27210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.063691435211128e-05,
+      "loss": 0.9855,
+      "step": 27215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.060521842429347e-05,
+      "loss": 0.9978,
+      "step": 27220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.057354405983345e-05,
+      "loss": 1.0309,
+      "step": 27225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0541891267333813e-05,
+      "loss": 1.0717,
+      "step": 27230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0510260055391418e-05,
+      "loss": 1.0358,
+      "step": 27235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0478650432597145e-05,
+      "loss": 1.0544,
+      "step": 27240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.044706240753611e-05,
+      "loss": 1.0294,
+      "step": 27245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.041549598878746e-05,
+      "loss": 0.9081,
+      "step": 27250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0383951184924578e-05,
+      "loss": 0.9869,
+      "step": 27255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.035242800451489e-05,
+      "loss": 1.0286,
+      "step": 27260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0320926456120027e-05,
+      "loss": 1.0271,
+      "step": 27265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0289446548295643e-05,
+      "loss": 0.9797,
+      "step": 27270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0257988289591622e-05,
+      "loss": 1.0186,
+      "step": 27275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0226551688551954e-05,
+      "loss": 1.0778,
+      "step": 27280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.019513675371466e-05,
+      "loss": 1.0309,
+      "step": 27285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.016374349361192e-05,
+      "loss": 1.124,
+      "step": 27290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0132371916770076e-05,
+      "loss": 1.0279,
+      "step": 27295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.010102203170955e-05,
+      "loss": 1.0553,
+      "step": 27300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.006969384694485e-05,
+      "loss": 1.0364,
+      "step": 27305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.003838737098458e-05,
+      "loss": 1.0,
+      "step": 27310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0007102612331497e-05,
+      "loss": 1.0449,
+      "step": 27315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.997583957948247e-05,
+      "loss": 1.0565,
+      "step": 27320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9944598280928408e-05,
+      "loss": 0.9891,
+      "step": 27325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.991337872515431e-05,
+      "loss": 0.9986,
+      "step": 27330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9882180920639327e-05,
+      "loss": 1.0133,
+      "step": 27335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9851004875856717e-05,
+      "loss": 1.0193,
+      "step": 27340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.981985059927376e-05,
+      "loss": 1.0163,
+      "step": 27345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.978871809935179e-05,
+      "loss": 0.9787,
+      "step": 27350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9757607384546407e-05,
+      "loss": 1.0372,
+      "step": 27355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972651846330713e-05,
+      "loss": 0.9938,
+      "step": 27360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9695451344077564e-05,
+      "loss": 1.111,
+      "step": 27365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.966440603529549e-05,
+      "loss": 0.986,
+      "step": 27370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9633382545392665e-05,
+      "loss": 0.9983,
+      "step": 27375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9602380882795023e-05,
+      "loss": 0.9977,
+      "step": 27380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9571401055922446e-05,
+      "loss": 1.062,
+      "step": 27385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9540443073189008e-05,
+      "loss": 1.0474,
+      "step": 27390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9509506943002752e-05,
+      "loss": 1.0369,
+      "step": 27395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.947859267376586e-05,
+      "loss": 1.0907,
+      "step": 27400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9447700273874514e-05,
+      "loss": 0.9748,
+      "step": 27405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9416829751719023e-05,
+      "loss": 1.1434,
+      "step": 27410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9385981115683672e-05,
+      "loss": 1.0793,
+      "step": 27415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.935515437414688e-05,
+      "loss": 0.9201,
+      "step": 27420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.932434953548109e-05,
+      "loss": 0.9769,
+      "step": 27425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9293566608052794e-05,
+      "loss": 1.0236,
+      "step": 27430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.92628056002225e-05,
+      "loss": 0.9949,
+      "step": 27435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9232066520344817e-05,
+      "loss": 1.0628,
+      "step": 27440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9201349376768396e-05,
+      "loss": 0.9804,
+      "step": 27445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.917065417783589e-05,
+      "loss": 1.0584,
+      "step": 27450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9139980931883993e-05,
+      "loss": 1.0286,
+      "step": 27455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.910932964724347e-05,
+      "loss": 1.0801,
+      "step": 27460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9078700332239154e-05,
+      "loss": 1.0237,
+      "step": 27465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9048092995189814e-05,
+      "loss": 1.0433,
+      "step": 27470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9017507644408285e-05,
+      "loss": 1.0249,
+      "step": 27475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.898694428820148e-05,
+      "loss": 1.0478,
+      "step": 27480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.895640293487032e-05,
+      "loss": 1.0579,
+      "step": 27485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8925883592709714e-05,
+      "loss": 1.0685,
+      "step": 27490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8895386270008564e-05,
+      "loss": 0.9896,
+      "step": 27495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8864910975049933e-05,
+      "loss": 1.0321,
+      "step": 27500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8834457716110778e-05,
+      "loss": 0.9396,
+      "step": 27505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8804026501462047e-05,
+      "loss": 1.0297,
+      "step": 27510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.877361733936881e-05,
+      "loss": 1.0638,
+      "step": 27515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8743230238090115e-05,
+      "loss": 1.1102,
+      "step": 27520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.871286520587895e-05,
+      "loss": 1.0131,
+      "step": 27525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.868252225098236e-05,
+      "loss": 0.9769,
+      "step": 27530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8652201381641398e-05,
+      "loss": 1.0752,
+      "step": 27535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8621902606091136e-05,
+      "loss": 1.0212,
+      "step": 27540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8591625932560607e-05,
+      "loss": 0.9927,
+      "step": 27545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8561371369272818e-05,
+      "loss": 0.9938,
+      "step": 27550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8531138924444837e-05,
+      "loss": 1.0038,
+      "step": 27555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8500928606287726e-05,
+      "loss": 1.0646,
+      "step": 27560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.847074042300644e-05,
+      "loss": 1.0039,
+      "step": 27565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8440574382800057e-05,
+      "loss": 1.0066,
+      "step": 27570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.841043049386154e-05,
+      "loss": 1.0108,
+      "step": 27575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8380308764377842e-05,
+      "loss": 1.0333,
+      "step": 27580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8350209202529946e-05,
+      "loss": 1.0425,
+      "step": 27585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8320131816492824e-05,
+      "loss": 1.0079,
+      "step": 27590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.829007661443538e-05,
+      "loss": 1.0192,
+      "step": 27595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.826004360452046e-05,
+      "loss": 1.0579,
+      "step": 27600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.823003279490496e-05,
+      "loss": 1.0532,
+      "step": 27605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.820004419373975e-05,
+      "loss": 0.9751,
+      "step": 27610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8170077809169593e-05,
+      "loss": 0.9586,
+      "step": 27615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8140133649333258e-05,
+      "loss": 1.0643,
+      "step": 27620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.811021172236348e-05,
+      "loss": 1.0345,
+      "step": 27625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8080312036386994e-05,
+      "loss": 1.0339,
+      "step": 27630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8050434599524423e-05,
+      "loss": 0.9934,
+      "step": 27635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8020579419890326e-05,
+      "loss": 1.0862,
+      "step": 27640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7990746505593393e-05,
+      "loss": 1.0523,
+      "step": 27645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7960935864736063e-05,
+      "loss": 1.0975,
+      "step": 27650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7931147505414792e-05,
+      "loss": 1.0015,
+      "step": 27655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7901381435720032e-05,
+      "loss": 1.0228,
+      "step": 27660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7871637663736174e-05,
+      "loss": 0.9851,
+      "step": 27665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7841916197541497e-05,
+      "loss": 1.0008,
+      "step": 27670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7812217045208224e-05,
+      "loss": 1.0621,
+      "step": 27675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.778254021480257e-05,
+      "loss": 1.0256,
+      "step": 27680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7752885714384694e-05,
+      "loss": 1.0626,
+      "step": 27685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7723253552008632e-05,
+      "loss": 1.0999,
+      "step": 27690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7693643735722352e-05,
+      "loss": 0.9905,
+      "step": 27695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.766405627356781e-05,
+      "loss": 1.0061,
+      "step": 27700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.763449117358089e-05,
+      "loss": 1.0145,
+      "step": 27705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7604948443791325e-05,
+      "loss": 0.9774,
+      "step": 27710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7575428092222868e-05,
+      "loss": 1.0192,
+      "step": 27715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7545930126893084e-05,
+      "loss": 1.0671,
+      "step": 27720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.75164545558136e-05,
+      "loss": 0.9947,
+      "step": 27725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.748700138698982e-05,
+      "loss": 1.023,
+      "step": 27730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7457570628421172e-05,
+      "loss": 1.0785,
+      "step": 27735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7428162288100903e-05,
+      "loss": 1.0384,
+      "step": 27740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.739877637401627e-05,
+      "loss": 0.9922,
+      "step": 27745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.736941289414834e-05,
+      "loss": 0.9896,
+      "step": 27750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7340071856472195e-05,
+      "loss": 1.1099,
+      "step": 27755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7310753268956693e-05,
+      "loss": 1.022,
+      "step": 27760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7281457139564737e-05,
+      "loss": 1.0709,
+      "step": 27765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7252183476253003e-05,
+      "loss": 1.1128,
+      "step": 27770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.722293228697216e-05,
+      "loss": 1.0421,
+      "step": 27775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.719370357966671e-05,
+      "loss": 1.0009,
+      "step": 27780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7164497362275022e-05,
+      "loss": 1.0556,
+      "step": 27785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7135313642729523e-05,
+      "loss": 1.0064,
+      "step": 27790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7106152428956347e-05,
+      "loss": 1.0219,
+      "step": 27795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7077013728875556e-05,
+      "loss": 1.07,
+      "step": 27800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.704789755040116e-05,
+      "loss": 0.9593,
+      "step": 27805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701880390144104e-05,
+      "loss": 1.0281,
+      "step": 27810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6989732789896907e-05,
+      "loss": 1.0145,
+      "step": 27815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6960684223664337e-05,
+      "loss": 1.0423,
+      "step": 27820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6931658210632873e-05,
+      "loss": 1.0445,
+      "step": 27825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6902654758685897e-05,
+      "loss": 1.0456,
+      "step": 27830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6873673875700625e-05,
+      "loss": 1.0002,
+      "step": 27835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.684471556954814e-05,
+      "loss": 0.9514,
+      "step": 27840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.681577984809346e-05,
+      "loss": 1.0272,
+      "step": 27845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.678686671919544e-05,
+      "loss": 0.9692,
+      "step": 27850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6757976190706746e-05,
+      "loss": 1.0692,
+      "step": 27855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6729108270473992e-05,
+      "loss": 1.0373,
+      "step": 27860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6700262966337578e-05,
+      "loss": 1.0282,
+      "step": 27865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6671440286131813e-05,
+      "loss": 1.0402,
+      "step": 27870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6642640237684816e-05,
+      "loss": 1.071,
+      "step": 27875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6613862828818628e-05,
+      "loss": 1.0472,
+      "step": 27880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.658510806734903e-05,
+      "loss": 1.0519,
+      "step": 27885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6556375961085803e-05,
+      "loss": 1.007,
+      "step": 27890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6527666517832408e-05,
+      "loss": 1.0224,
+      "step": 27895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.64989797453863e-05,
+      "loss": 1.0284,
+      "step": 27900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.647031565153866e-05,
+      "loss": 0.9984,
+      "step": 27905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.644167424407461e-05,
+      "loss": 0.9897,
+      "step": 27910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6413055530773004e-05,
+      "loss": 1.0294,
+      "step": 27915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6384459519406648e-05,
+      "loss": 1.0998,
+      "step": 27920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.635588621774208e-05,
+      "loss": 0.9808,
+      "step": 27925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.632733563353971e-05,
+      "loss": 1.0193,
+      "step": 27930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.629880777455384e-05,
+      "loss": 1.0457,
+      "step": 27935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6270302648532488e-05,
+      "loss": 1.0206,
+      "step": 27940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6241820263217544e-05,
+      "loss": 1.0674,
+      "step": 27945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6213360626344754e-05,
+      "loss": 1.0253,
+      "step": 27950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6184923745643667e-05,
+      "loss": 1.0026,
+      "step": 27955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6156509628837644e-05,
+      "loss": 1.0119,
+      "step": 27960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6128118283643824e-05,
+      "loss": 1.0511,
+      "step": 27965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6099749717773238e-05,
+      "loss": 1.0831,
+      "step": 27970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6071403938930708e-05,
+      "loss": 1.0253,
+      "step": 27975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6043080954814828e-05,
+      "loss": 1.0624,
+      "step": 27980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6014780773118e-05,
+      "loss": 0.9995,
+      "step": 27985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5986503401526497e-05,
+      "loss": 1.0632,
+      "step": 27990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.595824884772037e-05,
+      "loss": 1.0359,
+      "step": 27995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.593001711937343e-05,
+      "loss": 0.9918,
+      "step": 28000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5901808224153347e-05,
+      "loss": 1.0573,
+      "step": 28005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5873622169721523e-05,
+      "loss": 0.9937,
+      "step": 28010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5845458963733238e-05,
+      "loss": 1.0302,
+      "step": 28015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5817318613837493e-05,
+      "loss": 1.0473,
+      "step": 28020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5789201127677156e-05,
+      "loss": 0.9898,
+      "step": 28025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.576110651288879e-05,
+      "loss": 0.9981,
+      "step": 28030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.573303477710284e-05,
+      "loss": 1.0857,
+      "step": 28035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.570498592794346e-05,
+      "loss": 1.027,
+      "step": 28040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.567695997302867e-05,
+      "loss": 1.0621,
+      "step": 28045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.564895691997017e-05,
+      "loss": 0.9847,
+      "step": 28050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.562097677637353e-05,
+      "loss": 1.0856,
+      "step": 28055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.559301954983807e-05,
+      "loss": 0.967,
+      "step": 28060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5565085247956878e-05,
+      "loss": 1.0348,
+      "step": 28065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.553717387831678e-05,
+      "loss": 1.031,
+      "step": 28070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5509285448498446e-05,
+      "loss": 1.0164,
+      "step": 28075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.548141996607628e-05,
+      "loss": 0.9814,
+      "step": 28080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5453577438618462e-05,
+      "loss": 1.0759,
+      "step": 28085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5425757873686875e-05,
+      "loss": 1.0123,
+      "step": 28090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.539796127883727e-05,
+      "loss": 1.0785,
+      "step": 28095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5370187661619117e-05,
+      "loss": 1.011,
+      "step": 28100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5342437029575618e-05,
+      "loss": 1.0627,
+      "step": 28105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.531470939024373e-05,
+      "loss": 0.9686,
+      "step": 28110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.528700475115422e-05,
+      "loss": 1.0189,
+      "step": 28115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.525932311983158e-05,
+      "loss": 1.109,
+      "step": 28120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5231664503794042e-05,
+      "loss": 1.0025,
+      "step": 28125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5204028910553526e-05,
+      "loss": 1.0115,
+      "step": 28130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5176416347615885e-05,
+      "loss": 1.0111,
+      "step": 28135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5148826822480533e-05,
+      "loss": 1.1188,
+      "step": 28140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5121260342640686e-05,
+      "loss": 1.008,
+      "step": 28145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5093716915583323e-05,
+      "loss": 1.0404,
+      "step": 28150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.506619654878917e-05,
+      "loss": 1.0092,
+      "step": 28155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5038699249732636e-05,
+      "loss": 1.0153,
+      "step": 28160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5011225025881893e-05,
+      "loss": 1.0057,
+      "step": 28165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4983773884698849e-05,
+      "loss": 0.9793,
+      "step": 28170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4956345833639185e-05,
+      "loss": 1.0471,
+      "step": 28175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4928940880152232e-05,
+      "loss": 1.0523,
+      "step": 28180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4901559031681056e-05,
+      "loss": 1.0299,
+      "step": 28185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4874200295662533e-05,
+      "loss": 1.1044,
+      "step": 28190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4846864679527162e-05,
+      "loss": 1.004,
+      "step": 28195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4819552190699215e-05,
+      "loss": 0.9262,
+      "step": 28200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4792262836596715e-05,
+      "loss": 1.0379,
+      "step": 28205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.476499662463131e-05,
+      "loss": 1.0461,
+      "step": 28210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4737753562208411e-05,
+      "loss": 1.0779,
+      "step": 28215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4710533656727154e-05,
+      "loss": 1.0497,
+      "step": 28220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4683336915580414e-05,
+      "loss": 1.0077,
+      "step": 28225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4656163346154684e-05,
+      "loss": 0.9981,
+      "step": 28230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4629012955830212e-05,
+      "loss": 1.0002,
+      "step": 28235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4601885751980982e-05,
+      "loss": 1.0671,
+      "step": 28240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4574781741974652e-05,
+      "loss": 0.9378,
+      "step": 28245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.454770093317258e-05,
+      "loss": 1.0896,
+      "step": 28250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4520643332929783e-05,
+      "loss": 1.0875,
+      "step": 28255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.449360894859505e-05,
+      "loss": 0.9998,
+      "step": 28260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.446659778751085e-05,
+      "loss": 1.032,
+      "step": 28265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4439609857013304e-05,
+      "loss": 1.0697,
+      "step": 28270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4412645164432181e-05,
+      "loss": 1.0267,
+      "step": 28275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4385703717091115e-05,
+      "loss": 1.088,
+      "step": 28280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.435878552230726e-05,
+      "loss": 0.9866,
+      "step": 28285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4331890587391484e-05,
+      "loss": 1.0554,
+      "step": 28290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4305018919648372e-05,
+      "loss": 0.9817,
+      "step": 28295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4278170526376233e-05,
+      "loss": 1.0928,
+      "step": 28300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4251345414866946e-05,
+      "loss": 1.0782,
+      "step": 28305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4224543592406114e-05,
+      "loss": 1.0246,
+      "step": 28310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.419776506627305e-05,
+      "loss": 1.0067,
+      "step": 28315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4171009843740723e-05,
+      "loss": 1.0644,
+      "step": 28320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4144277932075744e-05,
+      "loss": 1.0037,
+      "step": 28325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4117569338538395e-05,
+      "loss": 0.9673,
+      "step": 28330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4090884070382648e-05,
+      "loss": 1.0349,
+      "step": 28335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4064222134856175e-05,
+      "loss": 1.1054,
+      "step": 28340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4037583539200204e-05,
+      "loss": 1.0639,
+      "step": 28345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4010968290649762e-05,
+      "loss": 1.036,
+      "step": 28350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3984376396433385e-05,
+      "loss": 1.0074,
+      "step": 28355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3957807863773398e-05,
+      "loss": 1.0241,
+      "step": 28360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3931262699885695e-05,
+      "loss": 1.1143,
+      "step": 28365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3904740911979885e-05,
+      "loss": 1.0849,
+      "step": 28370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3878242507259143e-05,
+      "loss": 1.0656,
+      "step": 28375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.385176749292042e-05,
+      "loss": 0.9913,
+      "step": 28380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3825315876154176e-05,
+      "loss": 0.9479,
+      "step": 28385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3798887664144633e-05,
+      "loss": 1.0109,
+      "step": 28390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.377248286406958e-05,
+      "loss": 1.0096,
+      "step": 28395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3746101483100449e-05,
+      "loss": 0.9676,
+      "step": 28400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3719743528402362e-05,
+      "loss": 1.0478,
+      "step": 28405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3693409007134072e-05,
+      "loss": 1.009,
+      "step": 28410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3667097926447924e-05,
+      "loss": 1.0147,
+      "step": 28415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3640810293489881e-05,
+      "loss": 1.0228,
+      "step": 28420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.361454611539965e-05,
+      "loss": 1.1787,
+      "step": 28425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3588305399310475e-05,
+      "loss": 1.0154,
+      "step": 28430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3562088152349195e-05,
+      "loss": 1.0126,
+      "step": 28435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3535894381636372e-05,
+      "loss": 1.0735,
+      "step": 28440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3509724094286157e-05,
+      "loss": 1.078,
+      "step": 28445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3483577297406303e-05,
+      "loss": 1.0255,
+      "step": 28450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3457453998098145e-05,
+      "loss": 1.0796,
+      "step": 28455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3431354203456737e-05,
+      "loss": 0.9961,
+      "step": 28460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3405277920570702e-05,
+      "loss": 0.9999,
+      "step": 28465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3379225156522247e-05,
+      "loss": 1.0835,
+      "step": 28470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3353195918387207e-05,
+      "loss": 1.0221,
+      "step": 28475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3327190213235042e-05,
+      "loss": 1.0482,
+      "step": 28480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3301208048128843e-05,
+      "loss": 0.9903,
+      "step": 28485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3275249430125237e-05,
+      "loss": 1.0239,
+      "step": 28490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3249314366274546e-05,
+      "loss": 1.0476,
+      "step": 28495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3223402863620603e-05,
+      "loss": 1.0797,
+      "step": 28500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3197514929200917e-05,
+      "loss": 1.0183,
+      "step": 28505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3171650570046534e-05,
+      "loss": 1.0112,
+      "step": 28510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3145809793182162e-05,
+      "loss": 1.0847,
+      "step": 28515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3119992605626031e-05,
+      "loss": 1.0585,
+      "step": 28520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.309419901439004e-05,
+      "loss": 0.9641,
+      "step": 28525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3068429026479611e-05,
+      "loss": 1.0217,
+      "step": 28530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3042682648893833e-05,
+      "loss": 1.0853,
+      "step": 28535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3016959888625269e-05,
+      "loss": 1.0434,
+      "step": 28540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2991260752660207e-05,
+      "loss": 0.9911,
+      "step": 28545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2965585247978374e-05,
+      "loss": 1.0899,
+      "step": 28550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2939933381553216e-05,
+      "loss": 1.0117,
+      "step": 28555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2914305160351648e-05,
+      "loss": 1.0095,
+      "step": 28560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2888700591334223e-05,
+      "loss": 0.9692,
+      "step": 28565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2863119681455093e-05,
+      "loss": 0.986,
+      "step": 28570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.283756243766191e-05,
+      "loss": 1.0234,
+      "step": 28575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.281202886689592e-05,
+      "loss": 1.0126,
+      "step": 28580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2786518976091977e-05,
+      "loss": 0.996,
+      "step": 28585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2761032772178506e-05,
+      "loss": 0.9891,
+      "step": 28590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2735570262077456e-05,
+      "loss": 0.9782,
+      "step": 28595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2710131452704333e-05,
+      "loss": 1.0157,
+      "step": 28600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.268471635096825e-05,
+      "loss": 1.0324,
+      "step": 28605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2659324963771902e-05,
+      "loss": 1.0042,
+      "step": 28610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2633957298011467e-05,
+      "loss": 1.0503,
+      "step": 28615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2608613360576705e-05,
+      "loss": 1.0469,
+      "step": 28620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.258329315835095e-05,
+      "loss": 1.0778,
+      "step": 28625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2557996698211139e-05,
+      "loss": 1.0316,
+      "step": 28630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2532723987027628e-05,
+      "loss": 1.0513,
+      "step": 28635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2507475031664473e-05,
+      "loss": 1.0377,
+      "step": 28640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2482249838979142e-05,
+      "loss": 0.9151,
+      "step": 28645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2457048415822769e-05,
+      "loss": 0.9903,
+      "step": 28650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2431870769039922e-05,
+      "loss": 1.1097,
+      "step": 28655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2406716905468828e-05,
+      "loss": 0.9961,
+      "step": 28660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2381586831941128e-05,
+      "loss": 0.97,
+      "step": 28665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2356480555282113e-05,
+      "loss": 0.9823,
+      "step": 28670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.233139808231053e-05,
+      "loss": 0.9611,
+      "step": 28675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2306339419838742e-05,
+      "loss": 0.9893,
+      "step": 28680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2281304574672537e-05,
+      "loss": 1.0319,
+      "step": 28685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2256293553611353e-05,
+      "loss": 0.9922,
+      "step": 28690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2231306363448059e-05,
+      "loss": 1.0483,
+      "step": 28695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2206343010969124e-05,
+      "loss": 0.9704,
+      "step": 28700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2181403502954469e-05,
+      "loss": 1.0138,
+      "step": 28705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.215648784617761e-05,
+      "loss": 1.0773,
+      "step": 28710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.213159604740557e-05,
+      "loss": 1.0845,
+      "step": 28715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2106728113398869e-05,
+      "loss": 1.0434,
+      "step": 28720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2081884050911518e-05,
+      "loss": 1.0453,
+      "step": 28725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.205706386669112e-05,
+      "loss": 1.0684,
+      "step": 28730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.203226756747876e-05,
+      "loss": 0.9965,
+      "step": 28735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.200749516000902e-05,
+      "loss": 0.9462,
+      "step": 28740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1982746651009969e-05,
+      "loss": 0.9895,
+      "step": 28745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1958022047203254e-05,
+      "loss": 1.0961,
+      "step": 28750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1933321355304006e-05,
+      "loss": 1.0407,
+      "step": 28755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1908644582020845e-05,
+      "loss": 0.9402,
+      "step": 28760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1883991734055844e-05,
+      "loss": 1.1078,
+      "step": 28765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1859362818104714e-05,
+      "loss": 1.0555,
+      "step": 28770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1834757840856559e-05,
+      "loss": 1.0341,
+      "step": 28775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1810176808993979e-05,
+      "loss": 1.0119,
+      "step": 28780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.178561972919312e-05,
+      "loss": 1.0778,
+      "step": 28785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1761086608123638e-05,
+      "loss": 0.9924,
+      "step": 28790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1736577452448604e-05,
+      "loss": 1.0209,
+      "step": 28795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1712092268824626e-05,
+      "loss": 1.0213,
+      "step": 28800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1687631063901817e-05,
+      "loss": 1.0446,
+      "step": 28805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1663193844323727e-05,
+      "loss": 1.0702,
+      "step": 28810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1638780616727463e-05,
+      "loss": 1.0332,
+      "step": 28815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.161439138774354e-05,
+      "loss": 1.0109,
+      "step": 28820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1590026163996026e-05,
+      "loss": 0.979,
+      "step": 28825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1565684952102395e-05,
+      "loss": 0.9863,
+      "step": 28830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1541367758673693e-05,
+      "loss": 1.0271,
+      "step": 28835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1517074590314324e-05,
+      "loss": 0.9968,
+      "step": 28840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1492805453622291e-05,
+      "loss": 1.0801,
+      "step": 28845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1468560355188961e-05,
+      "loss": 0.9222,
+      "step": 28850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1444339301599239e-05,
+      "loss": 1.0151,
+      "step": 28855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1420142299431513e-05,
+      "loss": 0.994,
+      "step": 28860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1395969355257586e-05,
+      "loss": 1.0436,
+      "step": 28865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1371820475642725e-05,
+      "loss": 0.9966,
+      "step": 28870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1347695667145697e-05,
+      "loss": 1.0715,
+      "step": 28875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.132359493631875e-05,
+      "loss": 0.974,
+      "step": 28880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1299518289707545e-05,
+      "loss": 1.0343,
+      "step": 28885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1275465733851199e-05,
+      "loss": 0.9999,
+      "step": 28890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1251437275282306e-05,
+      "loss": 1.0461,
+      "step": 28895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.122743292052697e-05,
+      "loss": 1.0052,
+      "step": 28900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1203452676104642e-05,
+      "loss": 1.002,
+      "step": 28905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1179496548528246e-05,
+      "loss": 1.0442,
+      "step": 28910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1155564544304275e-05,
+      "loss": 1.0645,
+      "step": 28915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1131656669932532e-05,
+      "loss": 1.0118,
+      "step": 28920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1107772931906301e-05,
+      "loss": 1.0461,
+      "step": 28925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1083913336712337e-05,
+      "loss": 1.0071,
+      "step": 28930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1060077890830856e-05,
+      "loss": 0.9882,
+      "step": 28935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1036266600735468e-05,
+      "loss": 1.0393,
+      "step": 28940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1012479472893212e-05,
+      "loss": 1.0415,
+      "step": 28945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0988716513764618e-05,
+      "loss": 0.9953,
+      "step": 28950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0964977729803638e-05,
+      "loss": 1.0169,
+      "step": 28955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0941263127457635e-05,
+      "loss": 1.0323,
+      "step": 28960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.091757271316739e-05,
+      "loss": 1.0477,
+      "step": 28965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0893906493367168e-05,
+      "loss": 0.9649,
+      "step": 28970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0870264474484648e-05,
+      "loss": 1.0664,
+      "step": 28975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0846646662940907e-05,
+      "loss": 0.9779,
+      "step": 28980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0823053065150435e-05,
+      "loss": 1.0893,
+      "step": 28985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0799483687521216e-05,
+      "loss": 1.1045,
+      "step": 28990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.077593853645461e-05,
+      "loss": 1.0906,
+      "step": 28995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0752417618345378e-05,
+      "loss": 1.0487,
+      "step": 29000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0728920939581755e-05,
+      "loss": 1.1194,
+      "step": 29005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0705448506545346e-05,
+      "loss": 1.0141,
+      "step": 29010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0682000325611163e-05,
+      "loss": 1.0628,
+      "step": 29015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0658576403147668e-05,
+      "loss": 1.0212,
+      "step": 29020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.063517674551674e-05,
+      "loss": 1.0414,
+      "step": 29025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0611801359073648e-05,
+      "loss": 1.0779,
+      "step": 29030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0588450250167026e-05,
+      "loss": 1.067,
+      "step": 29035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0565123425138978e-05,
+      "loss": 0.9864,
+      "step": 29040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0541820890325027e-05,
+      "loss": 0.9991,
+      "step": 29045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0518542652054031e-05,
+      "loss": 0.9543,
+      "step": 29050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0495288716648256e-05,
+      "loss": 1.0237,
+      "step": 29055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0472059090423458e-05,
+      "loss": 0.9585,
+      "step": 29060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0448853779688694e-05,
+      "loss": 0.9428,
+      "step": 29065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0425672790746421e-05,
+      "loss": 0.9818,
+      "step": 29070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0402516129892548e-05,
+      "loss": 1.0123,
+      "step": 29075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0379383803416364e-05,
+      "loss": 0.9868,
+      "step": 29080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0356275817600503e-05,
+      "loss": 0.928,
+      "step": 29085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0333192178721007e-05,
+      "loss": 1.0086,
+      "step": 29090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0310132893047342e-05,
+      "loss": 1.0187,
+      "step": 29095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0287097966842341e-05,
+      "loss": 0.9889,
+      "step": 29100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0264087406362188e-05,
+      "loss": 0.9902,
+      "step": 29105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0241101217856464e-05,
+      "loss": 1.0069,
+      "step": 29110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0218139407568161e-05,
+      "loss": 0.9962,
+      "step": 29115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0195201981733659e-05,
+      "loss": 1.0604,
+      "step": 29120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.017228894658262e-05,
+      "loss": 1.03,
+      "step": 29125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0149400308338209e-05,
+      "loss": 0.9777,
+      "step": 29130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0126536073216863e-05,
+      "loss": 1.0084,
+      "step": 29135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0103696247428463e-05,
+      "loss": 1.1209,
+      "step": 29140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0080880837176198e-05,
+      "loss": 1.1022,
+      "step": 29145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0058089848656694e-05,
+      "loss": 1.0183,
+      "step": 29150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.003532328805986e-05,
+      "loss": 1.0363,
+      "step": 29155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0012581161569067e-05,
+      "loss": 1.0147,
+      "step": 29160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.989863475360961e-06,
+      "loss": 0.9523,
+      "step": 29165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.967170235605616e-06,
+      "loss": 1.0281,
+      "step": 29170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.944501448466426e-06,
+      "loss": 1.0638,
+      "step": 29175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.921857120100174e-06,
+      "loss": 1.0431,
+      "step": 29180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.899237256656958e-06,
+      "loss": 1.0272,
+      "step": 29185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.87664186428029e-06,
+      "loss": 1.023,
+      "step": 29190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.854070949106964e-06,
+      "loss": 1.0809,
+      "step": 29195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.831524517267188e-06,
+      "loss": 1.0156,
+      "step": 29200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.809002574884518e-06,
+      "loss": 1.0842,
+      "step": 29205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.786505128075817e-06,
+      "loss": 0.9836,
+      "step": 29210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.7640321829513e-06,
+      "loss": 0.9605,
+      "step": 29215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.741583745614557e-06,
+      "loss": 1.0973,
+      "step": 29220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.719159822162538e-06,
+      "loss": 1.072,
+      "step": 29225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.69676041868547e-06,
+      "loss": 1.0483,
+      "step": 29230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.674385541266951e-06,
+      "loss": 1.0286,
+      "step": 29235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.652035195983943e-06,
+      "loss": 0.9802,
+      "step": 29240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.629709388906728e-06,
+      "loss": 1.0628,
+      "step": 29245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.607408126098927e-06,
+      "loss": 0.9956,
+      "step": 29250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.585131413617442e-06,
+      "loss": 1.0854,
+      "step": 29255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.562879257512602e-06,
+      "loss": 1.0574,
+      "step": 29260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.540651663828004e-06,
+      "loss": 1.0655,
+      "step": 29265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.518448638600585e-06,
+      "loss": 1.0279,
+      "step": 29270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.496270187860622e-06,
+      "loss": 1.0302,
+      "step": 29275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.474116317631687e-06,
+      "loss": 1.0097,
+      "step": 29280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.451987033930732e-06,
+      "loss": 1.039,
+      "step": 29285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.429882342767949e-06,
+      "loss": 1.0546,
+      "step": 29290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.407802250146946e-06,
+      "loss": 0.9941,
+      "step": 29295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.385746762064573e-06,
+      "loss": 0.979,
+      "step": 29300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.363715884511049e-06,
+      "loss": 1.0566,
+      "step": 29305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.341709623469851e-06,
+      "loss": 0.973,
+      "step": 29310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.319727984917848e-06,
+      "loss": 1.0016,
+      "step": 29315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.297770974825149e-06,
+      "loss": 0.9822,
+      "step": 29320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.275838599155217e-06,
+      "loss": 1.0752,
+      "step": 29325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.253930863864813e-06,
+      "loss": 1.0631,
+      "step": 29330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.232047774904007e-06,
+      "loss": 1.0858,
+      "step": 29335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.210189338216158e-06,
+      "loss": 1.0597,
+      "step": 29340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.188355559737949e-06,
+      "loss": 1.0116,
+      "step": 29345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.166546445399394e-06,
+      "loss": 1.0643,
+      "step": 29350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.144762001123742e-06,
+      "loss": 1.0158,
+      "step": 29355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.123002232827571e-06,
+      "loss": 1.1442,
+      "step": 29360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.101267146420767e-06,
+      "loss": 0.9952,
+      "step": 29365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.079556747806527e-06,
+      "loss": 1.009,
+      "step": 29370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.057871042881317e-06,
+      "loss": 1.0316,
+      "step": 29375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.036210037534875e-06,
+      "loss": 1.0294,
+      "step": 29380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.014573737650278e-06,
+      "loss": 0.9652,
+      "step": 29385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.99296214910389e-06,
+      "loss": 0.9561,
+      "step": 29390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.971375277765326e-06,
+      "loss": 1.0152,
+      "step": 29395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.949813129497464e-06,
+      "loss": 1.025,
+      "step": 29400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.928275710156608e-06,
+      "loss": 1.1015,
+      "step": 29405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.90676302559219e-06,
+      "loss": 1.0681,
+      "step": 29410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.885275081646982e-06,
+      "loss": 0.972,
+      "step": 29415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.863811884157058e-06,
+      "loss": 1.0367,
+      "step": 29420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.842373438951735e-06,
+      "loss": 1.0551,
+      "step": 29425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.82095975185364e-06,
+      "loss": 0.9054,
+      "step": 29430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.799570828678627e-06,
+      "loss": 1.0166,
+      "step": 29435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.7782066752359e-06,
+      "loss": 1.0402,
+      "step": 29440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.756867297327842e-06,
+      "loss": 1.0522,
+      "step": 29445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.735552700750215e-06,
+      "loss": 1.0613,
+      "step": 29450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.714262891291935e-06,
+      "loss": 1.015,
+      "step": 29455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.692997874735287e-06,
+      "loss": 1.0503,
+      "step": 29460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.671757656855751e-06,
+      "loss": 1.0676,
+      "step": 29465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.650542243422122e-06,
+      "loss": 0.9613,
+      "step": 29470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.629351640196404e-06,
+      "loss": 1.0183,
+      "step": 29475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.608185852933925e-06,
+      "loss": 1.0021,
+      "step": 29480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.587044887383221e-06,
+      "loss": 0.9626,
+      "step": 29485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.565928749286111e-06,
+      "loss": 0.9981,
+      "step": 29490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.544837444377695e-06,
+      "loss": 1.037,
+      "step": 29495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.523770978386281e-06,
+      "loss": 1.0432,
+      "step": 29500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.502729357033434e-06,
+      "loss": 0.9933,
+      "step": 29505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.481712586033997e-06,
+      "loss": 0.9996,
+      "step": 29510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.46072067109609e-06,
+      "loss": 1.0697,
+      "step": 29515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.43975361792101e-06,
+      "loss": 1.0159,
+      "step": 29520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.418811432203322e-06,
+      "loss": 1.0305,
+      "step": 29525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.39789411963089e-06,
+      "loss": 1.0169,
+      "step": 29530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.377001685884789e-06,
+      "loss": 1.0061,
+      "step": 29535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.356134136639326e-06,
+      "loss": 1.0258,
+      "step": 29540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.335291477562012e-06,
+      "loss": 1.053,
+      "step": 29545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.314473714313719e-06,
+      "loss": 1.0076,
+      "step": 29550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.293680852548447e-06,
+      "loss": 1.0291,
+      "step": 29555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.272912897913443e-06,
+      "loss": 0.9826,
+      "step": 29560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.252169856049253e-06,
+      "loss": 1.0395,
+      "step": 29565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.231451732589613e-06,
+      "loss": 1.0104,
+      "step": 29570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.210758533161488e-06,
+      "loss": 0.9514,
+      "step": 29575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.190090263385063e-06,
+      "loss": 1.016,
+      "step": 29580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.169446928873792e-06,
+      "loss": 1.006,
+      "step": 29585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.148828535234353e-06,
+      "loss": 0.9663,
+      "step": 29590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.128235088066605e-06,
+      "loss": 0.9892,
+      "step": 29595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.107666592963659e-06,
+      "loss": 0.9732,
+      "step": 29600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.08712305551187e-06,
+      "loss": 1.0466,
+      "step": 29605
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.066604481290762e-06,
+      "loss": 0.9989,
+      "step": 29610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.046110875873159e-06,
+      "loss": 1.0269,
+      "step": 29615
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.025642244825005e-06,
+      "loss": 0.9953,
+      "step": 29620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.005198593705554e-06,
+      "loss": 1.0127,
+      "step": 29625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.984779928067198e-06,
+      "loss": 1.0516,
+      "step": 29630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.964386253455591e-06,
+      "loss": 0.9699,
+      "step": 29635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.944017575409612e-06,
+      "loss": 1.0953,
+      "step": 29640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.923673899461303e-06,
+      "loss": 1.0265,
+      "step": 29645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.903355231135923e-06,
+      "loss": 1.071,
+      "step": 29650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.883061575951978e-06,
+      "loss": 1.0221,
+      "step": 29655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.86279293942116e-06,
+      "loss": 1.0335,
+      "step": 29660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.842549327048365e-06,
+      "loss": 1.008,
+      "step": 29665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.822330744331653e-06,
+      "loss": 1.0719,
+      "step": 29670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.80213719676235e-06,
+      "loss": 0.998,
+      "step": 29675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.781968689824981e-06,
+      "loss": 1.0141,
+      "step": 29680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.761825228997222e-06,
+      "loss": 1.0259,
+      "step": 29685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.741706819749928e-06,
+      "loss": 1.0285,
+      "step": 29690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.721613467547274e-06,
+      "loss": 0.9429,
+      "step": 29695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.701545177846492e-06,
+      "loss": 1.0391,
+      "step": 29700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.68150195609807e-06,
+      "loss": 0.9707,
+      "step": 29705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.661483807745683e-06,
+      "loss": 1.014,
+      "step": 29710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.641490738226221e-06,
+      "loss": 1.0049,
+      "step": 29715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.621522752969701e-06,
+      "loss": 1.1093,
+      "step": 29720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.601579857399354e-06,
+      "loss": 1.125,
+      "step": 29725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.581662056931626e-06,
+      "loss": 1.0378,
+      "step": 29730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.561769356976145e-06,
+      "loss": 1.0284,
+      "step": 29735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.541901762935666e-06,
+      "loss": 1.0965,
+      "step": 29740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.522059280206151e-06,
+      "loss": 1.1078,
+      "step": 29745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.502241914176789e-06,
+      "loss": 1.1361,
+      "step": 29750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.482449670229896e-06,
+      "loss": 1.0439,
+      "step": 29755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.462682553740985e-06,
+      "loss": 0.986,
+      "step": 29760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.4429405700787155e-06,
+      "loss": 1.0304,
+      "step": 29765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.423223724604967e-06,
+      "loss": 1.0031,
+      "step": 29770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.403532022674764e-06,
+      "loss": 0.9565,
+      "step": 29775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.383865469636298e-06,
+      "loss": 1.036,
+      "step": 29780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.3642240708309715e-06,
+      "loss": 1.0881,
+      "step": 29785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.3446078315932755e-06,
+      "loss": 1.0737,
+      "step": 29790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.325016757250946e-06,
+      "loss": 0.9994,
+      "step": 29795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.30545085312484e-06,
+      "loss": 1.1036,
+      "step": 29800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.285910124529016e-06,
+      "loss": 1.0143,
+      "step": 29805
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.266394576770641e-06,
+      "loss": 0.9984,
+      "step": 29810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.2469042151500745e-06,
+      "loss": 1.0046,
+      "step": 29815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.227439044960838e-06,
+      "loss": 0.9868,
+      "step": 29820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.207999071489635e-06,
+      "loss": 0.9912,
+      "step": 29825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.188584300016266e-06,
+      "loss": 0.9825,
+      "step": 29830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.1691947358137005e-06,
+      "loss": 1.0695,
+      "step": 29835
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.149830384148137e-06,
+      "loss": 0.9898,
+      "step": 29840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.1304912502788365e-06,
+      "loss": 0.9602,
+      "step": 29845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.11117733945823e-06,
+      "loss": 1.0741,
+      "step": 29850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.091888656931922e-06,
+      "loss": 0.9869,
+      "step": 29855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.072625207938665e-06,
+      "loss": 1.0476,
+      "step": 29860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.053386997710354e-06,
+      "loss": 0.9677,
+      "step": 29865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.034174031471985e-06,
+      "loss": 0.9455,
+      "step": 29870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.014986314441762e-06,
+      "loss": 0.9775,
+      "step": 29875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.995823851831007e-06,
+      "loss": 1.0326,
+      "step": 29880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.976686648844188e-06,
+      "loss": 1.0552,
+      "step": 29885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.9575747106788715e-06,
+      "loss": 1.0421,
+      "step": 29890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.9384880425258255e-06,
+      "loss": 1.0043,
+      "step": 29895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.919426649568927e-06,
+      "loss": 0.9058,
+      "step": 29900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.900390536985179e-06,
+      "loss": 0.9703,
+      "step": 29905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.881379709944713e-06,
+      "loss": 1.0117,
+      "step": 29910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.862394173610809e-06,
+      "loss": 1.0065,
+      "step": 29915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.843433933139909e-06,
+      "loss": 1.0246,
+      "step": 29920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.8244989936815054e-06,
+      "loss": 1.0314,
+      "step": 29925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.805589360378306e-06,
+      "loss": 1.0423,
+      "step": 29930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.786705038366059e-06,
+      "loss": 1.0474,
+      "step": 29935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.767846032773728e-06,
+      "loss": 1.0135,
+      "step": 29940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.749012348723327e-06,
+      "loss": 1.0369,
+      "step": 29945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7302039913300304e-06,
+      "loss": 1.0016,
+      "step": 29950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7114209657021195e-06,
+      "loss": 1.0689,
+      "step": 29955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6926632769410245e-06,
+      "loss": 1.0176,
+      "step": 29960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.673930930141237e-06,
+      "loss": 0.9772,
+      "step": 29965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.655223930390431e-06,
+      "loss": 1.0156,
+      "step": 29970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.636542282769343e-06,
+      "loss": 0.9712,
+      "step": 29975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.617885992351846e-06,
+      "loss": 0.9406,
+      "step": 29980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.599255064204946e-06,
+      "loss": 1.0461,
+      "step": 29985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.580649503388736e-06,
+      "loss": 1.0436,
+      "step": 29990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.562069314956409e-06,
+      "loss": 1.1097,
+      "step": 29995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.543514503954273e-06,
+      "loss": 0.9697,
+      "step": 30000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.5249850754217965e-06,
+      "loss": 1.0786,
+      "step": 30005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.5064810343914875e-06,
+      "loss": 1.0166,
+      "step": 30010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.488002385888959e-06,
+      "loss": 1.0406,
+      "step": 30015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.469549134932973e-06,
+      "loss": 0.9545,
+      "step": 30020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.451121286535378e-06,
+      "loss": 1.0461,
+      "step": 30025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.432718845701114e-06,
+      "loss": 1.0034,
+      "step": 30030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.414341817428193e-06,
+      "loss": 1.0189,
+      "step": 30035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.39599020670778e-06,
+      "loss": 1.0154,
+      "step": 30040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.377664018524121e-06,
+      "loss": 0.9909,
+      "step": 30045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.359363257854534e-06,
+      "loss": 0.9915,
+      "step": 30050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.34108792966942e-06,
+      "loss": 0.9411,
+      "step": 30055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.3228380389323325e-06,
+      "loss": 1.0642,
+      "step": 30060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.304613590599884e-06,
+      "loss": 0.9811,
+      "step": 30065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.2864145896217475e-06,
+      "loss": 1.1242,
+      "step": 30070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.268241040940737e-06,
+      "loss": 1.0141,
+      "step": 30075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.250092949492692e-06,
+      "loss": 1.0918,
+      "step": 30080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.231970320206615e-06,
+      "loss": 0.9903,
+      "step": 30085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.213873158004524e-06,
+      "loss": 1.0144,
+      "step": 30090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.1958014678015655e-06,
+      "loss": 1.0565,
+      "step": 30095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.1777552545059235e-06,
+      "loss": 1.0393,
+      "step": 30100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.159734523018923e-06,
+      "loss": 1.0371,
+      "step": 30105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.141739278234904e-06,
+      "loss": 0.9804,
+      "step": 30110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.123769525041345e-06,
+      "loss": 1.0075,
+      "step": 30115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.105825268318743e-06,
+      "loss": 1.0249,
+      "step": 30120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.0879065129407e-06,
+      "loss": 0.9814,
+      "step": 30125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.070013263773921e-06,
+      "loss": 0.9834,
+      "step": 30130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.052145525678144e-06,
+      "loss": 1.0159,
+      "step": 30135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.03430330350615e-06,
+      "loss": 1.0173,
+      "step": 30140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.016486602103855e-06,
+      "loss": 0.9885,
+      "step": 30145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.998695426310219e-06,
+      "loss": 1.0852,
+      "step": 30150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.980929780957267e-06,
+      "loss": 0.9527,
+      "step": 30155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.963189670870073e-06,
+      "loss": 1.0285,
+      "step": 30160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.945475100866793e-06,
+      "loss": 1.0206,
+      "step": 30165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.927786075758668e-06,
+      "loss": 0.97,
+      "step": 30170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.910122600349965e-06,
+      "loss": 0.9686,
+      "step": 30175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.892484679438004e-06,
+      "loss": 1.0289,
+      "step": 30180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.874872317813218e-06,
+      "loss": 0.9962,
+      "step": 30185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.857285520259059e-06,
+      "loss": 1.0703,
+      "step": 30190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.839724291552018e-06,
+      "loss": 0.9323,
+      "step": 30195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8221886364616895e-06,
+      "loss": 0.963,
+      "step": 30200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8046785597507096e-06,
+      "loss": 1.0458,
+      "step": 30205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.787194066174739e-06,
+      "loss": 1.0356,
+      "step": 30210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7697351604824925e-06,
+      "loss": 1.0753,
+      "step": 30215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.752301847415775e-06,
+      "loss": 0.9927,
+      "step": 30220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.734894131709401e-06,
+      "loss": 0.9767,
+      "step": 30225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7175120180912755e-06,
+      "loss": 1.0294,
+      "step": 30230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.700155511282279e-06,
+      "loss": 1.0514,
+      "step": 30235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.682824615996418e-06,
+      "loss": 1.0034,
+      "step": 30240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6655193369406835e-06,
+      "loss": 1.0552,
+      "step": 30245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.64823967881516e-06,
+      "loss": 0.9603,
+      "step": 30250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.630985646312903e-06,
+      "loss": 1.0037,
+      "step": 30255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.613757244120088e-06,
+      "loss": 1.111,
+      "step": 30260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.596554476915861e-06,
+      "loss": 1.068,
+      "step": 30265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.57937734937245e-06,
+      "loss": 0.9845,
+      "step": 30270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.562225866155124e-06,
+      "loss": 1.0597,
+      "step": 30275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5451000319221456e-06,
+      "loss": 0.9797,
+      "step": 30280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.527999851324839e-06,
+      "loss": 1.0561,
+      "step": 30285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.510925329007544e-06,
+      "loss": 0.9576,
+      "step": 30290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4938764696076726e-06,
+      "loss": 1.0983,
+      "step": 30295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.476853277755634e-06,
+      "loss": 1.0115,
+      "step": 30300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.459855758074839e-06,
+      "loss": 0.9431,
+      "step": 30305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.4428839151817735e-06,
+      "loss": 1.0721,
+      "step": 30310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.42593775368595e-06,
+      "loss": 1.0002,
+      "step": 30315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.409017278189876e-06,
+      "loss": 0.9735,
+      "step": 30320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.392122493289076e-06,
+      "loss": 0.9894,
+      "step": 30325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.375253403572167e-06,
+      "loss": 1.0444,
+      "step": 30330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.3584100136206965e-06,
+      "loss": 0.9873,
+      "step": 30335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.341592328009282e-06,
+      "loss": 1.0287,
+      "step": 30340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.324800351305548e-06,
+      "loss": 1.0381,
+      "step": 30345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.308034088070157e-06,
+      "loss": 0.9822,
+      "step": 30350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.291293542856768e-06,
+      "loss": 1.0641,
+      "step": 30355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.27457872021202e-06,
+      "loss": 1.0431,
+      "step": 30360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.257889624675638e-06,
+      "loss": 1.0115,
+      "step": 30365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.24122626078033e-06,
+      "loss": 1.0226,
+      "step": 30370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2245886330517966e-06,
+      "loss": 0.9593,
+      "step": 30375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.207976746008758e-06,
+      "loss": 0.9483,
+      "step": 30380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.191390604162938e-06,
+      "loss": 1.0203,
+      "step": 30385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.174830212019111e-06,
+      "loss": 1.0085,
+      "step": 30390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.158295574075e-06,
+      "loss": 1.0406,
+      "step": 30395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.14178669482136e-06,
+      "loss": 0.9713,
+      "step": 30400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.1253035787419356e-06,
+      "loss": 1.0611,
+      "step": 30405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.1088462303135245e-06,
+      "loss": 0.9981,
+      "step": 30410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.092414654005839e-06,
+      "loss": 0.9874,
+      "step": 30415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.076008854281689e-06,
+      "loss": 1.0721,
+      "step": 30420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.059628835596808e-06,
+      "loss": 1.0144,
+      "step": 30425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.043274602399939e-06,
+      "loss": 1.0652,
+      "step": 30430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.026946159132851e-06,
+      "loss": 1.026,
+      "step": 30435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.010643510230317e-06,
+      "loss": 1.0192,
+      "step": 30440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.994366660120064e-06,
+      "loss": 1.0525,
+      "step": 30445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.978115613222823e-06,
+      "loss": 1.063,
+      "step": 30450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.961890373952327e-06,
+      "loss": 1.0349,
+      "step": 30455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.945690946715309e-06,
+      "loss": 1.0278,
+      "step": 30460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.929517335911482e-06,
+      "loss": 1.0273,
+      "step": 30465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.913369545933499e-06,
+      "loss": 0.9807,
+      "step": 30470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.897247581167108e-06,
+      "loss": 1.0034,
+      "step": 30475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8811514459909505e-06,
+      "loss": 0.9906,
+      "step": 30480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.865081144776684e-06,
+      "loss": 1.0638,
+      "step": 30485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.849036681888941e-06,
+      "loss": 1.0657,
+      "step": 30490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8330180616853686e-06,
+      "loss": 1.0897,
+      "step": 30495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.817025288516563e-06,
+      "loss": 1.0893,
+      "step": 30500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8010583667260856e-06,
+      "loss": 1.0758,
+      "step": 30505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.78511730065051e-06,
+      "loss": 1.0609,
+      "step": 30510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7692020946193964e-06,
+      "loss": 1.0287,
+      "step": 30515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.753312752955252e-06,
+      "loss": 1.0499,
+      "step": 30520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7374492799735385e-06,
+      "loss": 1.009,
+      "step": 30525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.721611679982763e-06,
+      "loss": 1.0475,
+      "step": 30530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705799957284351e-06,
+      "loss": 1.0604,
+      "step": 30535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.690014116172714e-06,
+      "loss": 0.9854,
+      "step": 30540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.674254160935221e-06,
+      "loss": 1.012,
+      "step": 30545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.658520095852225e-06,
+      "loss": 1.0591,
+      "step": 30550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.642811925197077e-06,
+      "loss": 1.0596,
+      "step": 30555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.627129653236028e-06,
+      "loss": 0.9995,
+      "step": 30560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6114732842283515e-06,
+      "loss": 0.9355,
+      "step": 30565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.595842822426244e-06,
+      "loss": 1.0731,
+      "step": 30570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5802382720749215e-06,
+      "loss": 1.005,
+      "step": 30575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.564659637412483e-06,
+      "loss": 1.048,
+      "step": 30580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5491069226700764e-06,
+      "loss": 0.9765,
+      "step": 30585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.533580132071736e-06,
+      "loss": 1.0934,
+      "step": 30590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.518079269834508e-06,
+      "loss": 1.0305,
+      "step": 30595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.502604340168349e-06,
+      "loss": 0.9488,
+      "step": 30600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.487155347276228e-06,
+      "loss": 1.0143,
+      "step": 30605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.471732295354014e-06,
+      "loss": 0.966,
+      "step": 30610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.456335188590577e-06,
+      "loss": 1.0043,
+      "step": 30615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.440964031167716e-06,
+      "loss": 0.9505,
+      "step": 30620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.425618827260181e-06,
+      "loss": 1.0688,
+      "step": 30625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.41029958103567e-06,
+      "loss": 1.0354,
+      "step": 30630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3950062966548465e-06,
+      "loss": 0.9514,
+      "step": 30635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.37973897827133e-06,
+      "loss": 1.021,
+      "step": 30640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.364497630031661e-06,
+      "loss": 1.0498,
+      "step": 30645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.349282256075316e-06,
+      "loss": 1.028,
+      "step": 30650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.334092860534777e-06,
+      "loss": 0.9736,
+      "step": 30655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.318929447535425e-06,
+      "loss": 1.0622,
+      "step": 30660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.303792021195574e-06,
+      "loss": 1.0702,
+      "step": 30665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.288680585626514e-06,
+      "loss": 1.0289,
+      "step": 30670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.27359514493244e-06,
+      "loss": 1.0017,
+      "step": 30675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.258535703210553e-06,
+      "loss": 1.0773,
+      "step": 30680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2435022645509025e-06,
+      "loss": 1.0107,
+      "step": 30685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.228494833036534e-06,
+      "loss": 1.0461,
+      "step": 30690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.213513412743419e-06,
+      "loss": 1.0325,
+      "step": 30695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.198558007740461e-06,
+      "loss": 1.0663,
+      "step": 30700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.183628622089486e-06,
+      "loss": 0.9985,
+      "step": 30705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.168725259845296e-06,
+      "loss": 0.9302,
+      "step": 30710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.153847925055554e-06,
+      "loss": 1.0977,
+      "step": 30715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.138996621760938e-06,
+      "loss": 1.0143,
+      "step": 30720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1241713539949656e-06,
+      "loss": 0.9493,
+      "step": 30725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1093721257841725e-06,
+      "loss": 0.9799,
+      "step": 30730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.094598941147954e-06,
+      "loss": 1.0643,
+      "step": 30735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.079851804098666e-06,
+      "loss": 1.0517,
+      "step": 30740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.065130718641574e-06,
+      "loss": 1.0464,
+      "step": 30745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0504356887748895e-06,
+      "loss": 1.0213,
+      "step": 30750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.03576671848972e-06,
+      "loss": 1.0531,
+      "step": 30755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.021123811770111e-06,
+      "loss": 0.9338,
+      "step": 30760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.006506972593049e-06,
+      "loss": 1.0024,
+      "step": 30765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.991916204928403e-06,
+      "loss": 1.0947,
+      "step": 30770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.977351512738958e-06,
+      "loss": 1.0693,
+      "step": 30775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.962812899980451e-06,
+      "loss": 0.9643,
+      "step": 30780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.948300370601532e-06,
+      "loss": 1.0943,
+      "step": 30785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.93381392854375e-06,
+      "loss": 1.009,
+      "step": 30790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.919353577741558e-06,
+      "loss": 1.0669,
+      "step": 30795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.904919322122346e-06,
+      "loss": 1.0138,
+      "step": 30800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.890511165606437e-06,
+      "loss": 1.1009,
+      "step": 30805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.876129112107008e-06,
+      "loss": 1.0571,
+      "step": 30810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.861773165530169e-06,
+      "loss": 1.092,
+      "step": 30815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.84744332977498e-06,
+      "loss": 0.9928,
+      "step": 30820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.833139608733361e-06,
+      "loss": 1.078,
+      "step": 30825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.818862006290158e-06,
+      "loss": 1.0385,
+      "step": 30830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8046105263231048e-06,
+      "loss": 1.026,
+      "step": 30835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7903851727028596e-06,
+      "loss": 1.0774,
+      "step": 30840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7761859492929986e-06,
+      "loss": 1.0394,
+      "step": 30845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7620128599499706e-06,
+      "loss": 1.0073,
+      "step": 30850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7478659085231514e-06,
+      "loss": 0.9867,
+      "step": 30855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7337450988547793e-06,
+      "loss": 1.0024,
+      "step": 30860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.719650434780053e-06,
+      "loss": 1.0284,
+      "step": 30865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7055819201269992e-06,
+      "loss": 1.06,
+      "step": 30870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.691539558716617e-06,
+      "loss": 1.1062,
+      "step": 30875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.677523354362733e-06,
+      "loss": 1.0175,
+      "step": 30880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.663533310872136e-06,
+      "loss": 0.9485,
+      "step": 30885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.649569432044442e-06,
+      "loss": 1.0013,
+      "step": 30890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.635631721672206e-06,
+      "loss": 0.9587,
+      "step": 30895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6217201835408666e-06,
+      "loss": 0.995,
+      "step": 30900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.607834821428746e-06,
+      "loss": 0.9713,
+      "step": 30905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5939756391070722e-06,
+      "loss": 0.993,
+      "step": 30910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.580142640339956e-06,
+      "loss": 1.0662,
+      "step": 30915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5663358288843596e-06,
+      "loss": 1.0542,
+      "step": 30920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.552555208490205e-06,
+      "loss": 1.0271,
+      "step": 30925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5388007829002533e-06,
+      "loss": 1.1152,
+      "step": 30930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.525072555850151e-06,
+      "loss": 0.9565,
+      "step": 30935
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5113705310684363e-06,
+      "loss": 0.9986,
+      "step": 30940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4976947122765447e-06,
+      "loss": 1.0819,
+      "step": 30945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4840451031887933e-06,
+      "loss": 1.0837,
+      "step": 30950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4704217075123502e-06,
+      "loss": 0.9727,
+      "step": 30955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4568245289472667e-06,
+      "loss": 1.012,
+      "step": 30960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.443253571186533e-06,
+      "loss": 1.1164,
+      "step": 30965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.429708837915968e-06,
+      "loss": 0.9704,
+      "step": 30970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4161903328142618e-06,
+      "loss": 1.0039,
+      "step": 30975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4026980595529777e-06,
+      "loss": 0.987,
+      "step": 30980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3892320217966176e-06,
+      "loss": 0.9569,
+      "step": 30985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3757922232024898e-06,
+      "loss": 1.033,
+      "step": 30990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.362378667420785e-06,
+      "loss": 1.0237,
+      "step": 30995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.348991358094589e-06,
+      "loss": 0.9458,
+      "step": 31000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.335630298859871e-06,
+      "loss": 1.0487,
+      "step": 31005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3222954933454285e-06,
+      "loss": 1.0234,
+      "step": 31010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.308986945172943e-06,
+      "loss": 0.9889,
+      "step": 31015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2957046579569883e-06,
+      "loss": 0.9772,
+      "step": 31020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.282448635304991e-06,
+      "loss": 0.9908,
+      "step": 31025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2692188808172376e-06,
+      "loss": 1.0165,
+      "step": 31030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.256015398086876e-06,
+      "loss": 1.0194,
+      "step": 31035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.242838190699948e-06,
+      "loss": 0.9882,
+      "step": 31040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2296872622353146e-06,
+      "loss": 0.952,
+      "step": 31045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.216562616264729e-06,
+      "loss": 1.0454,
+      "step": 31050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2034642563528285e-06,
+      "loss": 1.0446,
+      "step": 31055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1903921860570564e-06,
+      "loss": 1.058,
+      "step": 31060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1773464089277504e-06,
+      "loss": 1.019,
+      "step": 31065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.164326928508099e-06,
+      "loss": 0.9848,
+      "step": 31070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.151333748334162e-06,
+      "loss": 1.0357,
+      "step": 31075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.138366871934828e-06,
+      "loss": 1.0103,
+      "step": 31080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1254263028318576e-06,
+      "loss": 1.0003,
+      "step": 31085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1125120445398614e-06,
+      "loss": 1.0292,
+      "step": 31090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.099624100566345e-06,
+      "loss": 1.0118,
+      "step": 31095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0867624744115863e-06,
+      "loss": 1.0103,
+      "step": 31100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0739271695687688e-06,
+      "loss": 1.0523,
+      "step": 31105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0611181895239483e-06,
+      "loss": 0.9929,
+      "step": 31110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0483355377559864e-06,
+      "loss": 1.0506,
+      "step": 31115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0355792177365948e-06,
+      "loss": 0.9536,
+      "step": 31120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.022849232930336e-06,
+      "loss": 1.0199,
+      "step": 31125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.010145586794677e-06,
+      "loss": 1.088,
+      "step": 31130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.997468282779858e-06,
+      "loss": 1.0169,
+      "step": 31135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.984817324328981e-06,
+      "loss": 0.9757,
+      "step": 31140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9721927148780193e-06,
+      "loss": 0.9293,
+      "step": 31145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9595944578557855e-06,
+      "loss": 1.0413,
+      "step": 31150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9470225566839092e-06,
+      "loss": 1.0102,
+      "step": 31155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.934477014776871e-06,
+      "loss": 1.0021,
+      "step": 31160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9219578355419996e-06,
+      "loss": 0.9867,
+      "step": 31165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.909465022379487e-06,
+      "loss": 1.0863,
+      "step": 31170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8969985786823194e-06,
+      "loss": 1.0464,
+      "step": 31175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8845585078363326e-06,
+      "loss": 1.0009,
+      "step": 31180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8721448132202233e-06,
+      "loss": 1.0306,
+      "step": 31185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8597574982055176e-06,
+      "loss": 1.0737,
+      "step": 31190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8473965661565347e-06,
+      "loss": 1.0978,
+      "step": 31195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.835062020430501e-06,
+      "loss": 1.0448,
+      "step": 31200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8227538643774144e-06,
+      "loss": 1.0804,
+      "step": 31205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8104721013401447e-06,
+      "loss": 1.106,
+      "step": 31210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7982167346543576e-06,
+      "loss": 1.0801,
+      "step": 31215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7859877676486014e-06,
+      "loss": 1.0355,
+      "step": 31220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7737852036441857e-06,
+      "loss": 0.9861,
+      "step": 31225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7616090459553046e-06,
+      "loss": 1.0149,
+      "step": 31230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7494592978889676e-06,
+      "loss": 1.0767,
+      "step": 31235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.737335962744991e-06,
+      "loss": 1.0796,
+      "step": 31240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7252390438160523e-06,
+      "loss": 1.0218,
+      "step": 31245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.713168544387601e-06,
+      "loss": 0.9711,
+      "step": 31250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.701124467737981e-06,
+      "loss": 0.988,
+      "step": 31255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6891068171383093e-06,
+      "loss": 1.0307,
+      "step": 31260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6771155958525307e-06,
+      "loss": 0.9813,
+      "step": 31265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.665150807137429e-06,
+      "loss": 0.9782,
+      "step": 31270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.653212454242604e-06,
+      "loss": 0.9926,
+      "step": 31275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.641300540410474e-06,
+      "loss": 1.0102,
+      "step": 31280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.629415068876262e-06,
+      "loss": 0.9902,
+      "step": 31285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.617556042868019e-06,
+      "loss": 0.9674,
+      "step": 31290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6057234656066465e-06,
+      "loss": 0.9718,
+      "step": 31295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.593917340305818e-06,
+      "loss": 1.0333,
+      "step": 31300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5821376701720246e-06,
+      "loss": 0.9976,
+      "step": 31305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5703844584045846e-06,
+      "loss": 0.9923,
+      "step": 31310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5586577081956664e-06,
+      "loss": 1.0728,
+      "step": 31315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5469574227301782e-06,
+      "loss": 1.0178,
+      "step": 31320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5352836051858897e-06,
+      "loss": 1.0391,
+      "step": 31325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.523636258733375e-06,
+      "loss": 1.0578,
+      "step": 31330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.512015386536015e-06,
+      "loss": 1.0036,
+      "step": 31335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5004209917499852e-06,
+      "loss": 1.0508,
+      "step": 31340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4888530775243112e-06,
+      "loss": 1.0074,
+      "step": 31345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4773116470007797e-06,
+      "loss": 1.0386,
+      "step": 31350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.465796703314016e-06,
+      "loss": 1.0279,
+      "step": 31355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4543082495914303e-06,
+      "loss": 1.0601,
+      "step": 31360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.44284628895326e-06,
+      "loss": 1.0745,
+      "step": 31365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4314108245125147e-06,
+      "loss": 0.9753,
+      "step": 31370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.420001859375054e-06,
+      "loss": 1.0295,
+      "step": 31375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.408619396639489e-06,
+      "loss": 1.0434,
+      "step": 31380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.397263439397279e-06,
+      "loss": 0.9594,
+      "step": 31385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3859339907326582e-06,
+      "loss": 1.0566,
+      "step": 31390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3746310537226648e-06,
+      "loss": 0.9802,
+      "step": 31395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3633546314371422e-06,
+      "loss": 1.0097,
+      "step": 31400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.352104726938731e-06,
+      "loss": 1.0577,
+      "step": 31405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3408813432828523e-06,
+      "loss": 1.0559,
+      "step": 31410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.329684483517758e-06,
+      "loss": 1.0703,
+      "step": 31415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3185141506844697e-06,
+      "loss": 1.0315,
+      "step": 31420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3073703478168174e-06,
+      "loss": 1.0333,
+      "step": 31425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2962530779414126e-06,
+      "loss": 1.0154,
+      "step": 31430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2851623440776737e-06,
+      "loss": 1.0626,
+      "step": 31435
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2740981492378023e-06,
+      "loss": 1.1136,
+      "step": 31440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2630604964268163e-06,
+      "loss": 1.0165,
+      "step": 31445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2520493886424743e-06,
+      "loss": 0.9959,
+      "step": 31450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.241064828875372e-06,
+      "loss": 0.9569,
+      "step": 31455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2301068201088905e-06,
+      "loss": 0.994,
+      "step": 31460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2191753653191817e-06,
+      "loss": 1.0229,
+      "step": 31465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.208270467475171e-06,
+      "loss": 1.0079,
+      "step": 31470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1973921295386114e-06,
+      "loss": 1.0266,
+      "step": 31475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1865403544640286e-06,
+      "loss": 1.0276,
+      "step": 31480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.175715145198709e-06,
+      "loss": 1.0862,
+      "step": 31485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1649165046827792e-06,
+      "loss": 1.0098,
+      "step": 31490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1541444358490703e-06,
+      "loss": 1.0495,
+      "step": 31495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1433989416232646e-06,
+      "loss": 1.0589,
+      "step": 31500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1326800249237943e-06,
+      "loss": 1.0129,
+      "step": 31505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.121987688661886e-06,
+      "loss": 1.0575,
+      "step": 31510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1113219357415393e-06,
+      "loss": 1.0522,
+      "step": 31515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.100682769059548e-06,
+      "loss": 1.0435,
+      "step": 31520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0900701915054666e-06,
+      "loss": 1.0267,
+      "step": 31525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0794842059616237e-06,
+      "loss": 1.0417,
+      "step": 31530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.068924815303153e-06,
+      "loss": 1.0006,
+      "step": 31535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.058392022397937e-06,
+      "loss": 1.0368,
+      "step": 31540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0478858301066774e-06,
+      "loss": 0.9859,
+      "step": 31545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.037406241282791e-06,
+      "loss": 0.9958,
+      "step": 31550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0269532587724905e-06,
+      "loss": 0.9624,
+      "step": 31555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0165268854147934e-06,
+      "loss": 1.0006,
+      "step": 31560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.006127124041479e-06,
+      "loss": 1.0349,
+      "step": 31565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.995753977477066e-06,
+      "loss": 1.076,
+      "step": 31570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.985407448538856e-06,
+      "loss": 1.0391,
+      "step": 31575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9750875400369574e-06,
+      "loss": 1.0307,
+      "step": 31580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.964794254774216e-06,
+      "loss": 1.0838,
+      "step": 31585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9545275955462515e-06,
+      "loss": 1.0399,
+      "step": 31590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.944287565141445e-06,
+      "loss": 0.9835,
+      "step": 31595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.93407416634096e-06,
+      "loss": 1.0287,
+      "step": 31600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9238874019187337e-06,
+      "loss": 1.0004,
+      "step": 31605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9137272746414526e-06,
+      "loss": 1.0125,
+      "step": 31610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9035937872685428e-06,
+      "loss": 1.052,
+      "step": 31615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8934869425522805e-06,
+      "loss": 1.0528,
+      "step": 31620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8834067432376146e-06,
+      "loss": 0.9839,
+      "step": 31625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8733531920622882e-06,
+      "loss": 1.0657,
+      "step": 31630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8633262917568284e-06,
+      "loss": 1.0155,
+      "step": 31635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8533260450445233e-06,
+      "loss": 1.0006,
+      "step": 31640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8433524546413783e-06,
+      "loss": 0.9867,
+      "step": 31645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.833405523256193e-06,
+      "loss": 1.0397,
+      "step": 31650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8234852535905289e-06,
+      "loss": 1.0558,
+      "step": 31655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8135916483386862e-06,
+      "loss": 1.1248,
+      "step": 31660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8037247101877485e-06,
+      "loss": 1.0147,
+      "step": 31665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7938844418175283e-06,
+      "loss": 1.0625,
+      "step": 31670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7840708459006317e-06,
+      "loss": 1.0189,
+      "step": 31675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7742839251023825e-06,
+      "loss": 1.0023,
+      "step": 31680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7645236820808763e-06,
+      "loss": 1.0383,
+      "step": 31685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7547901194869708e-06,
+      "loss": 1.075,
+      "step": 31690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7450832399642624e-06,
+      "loss": 1.0485,
+      "step": 31695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7354030461490977e-06,
+      "loss": 1.0722,
+      "step": 31700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.725749540670596e-06,
+      "loss": 0.9609,
+      "step": 31705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7161227261506262e-06,
+      "loss": 1.0714,
+      "step": 31710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7065226052037974e-06,
+      "loss": 1.0413,
+      "step": 31715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.696949180437446e-06,
+      "loss": 1.0232,
+      "step": 31720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6874024544517031e-06,
+      "loss": 1.0165,
+      "step": 31725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.677882429839428e-06,
+      "loss": 1.0173,
+      "step": 31730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6683891091862303e-06,
+      "loss": 0.9346,
+      "step": 31735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6589224950704364e-06,
+      "loss": 1.0324,
+      "step": 31740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6494825900631783e-06,
+      "loss": 0.9628,
+      "step": 31745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6400693967282943e-06,
+      "loss": 1.0238,
+      "step": 31750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6306829176223836e-06,
+      "loss": 1.057,
+      "step": 31755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6213231552947407e-06,
+      "loss": 1.0534,
+      "step": 31760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6119901122875091e-06,
+      "loss": 1.0536,
+      "step": 31765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602683791135473e-06,
+      "loss": 1.0999,
+      "step": 31770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5934041943661993e-06,
+      "loss": 1.0266,
+      "step": 31775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.584151324500005e-06,
+      "loss": 1.0348,
+      "step": 31780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5749251840499468e-06,
+      "loss": 1.0146,
+      "step": 31785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.565725775521798e-06,
+      "loss": 0.9694,
+      "step": 31790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5565531014140933e-06,
+      "loss": 0.9885,
+      "step": 31795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5474071642180955e-06,
+      "loss": 1.036,
+      "step": 31800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5382879664178173e-06,
+      "loss": 0.9918,
+      "step": 31805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5291955104900002e-06,
+      "loss": 1.0029,
+      "step": 31810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5201297989041242e-06,
+      "loss": 1.0074,
+      "step": 31815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5110908341223973e-06,
+      "loss": 0.9416,
+      "step": 31820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5020786185997892e-06,
+      "loss": 0.9935,
+      "step": 31825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4930931547839755e-06,
+      "loss": 1.0025,
+      "step": 31830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4841344451153926e-06,
+      "loss": 1.0473,
+      "step": 31835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4752024920271834e-06,
+      "loss": 0.9244,
+      "step": 31840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4662972979452406e-06,
+      "loss": 1.0574,
+      "step": 31845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4574188652881737e-06,
+      "loss": 0.9878,
+      "step": 31850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4485671964673542e-06,
+      "loss": 1.031,
+      "step": 31855
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4397422938868587e-06,
+      "loss": 1.0016,
+      "step": 31860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4309441599434926e-06,
+      "loss": 1.031,
+      "step": 31865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4221727970267995e-06,
+      "loss": 1.1121,
+      "step": 31870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.413428207519074e-06,
+      "loss": 1.0544,
+      "step": 31875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4047103937953055e-06,
+      "loss": 1.0319,
+      "step": 31880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3960193582231995e-06,
+      "loss": 1.0342,
+      "step": 31885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.387355103163257e-06,
+      "loss": 1.0571,
+      "step": 31890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.378717630968629e-06,
+      "loss": 1.0248,
+      "step": 31895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3701069439852388e-06,
+      "loss": 1.058,
+      "step": 31900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3615230445516935e-06,
+      "loss": 1.0246,
+      "step": 31905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.352965934999384e-06,
+      "loss": 1.0643,
+      "step": 31910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3444356176523843e-06,
+      "loss": 1.0222,
+      "step": 31915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3359320948274967e-06,
+      "loss": 1.0451,
+      "step": 31920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3274553688342405e-06,
+      "loss": 0.9478,
+      "step": 31925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3190054419748853e-06,
+      "loss": 1.0766,
+      "step": 31930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3105823165443843e-06,
+      "loss": 1.0978,
+      "step": 31935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3021859948304404e-06,
+      "loss": 1.0888,
+      "step": 31940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2938164791134632e-06,
+      "loss": 1.0472,
+      "step": 31945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2854737716665898e-06,
+      "loss": 1.0702,
+      "step": 31950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2771578747556744e-06,
+      "loss": 1.071,
+      "step": 31955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.268868790639277e-06,
+      "loss": 1.0957,
+      "step": 31960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2606065215686857e-06,
+      "loss": 1.0649,
+      "step": 31965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2523710697879165e-06,
+      "loss": 1.0995,
+      "step": 31970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2441624375336692e-06,
+      "loss": 1.016,
+      "step": 31975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2359806270354157e-06,
+      "loss": 0.9451,
+      "step": 31980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2278256405152677e-06,
+      "loss": 0.9604,
+      "step": 31985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2196974801881312e-06,
+      "loss": 1.1377,
+      "step": 31990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2115961482615513e-06,
+      "loss": 0.9887,
+      "step": 31995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2035216469358456e-06,
+      "loss": 1.047,
+      "step": 32000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.195473978404016e-06,
+      "loss": 0.9851,
+      "step": 32005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.187453144851791e-06,
+      "loss": 1.0507,
+      "step": 32010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1794591484575735e-06,
+      "loss": 1.0181,
+      "step": 32015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.171491991392548e-06,
+      "loss": 1.0494,
+      "step": 32020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.163551675820529e-06,
+      "loss": 1.08,
+      "step": 32025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1556382038981018e-06,
+      "loss": 1.0333,
+      "step": 32030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1477515777745363e-06,
+      "loss": 1.0454,
+      "step": 32035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.139891799591819e-06,
+      "loss": 1.0206,
+      "step": 32040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1320588714846203e-06,
+      "loss": 1.0503,
+      "step": 32045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1242527955803494e-06,
+      "loss": 1.0414,
+      "step": 32050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1164735739991217e-06,
+      "loss": 1.039,
+      "step": 32055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1087212088537357e-06,
+      "loss": 1.0918,
+      "step": 32060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1009957022496964e-06,
+      "loss": 0.9735,
+      "step": 32065
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0932970562852363e-06,
+      "loss": 1.0426,
+      "step": 32070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.085625273051294e-06,
+      "loss": 1.0185,
+      "step": 32075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0779803546314914e-06,
+      "loss": 1.0293,
+      "step": 32080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.070362303102146e-06,
+      "loss": 1.0111,
+      "step": 32085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0627711205323243e-06,
+      "loss": 0.9563,
+      "step": 32090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.055206808983744e-06,
+      "loss": 0.9391,
+      "step": 32095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.047669370510862e-06,
+      "loss": 1.0268,
+      "step": 32100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.040158807160807e-06,
+      "loss": 0.9797,
+      "step": 32105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0326751209734253e-06,
+      "loss": 1.1068,
+      "step": 32110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0252183139812798e-06,
+      "loss": 1.0893,
+      "step": 32115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0177883882095952e-06,
+      "loss": 1.1198,
+      "step": 32120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0103853456763234e-06,
+      "loss": 1.1513,
+      "step": 32125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0030091883920901e-06,
+      "loss": 1.0886,
+      "step": 32130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.95659918360259e-07,
+      "loss": 1.0344,
+      "step": 32135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.883375375768555e-07,
+      "loss": 1.0995,
+      "step": 32140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.810420480306114e-07,
+      "loss": 1.0451,
+      "step": 32145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.73773451702964e-07,
+      "loss": 0.993,
+      "step": 32150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.665317505680338e-07,
+      "loss": 1.0514,
+      "step": 32155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.593169465926367e-07,
+      "loss": 1.0695,
+      "step": 32160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.521290417363049e-07,
+      "loss": 1.0261,
+      "step": 32165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.449680379512438e-07,
+      "loss": 1.0046,
+      "step": 32170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.37833937182342e-07,
+      "loss": 1.0536,
+      "step": 32175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.307267413672272e-07,
+      "loss": 0.9976,
+      "step": 32180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.236464524361777e-07,
+      "loss": 0.983,
+      "step": 32185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.165930723121663e-07,
+      "loss": 1.0096,
+      "step": 32190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.095666029108718e-07,
+      "loss": 1.0738,
+      "step": 32195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.025670461406677e-07,
+      "loss": 1.0978,
+      "step": 32200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.955944039025999e-07,
+      "loss": 1.0489,
+      "step": 32205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.886486780904202e-07,
+      "loss": 1.013,
+      "step": 32210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.817298705905641e-07,
+      "loss": 1.085,
+      "step": 32215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.748379832821618e-07,
+      "loss": 1.0305,
+      "step": 32220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.679730180370271e-07,
+      "loss": 1.0548,
+      "step": 32225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.611349767196464e-07,
+      "loss": 0.9744,
+      "step": 32230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.543238611872117e-07,
+      "loss": 1.017,
+      "step": 32235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.475396732896213e-07,
+      "loss": 1.0419,
+      "step": 32240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.407824148694232e-07,
+      "loss": 1.0073,
+      "step": 32245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.340520877618496e-07,
+      "loss": 1.0156,
+      "step": 32250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.273486937948715e-07,
+      "loss": 1.0191,
+      "step": 32255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.206722347890772e-07,
+      "loss": 1.0857,
+      "step": 32260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.140227125577826e-07,
+      "loss": 0.9939,
+      "step": 32265
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.074001289069765e-07,
+      "loss": 1.0154,
+      "step": 32270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.008044856353313e-07,
+      "loss": 1.035,
+      "step": 32275
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.94235784534203e-07,
+      "loss": 1.0048,
+      "step": 32280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.876940273876199e-07,
+      "loss": 1.1056,
+      "step": 32285
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.811792159723052e-07,
+      "loss": 1.1132,
+      "step": 32290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.746913520576659e-07,
+      "loss": 1.0387,
+      "step": 32295
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.682304374057703e-07,
+      "loss": 1.0302,
+      "step": 32300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.617964737714034e-07,
+      "loss": 0.9582,
+      "step": 32305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.553894629019897e-07,
+      "loss": 1.0565,
+      "step": 32310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.490094065376596e-07,
+      "loss": 0.9845,
+      "step": 32315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.426563064112046e-07,
+      "loss": 0.994,
+      "step": 32320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.363301642481224e-07,
+      "loss": 1.074,
+      "step": 32325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.300309817665718e-07,
+      "loss": 1.0267,
+      "step": 32330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.237587606773733e-07,
+      "loss": 1.0613,
+      "step": 32335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.175135026840419e-07,
+      "loss": 1.0581,
+      "step": 32340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.112952094827985e-07,
+      "loss": 1.02,
+      "step": 32345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.051038827624701e-07,
+      "loss": 1.0211,
+      "step": 32350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.989395242046226e-07,
+      "loss": 1.0552,
+      "step": 32355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.928021354834724e-07,
+      "loss": 1.0421,
+      "step": 32360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.866917182659194e-07,
+      "loss": 1.052,
+      "step": 32365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.80608274211536e-07,
+      "loss": 1.0149,
+      "step": 32370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.745518049725342e-07,
+      "loss": 0.9806,
+      "step": 32375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.685223121938644e-07,
+      "loss": 1.025,
+      "step": 32380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.625197975131059e-07,
+      "loss": 0.9931,
+      "step": 32385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.565442625605211e-07,
+      "loss": 1.0155,
+      "step": 32390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.505957089590342e-07,
+      "loss": 0.9602,
+      "step": 32395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.446741383242638e-07,
+      "loss": 0.9849,
+      "step": 32400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.387795522645013e-07,
+      "loss": 1.0305,
+      "step": 32405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.32911952380677e-07,
+      "loss": 0.9896,
+      "step": 32410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.270713402664163e-07,
+      "loss": 1.0623,
+      "step": 32415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.212577175080169e-07,
+      "loss": 1.0462,
+      "step": 32420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.15471085684427e-07,
+      "loss": 0.9894,
+      "step": 32425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.097114463672892e-07,
+      "loss": 1.0602,
+      "step": 32430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.039788011208858e-07,
+      "loss": 0.9752,
+      "step": 32435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.982731515022044e-07,
+      "loss": 1.0304,
+      "step": 32440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.925944990608722e-07,
+      "loss": 0.9845,
+      "step": 32445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.86942845339189e-07,
+      "loss": 0.9608,
+      "step": 32450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.813181918721267e-07,
+      "loss": 0.9767,
+      "step": 32455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.757205401873189e-07,
+      "loss": 1.0355,
+      "step": 32460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.701498918050719e-07,
+      "loss": 1.0736,
+      "step": 32465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.64606248238364e-07,
+      "loss": 0.9918,
+      "step": 32470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.590896109928245e-07,
+      "loss": 0.9919,
+      "step": 32475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.535999815667436e-07,
+      "loss": 1.0201,
+      "step": 32480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.481373614510954e-07,
+      "loss": 0.9953,
+      "step": 32485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.427017521295153e-07,
+      "loss": 0.9967,
+      "step": 32490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.372931550782889e-07,
+      "loss": 1.063,
+      "step": 32495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.319115717663636e-07,
+      "loss": 1.0328,
+      "step": 32500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.265570036553813e-07,
+      "loss": 1.1129,
+      "step": 32505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.21229452199612e-07,
+      "loss": 1.0175,
+      "step": 32510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.159289188459982e-07,
+      "loss": 1.0648,
+      "step": 32515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.10655405034155e-07,
+      "loss": 0.9695,
+      "step": 32520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.054089121963368e-07,
+      "loss": 1.0289,
+      "step": 32525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.001894417574926e-07,
+      "loss": 0.9954,
+      "step": 32530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.949969951352107e-07,
+      "loss": 1.0295,
+      "step": 32535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.898315737397186e-07,
+      "loss": 1.0741,
+      "step": 32540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.846931789739606e-07,
+      "loss": 1.0239,
+      "step": 32545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.795818122334872e-07,
+      "loss": 1.0379,
+      "step": 32550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.744974749065323e-07,
+      "loss": 1.0666,
+      "step": 32555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.694401683739913e-07,
+      "loss": 1.0056,
+      "step": 32560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6440989400939884e-07,
+      "loss": 1.1186,
+      "step": 32565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5940665317897313e-07,
+      "loss": 1.0841,
+      "step": 32570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.544304472415717e-07,
+      "loss": 0.9992,
+      "step": 32575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.494812775487134e-07,
+      "loss": 1.0727,
+      "step": 32580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.445591454445897e-07,
+      "loss": 0.9975,
+      "step": 32585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3966405226602006e-07,
+      "loss": 1.0705,
+      "step": 32590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.347959993424966e-07,
+      "loss": 1.0233,
+      "step": 32595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2995498799617284e-07,
+      "loss": 1.0065,
+      "step": 32600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.251410195418526e-07,
+      "loss": 1.1057,
+      "step": 32605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.203540952869789e-07,
+      "loss": 0.9315,
+      "step": 32610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.155942165316784e-07,
+      "loss": 0.9815,
+      "step": 32615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1086138456871704e-07,
+      "loss": 1.0068,
+      "step": 32620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.061556006835221e-07,
+      "loss": 0.9568,
+      "step": 32625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.01476866154149e-07,
+      "loss": 0.9929,
+      "step": 32630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.96825182251348e-07,
+      "loss": 1.0437,
+      "step": 32635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.922005502384862e-07,
+      "loss": 0.9601,
+      "step": 32640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.876029713716034e-07,
+      "loss": 1.0354,
+      "step": 32645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.830324468993896e-07,
+      "loss": 1.0116,
+      "step": 32650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7848897806317394e-07,
+      "loss": 1.0049,
+      "step": 32655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7397256609695843e-07,
+      "loss": 1.0258,
+      "step": 32660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6948321222738393e-07,
+      "loss": 1.0545,
+      "step": 32665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.650209176737418e-07,
+      "loss": 0.9421,
+      "step": 32670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6058568364797373e-07,
+      "loss": 0.9724,
+      "step": 32675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.561775113546828e-07,
+      "loss": 0.9906,
+      "step": 32680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.517964019910891e-07,
+      "loss": 1.0109,
+      "step": 32685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4744235674711855e-07,
+      "loss": 1.0737,
+      "step": 32690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.431153768053141e-07,
+      "loss": 1.1039,
+      "step": 32695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.388154633408358e-07,
+      "loss": 1.0549,
+      "step": 32700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.345426175215605e-07,
+      "loss": 0.9412,
+      "step": 32705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3029684050797096e-07,
+      "loss": 1.1049,
+      "step": 32710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.260781334531893e-07,
+      "loss": 1.008,
+      "step": 32715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2188649750302113e-07,
+      "loss": 1.0106,
+      "step": 32720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.177219337958892e-07,
+      "loss": 1.0408,
+      "step": 32725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1358444346288874e-07,
+      "loss": 1.0594,
+      "step": 32730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.094740276277319e-07,
+      "loss": 1.024,
+      "step": 32735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.053906874068035e-07,
+      "loss": 1.0445,
+      "step": 32740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0133442390912757e-07,
+      "loss": 1.0856,
+      "step": 32745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.973052382363561e-07,
+      "loss": 1.0289,
+      "step": 32750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9330313148282497e-07,
+      "loss": 1.0559,
+      "step": 32755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8932810473547565e-07,
+      "loss": 1.0297,
+      "step": 32760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8538015907392244e-07,
+      "loss": 1.0459,
+      "step": 32765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.814592955704076e-07,
+      "loss": 0.9938,
+      "step": 32770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7756551528982376e-07,
+      "loss": 1.031,
+      "step": 32775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.736988192897139e-07,
+      "loss": 0.9878,
+      "step": 32780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6985920862026004e-07,
+      "loss": 1.0246,
+      "step": 32785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.660466843242837e-07,
+      "loss": 1.0699,
+      "step": 32790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6226124743724544e-07,
+      "loss": 1.0789,
+      "step": 32795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5850289898727844e-07,
+      "loss": 0.9432,
+      "step": 32800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5477163999513275e-07,
+      "loss": 1.006,
+      "step": 32805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5106747147418675e-07,
+      "loss": 1.119,
+      "step": 32810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4739039443049115e-07,
+      "loss": 1.0462,
+      "step": 32815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4374040986274714e-07,
+      "loss": 1.0925,
+      "step": 32820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4011751876223955e-07,
+      "loss": 1.0687,
+      "step": 32825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3652172211297008e-07,
+      "loss": 1.0658,
+      "step": 32830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3295302089152425e-07,
+      "loss": 0.9806,
+      "step": 32835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.29411416067149e-07,
+      "loss": 0.9645,
+      "step": 32840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.258969086017415e-07,
+      "loss": 1.0585,
+      "step": 32845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2240949944982714e-07,
+      "loss": 1.0871,
+      "step": 32850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1894918955858158e-07,
+      "loss": 1.0287,
+      "step": 32855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1551597986780859e-07,
+      "loss": 0.9769,
+      "step": 32860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.121098713099401e-07,
+      "loss": 1.0395,
+      "step": 32865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0873086481008053e-07,
+      "loss": 1.059,
+      "step": 32870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0537896128596245e-07,
+      "loss": 1.102,
+      "step": 32875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0205416164794655e-07,
+      "loss": 0.9808,
+      "step": 32880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.987564667990216e-07,
+      "loss": 1.0021,
+      "step": 32885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9548587763484893e-07,
+      "loss": 0.9878,
+      "step": 32890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9224239504370689e-07,
+      "loss": 1.0223,
+      "step": 32895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8902601990650192e-07,
+      "loss": 1.0479,
+      "step": 32900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8583675309681304e-07,
+      "loss": 1.0421,
+      "step": 32905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8267459548081402e-07,
+      "loss": 1.0738,
+      "step": 32910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7953954791734007e-07,
+      "loss": 1.0262,
+      "step": 32915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7643161125787677e-07,
+      "loss": 1.1363,
+      "step": 32920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7335078634650447e-07,
+      "loss": 0.9707,
+      "step": 32925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7029707401997606e-07,
+      "loss": 1.0555,
+      "step": 32930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6727047510768369e-07,
+      "loss": 1.0597,
+      "step": 32935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.642709904316142e-07,
+      "loss": 1.0569,
+      "step": 32940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.612986208064382e-07,
+      "loss": 1.0853,
+      "step": 32945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5835336703943216e-07,
+      "loss": 1.1017,
+      "step": 32950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5543522993052284e-07,
+      "loss": 1.0974,
+      "step": 32955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5254421027225408e-07,
+      "loss": 0.9823,
+      "step": 32960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.496803088498422e-07,
+      "loss": 1.0301,
+      "step": 32965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4684352644108723e-07,
+      "loss": 1.0211,
+      "step": 32970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4403386381646177e-07,
+      "loss": 1.0065,
+      "step": 32975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.412513217390554e-07,
+      "loss": 1.0621,
+      "step": 32980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3849590096460806e-07,
+      "loss": 1.1379,
+      "step": 32985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3576760224147667e-07,
+      "loss": 1.0065,
+      "step": 32990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3306642631064625e-07,
+      "loss": 1.0494,
+      "step": 32995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.303923739057633e-07,
+      "loss": 1.0373,
+      "step": 33000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2774544575309134e-07,
+      "loss": 1.0533,
+      "step": 33005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2512564257152193e-07,
+      "loss": 0.953,
+      "step": 33010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2253296507257482e-07,
+      "loss": 0.9525,
+      "step": 33015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.199674139604312e-07,
+      "loss": 1.0577,
+      "step": 33020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1742898993187812e-07,
+      "loss": 1.1482,
+      "step": 33025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1491769367634187e-07,
+      "loss": 1.1678,
+      "step": 33030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1243352587588796e-07,
+      "loss": 1.0416,
+      "step": 33035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0997648720519893e-07,
+      "loss": 1.0466,
+      "step": 33040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0754657833159653e-07,
+      "loss": 1.0364,
+      "step": 33045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0514379991504176e-07,
+      "loss": 0.9948,
+      "step": 33050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0276815260813478e-07,
+      "loss": 1.0748,
+      "step": 33055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.004196370560706e-07,
+      "loss": 1.084,
+      "step": 33060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.809825389670569e-08,
+      "loss": 1.0359,
+      "step": 33065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.580400376052456e-08,
+      "loss": 1.0372,
+      "step": 33070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.353688727063992e-08,
+      "loss": 1.0734,
+      "step": 33075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.129690504279254e-08,
+      "loss": 1.0507,
+      "step": 33080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.908405768534022e-08,
+      "loss": 1.0421,
+      "step": 33085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.689834579929112e-08,
+      "loss": 1.1419,
+      "step": 33090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.473976997829259e-08,
+      "loss": 1.0433,
+      "step": 33095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.260833080858677e-08,
+      "loss": 1.0026,
+      "step": 33100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.050402886907726e-08,
+      "loss": 1.0695,
+      "step": 33105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.842686473128469e-08,
+      "loss": 1.0675,
+      "step": 33110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.637683895934666e-08,
+      "loss": 1.0791,
+      "step": 33115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.435395211006225e-08,
+      "loss": 1.024,
+      "step": 33120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.235820473282529e-08,
+      "loss": 1.0324,
+      "step": 33125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.038959736969108e-08,
+      "loss": 1.0327,
+      "step": 33130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.844813055530974e-08,
+      "loss": 0.9654,
+      "step": 33135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.653380481698168e-08,
+      "loss": 1.0353,
+      "step": 33140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.464662067464656e-08,
+      "loss": 0.9622,
+      "step": 33145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.278657864083881e-08,
+      "loss": 1.0401,
+      "step": 33150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.095367922074325e-08,
+      "loss": 1.0605,
+      "step": 33155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9147922912161644e-08,
+      "loss": 0.999,
+      "step": 33160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.736931020555725e-08,
+      "loss": 0.9496,
+      "step": 33165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.561784158397698e-08,
+      "loss": 1.0708,
+      "step": 33170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.389351752310701e-08,
+      "loss": 1.0351,
+      "step": 33175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.21963384912838e-08,
+      "loss": 1.0477,
+      "step": 33180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.052630494944977e-08,
+      "loss": 0.9883,
+      "step": 33185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.888341735117541e-08,
+      "loss": 1.0154,
+      "step": 33190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.726767614267047e-08,
+      "loss": 0.9743,
+      "step": 33195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5679081762761697e-08,
+      "loss": 1.0818,
+      "step": 33200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.411763464290397e-08,
+      "loss": 0.9783,
+      "step": 33205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.258333520718028e-08,
+      "loss": 1.0376,
+      "step": 33210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.107618387231282e-08,
+      "loss": 1.0127,
+      "step": 33215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.959618104762974e-08,
+      "loss": 1.0251,
+      "step": 33220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8143327135087284e-08,
+      "loss": 1.0233,
+      "step": 33225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6717622529292006e-08,
+      "loss": 0.9605,
+      "step": 33230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.53190676174453e-08,
+      "loss": 1.0516,
+      "step": 33235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.394766277940997e-08,
+      "loss": 0.9519,
+      "step": 33240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.260340838763254e-08,
+      "loss": 0.9851,
+      "step": 33245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.128630480722095e-08,
+      "loss": 1.0326,
+      "step": 33250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.999635239588905e-08,
+      "loss": 0.9407,
+      "step": 33255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.873355150400103e-08,
+      "loss": 1.0578,
+      "step": 33260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7497902474504788e-08,
+      "loss": 0.9887,
+      "step": 33265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6289405643009635e-08,
+      "loss": 0.9677,
+      "step": 33270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5108061337741906e-08,
+      "loss": 1.0865,
+      "step": 33275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3953869879544956e-08,
+      "loss": 1.0035,
+      "step": 33280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2826831581901353e-08,
+      "loss": 1.0245,
+      "step": 33285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.172694675089959e-08,
+      "loss": 1.0148,
+      "step": 33290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0654215685278477e-08,
+      "loss": 1.0956,
+      "step": 33295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9608638676371638e-08,
+      "loss": 1.0079,
+      "step": 33300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8590216008174122e-08,
+      "loss": 1.0518,
+      "step": 33305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7598947957264687e-08,
+      "loss": 1.0643,
+      "step": 33310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6634834792894628e-08,
+      "loss": 1.0189,
+      "step": 33315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5697876776887832e-08,
+      "loss": 1.1239,
+      "step": 33320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4788074163729626e-08,
+      "loss": 1.0433,
+      "step": 33325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3905427200522348e-08,
+      "loss": 1.0062,
+      "step": 33330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3049936126985352e-08,
+      "loss": 1.0362,
+      "step": 33335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2221601175466113e-08,
+      "loss": 0.9938,
+      "step": 33340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1420422570951327e-08,
+      "loss": 1.0636,
+      "step": 33345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0646400531022504e-08,
+      "loss": 1.1497,
+      "step": 33350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.899535265900373e-09,
+      "loss": 1.0251,
+      "step": 33355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.179826978444884e-09,
+      "loss": 1.0716,
+      "step": 33360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.4872758641108e-09,
+      "loss": 0.9776,
+      "step": 33365
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.821882111003209e-09,
+      "loss": 0.9675,
+      "step": 33370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.183645899844216e-09,
+      "loss": 1.0777,
+      "step": 33375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.5725674039618376e-09,
+      "loss": 1.0456,
+      "step": 33380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.988646789334418e-09,
+      "loss": 0.9625,
+      "step": 33385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.431884214557315e-09,
+      "loss": 1.0344,
+      "step": 33390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.902279830831802e-09,
+      "loss": 1.0507,
+      "step": 33395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.399833781998375e-09,
+      "loss": 1.0283,
+      "step": 33400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9245462045367495e-09,
+      "loss": 0.9996,
+      "step": 33405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4764172275103536e-09,
+      "loss": 1.0546,
+      "step": 33410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0554469726440383e-09,
+      "loss": 1.0001,
+      "step": 33415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.661635554268571e-09,
+      "loss": 1.0731,
+      "step": 33420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2949830793428384e-09,
+      "loss": 1.0345,
+      "step": 33425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.955489647453845e-09,
+      "loss": 1.0085,
+      "step": 33430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6431553507834096e-09,
+      "loss": 0.9439,
+      "step": 33435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3579802741858772e-09,
+      "loss": 1.0171,
+      "step": 33440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.099964495110406e-09,
+      "loss": 1.0582,
+      "step": 33445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.691080836231713e-10,
+      "loss": 1.0641,
+      "step": 33450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.654111024295695e-10,
+      "loss": 0.9505,
+      "step": 33455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.888736068520139e-10,
+      "loss": 1.0263,
+      "step": 33460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3949564482993467e-10,
+      "loss": 0.9571,
+      "step": 33465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1727725695308566e-10,
+      "loss": 0.9431,
+      "step": 33470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2221847639493078e-10,
+      "loss": 1.0255,
+      "step": 33475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.431932899035985e-11,
+      "loss": 1.0394,
+      "step": 33480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3579833169075073e-11,
+      "loss": 1.0162,
+      "step": 33485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.9852,
+      "step": 33490
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.0256932973861694,
+      "eval_runtime": 2613.5895,
+      "eval_samples_per_second": 11.317,
+      "eval_steps_per_second": 1.415,
+      "step": 33490
+    },
+    {
+      "epoch": 1.0,
+      "step": 33490,
+      "total_flos": 6.646586827132109e+16,
+      "train_loss": 1.0557984271025294,
+      "train_runtime": 103963.1806,
+      "train_samples_per_second": 2.577,
+      "train_steps_per_second": 0.322
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 33490,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 6.646586827132109e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}