diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,21950 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 18257,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999837529454,
+      "loss": 1.1032,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999995938236562,
+      "loss": 1.0664,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999983752949535,
+      "loss": 1.0333,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999963444148814,
+      "loss": 1.1211,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999935011850908,
+      "loss": 1.1523,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999989845607891,
+      "loss": 1.0734,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999853776862518,
+      "loss": 1.1928,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999800974238022,
+      "loss": 1.1473,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999740048248324,
+      "loss": 1.1761,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999670998942912,
+      "loss": 1.036,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999593826377883,
+      "loss": 1.1627,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999950853061592,
+      "loss": 1.086,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999941511172632,
+      "loss": 1.0676,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999313569784967,
+      "loss": 1.1365,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999203904874353,
+      "loss": 1.1211,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999908611708357,
+      "loss": 1.043,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019998960206508287,
+      "loss": 1.0945,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019998826173250802,
+      "loss": 1.129,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019998684017419997,
+      "loss": 1.1267,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998533739131342,
+      "loss": 1.1279,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998375338506927,
+      "loss": 1.1077,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998208815675427,
+      "loss": 1.0842,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999803417077211,
+      "loss": 1.1657,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999785140393886,
+      "loss": 1.0209,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999766051532414,
+      "loss": 1.1343,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997461505083023,
+      "loss": 1.1039,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999725437337717,
+      "loss": 1.1007,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999703912037485,
+      "loss": 1.1289,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996815746250925,
+      "loss": 1.1169,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996584251186846,
+      "loss": 1.1366,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996344635370684,
+      "loss": 1.0876,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996096898997078,
+      "loss": 1.1273,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995841042267284,
+      "loss": 1.1308,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999557706538914,
+      "loss": 1.1608,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995304968577093,
+      "loss": 1.2318,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995024752052186,
+      "loss": 1.1642,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994736416042048,
+      "loss": 1.1404,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999443996078091,
+      "loss": 1.0635,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994135386509603,
+      "loss": 1.1754,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993822693475542,
+      "loss": 1.1087,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993501881932746,
+      "loss": 1.0764,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999317295214183,
+      "loss": 1.0868,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992835904369998,
+      "loss": 1.0977,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992490738891057,
+      "loss": 1.1653,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992137455985396,
+      "loss": 1.1657,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999177605594001,
+      "loss": 1.1186,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999140653904848,
+      "loss": 1.1123,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991028905610986,
+      "loss": 1.1111,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199906431559343,
+      "loss": 1.1037,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990249290331784,
+      "loss": 1.1314,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019989847309123397,
+      "loss": 1.0606,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019989437212635693,
+      "loss": 1.0934,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019989019001201815,
+      "loss": 1.0947,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019988592675161492,
+      "loss": 1.0364,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019988158234861058,
+      "loss": 1.1504,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998771568065343,
+      "loss": 1.2071,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998726501289811,
+      "loss": 1.1213,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986806231961208,
+      "loss": 1.1015,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986339338215416,
+      "loss": 1.0587,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985864332040012,
+      "loss": 1.0896,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985381213820876,
+      "loss": 1.2225,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998488998395046,
+      "loss": 1.1455,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998439064282782,
+      "loss": 1.1164,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983883190858602,
+      "loss": 1.1763,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998336762845503,
+      "loss": 1.1246,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998284395603592,
+      "loss": 1.091,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998231217402668,
+      "loss": 1.1619,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981772282859315,
+      "loss": 1.1208,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199812242829724,
+      "loss": 1.1494,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980668174811103,
+      "loss": 1.0512,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998010395882718,
+      "loss": 1.0742,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979531635478978,
+      "loss": 1.1415,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997895120523142,
+      "loss": 1.0945,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997836266855602,
+      "loss": 1.1632,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019977766025930878,
+      "loss": 1.0635,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997716127784068,
+      "loss": 1.0715,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997654842477669,
+      "loss": 1.0331,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997592746723677,
+      "loss": 1.0083,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997529840572535,
+      "loss": 1.0541,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974661240753448,
+      "loss": 1.1628,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974015972838678,
+      "loss": 1.1128,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019973362602505212,
+      "loss": 1.0925,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019972701130283822,
+      "loss": 1.1262,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019972031556711862,
+      "loss": 1.1006,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997135388233325,
+      "loss": 1.0861,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997066810769851,
+      "loss": 1.1397,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019969974233364723,
+      "loss": 1.1535,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019969272259895565,
+      "loss": 1.1099,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019968562187861277,
+      "loss": 1.1154,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199678440178387,
+      "loss": 1.0709,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019967117750411233,
+      "loss": 1.1128,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019966383386168867,
+      "loss": 1.134,
+      "step": 455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019965640925708158,
+      "loss": 1.0992,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019964890369632255,
+      "loss": 1.0872,
+      "step": 465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019964131718550873,
+      "loss": 1.0695,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199633649730803,
+      "loss": 1.1137,
+      "step": 475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019962590133843397,
+      "loss": 1.1437,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996180720146962,
+      "loss": 1.0785,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019961016176594978,
+      "loss": 1.0316,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019960217059862063,
+      "loss": 1.1335,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019959409851920043,
+      "loss": 1.0676,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019958594553424658,
+      "loss": 1.0861,
+      "step": 505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995777116503821,
+      "loss": 1.1432,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019956939687429582,
+      "loss": 1.0515,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019956100121274232,
+      "loss": 1.1532,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995525246725418,
+      "loss": 1.1023,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995439672605802,
+      "loss": 1.0932,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995353289838092,
+      "loss": 1.1421,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019952660984924608,
+      "loss": 1.2585,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019951780986397392,
+      "loss": 1.1386,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019950892903514129,
+      "loss": 1.152,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019949996736996265,
+      "loss": 1.0659,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019949092487571798,
+      "loss": 1.0348,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019948180155975303,
+      "loss": 1.1131,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019947259742947911,
+      "loss": 1.1788,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019946331249237327,
+      "loss": 1.064,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019945394675597807,
+      "loss": 1.202,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019944450022790184,
+      "loss": 1.124,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019943497291581845,
+      "loss": 1.1475,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994253648274675,
+      "loss": 1.0344,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019941567597065414,
+      "loss": 1.1871,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019940590635324915,
+      "loss": 1.0982,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019939605598318888,
+      "loss": 1.0975,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001993861248684753,
+      "loss": 1.1043,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199376113017176,
+      "loss": 1.131,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019936602043742408,
+      "loss": 1.0508,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019935584713741826,
+      "loss": 1.175,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019934559312542296,
+      "loss": 1.1795,
+      "step": 635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019933525840976795,
+      "loss": 1.1411,
+      "step": 640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993248429988487,
+      "loss": 1.1145,
+      "step": 645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019931434690112626,
+      "loss": 1.1791,
+      "step": 650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019930377012512707,
+      "loss": 1.1538,
+      "step": 655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019929311267944324,
+      "loss": 1.0565,
+      "step": 660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992823745727324,
+      "loss": 1.1262,
+      "step": 665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019927155581371767,
+      "loss": 1.088,
+      "step": 670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019926065641118763,
+      "loss": 1.1132,
+      "step": 675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992496763739965,
+      "loss": 1.1825,
+      "step": 680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019923861571106397,
+      "loss": 1.2013,
+      "step": 685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992274744313751,
+      "loss": 1.1262,
+      "step": 690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992162525439806,
+      "loss": 1.0849,
+      "step": 695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019920495005799663,
+      "loss": 1.1018,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991935669826048,
+      "loss": 1.1481,
+      "step": 705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019918210332705213,
+      "loss": 1.1132,
+      "step": 710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019917055910065117,
+      "loss": 1.1347,
+      "step": 715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019915893431277987,
+      "loss": 1.0726,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991472289728817,
+      "loss": 1.1456,
+      "step": 725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019913544309046555,
+      "loss": 1.1092,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019912357667510564,
+      "loss": 1.0654,
+      "step": 735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019911162973644165,
+      "loss": 1.1094,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001990996022841788,
+      "loss": 1.1564,
+      "step": 745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019908749432808766,
+      "loss": 1.1808,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019907530587800413,
+      "loss": 1.0844,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019906303694382948,
+      "loss": 1.0718,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019905068753553045,
+      "loss": 1.1844,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001990382576631391,
+      "loss": 1.0922,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019902574733675284,
+      "loss": 1.1269,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019901315656653445,
+      "loss": 1.1092,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001990004853627122,
+      "loss": 1.1223,
+      "step": 785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019898773373557948,
+      "loss": 1.0795,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001989749016954951,
+      "loss": 1.0622,
+      "step": 795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019896198925288322,
+      "loss": 1.1183,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001989489964182333,
+      "loss": 1.0611,
+      "step": 805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019893592320210013,
+      "loss": 1.1477,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019892276961510377,
+      "loss": 1.1472,
+      "step": 815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019890953566792957,
+      "loss": 1.1411,
+      "step": 820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988962213713281,
+      "loss": 1.0948,
+      "step": 825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988828267361153,
+      "loss": 1.1414,
+      "step": 830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988693517731724,
+      "loss": 1.1161,
+      "step": 835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019885579649344578,
+      "loss": 1.1123,
+      "step": 840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019884216090794711,
+      "loss": 1.1128,
+      "step": 845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019882844502775324,
+      "loss": 1.1177,
+      "step": 850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019881464886400637,
+      "loss": 1.0724,
+      "step": 855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019880077242791385,
+      "loss": 1.1243,
+      "step": 860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019878681573074817,
+      "loss": 1.0985,
+      "step": 865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019877277878384712,
+      "loss": 1.1704,
+      "step": 870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019875866159861375,
+      "loss": 1.2201,
+      "step": 875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019874446418651612,
+      "loss": 1.0622,
+      "step": 880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987301865590875,
+      "loss": 1.1618,
+      "step": 885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987158287279264,
+      "loss": 1.1283,
+      "step": 890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019870139070469643,
+      "loss": 1.1129,
+      "step": 895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019868687250112636,
+      "loss": 1.1868,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019867227412901007,
+      "loss": 1.119,
+      "step": 905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986575956002066,
+      "loss": 1.1817,
+      "step": 910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019864283692664015,
+      "loss": 1.1666,
+      "step": 915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986279981202999,
+      "loss": 1.0744,
+      "step": 920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019861307919324023,
+      "loss": 1.1244,
+      "step": 925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001985980801575805,
+      "loss": 1.0783,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001985830010255053,
+      "loss": 1.1358,
+      "step": 935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019856784180926417,
+      "loss": 1.1039,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001985526025211718,
+      "loss": 1.1694,
+      "step": 945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019853728317360773,
+      "loss": 1.2141,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019852188377901677,
+      "loss": 1.1452,
+      "step": 955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019850640434990869,
+      "loss": 1.0506,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019849084489885813,
+      "loss": 1.0857,
+      "step": 965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019847520543850498,
+      "loss": 1.1396,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019845948598155391,
+      "loss": 1.1202,
+      "step": 975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019844368654077467,
+      "loss": 1.112,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000198427807129002,
+      "loss": 1.0507,
+      "step": 985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001984118477591356,
+      "loss": 1.1932,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001983958084441401,
+      "loss": 1.0773,
+      "step": 995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019837968919704502,
+      "loss": 1.1981,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000198363490030945,
+      "loss": 1.0932,
+      "step": 1005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001983472109589993,
+      "loss": 1.0799,
+      "step": 1010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019833085199443245,
+      "loss": 1.0515,
+      "step": 1015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019831441315053359,
+      "loss": 1.0079,
+      "step": 1020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019829789444065696,
+      "loss": 1.1045,
+      "step": 1025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019828129587822144,
+      "loss": 1.1379,
+      "step": 1030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019826461747671098,
+      "loss": 1.0673,
+      "step": 1035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019824785924967432,
+      "loss": 1.1513,
+      "step": 1040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019823102121072506,
+      "loss": 1.0987,
+      "step": 1045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019821410337354168,
+      "loss": 1.0806,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981971057518673,
+      "loss": 1.075,
+      "step": 1055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019818002835951006,
+      "loss": 1.2132,
+      "step": 1060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019816287121034287,
+      "loss": 1.1387,
+      "step": 1065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981456343183033,
+      "loss": 1.0743,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019812831769739383,
+      "loss": 1.1505,
+      "step": 1075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019811092136168162,
+      "loss": 1.1212,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019809344532529865,
+      "loss": 1.0111,
+      "step": 1085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980758896024417,
+      "loss": 1.0533,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019805825420737215,
+      "loss": 1.1439,
+      "step": 1095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019804053915441615,
+      "loss": 1.0587,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019802274445796458,
+      "loss": 1.1231,
+      "step": 1105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019800487013247302,
+      "loss": 1.0135,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019798691619246173,
+      "loss": 1.05,
+      "step": 1115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019796888265251558,
+      "loss": 1.0607,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019795076952728425,
+      "loss": 1.1267,
+      "step": 1125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019793257683148194,
+      "loss": 1.1437,
+      "step": 1130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001979143045798876,
+      "loss": 1.1138,
+      "step": 1135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019789595278734464,
+      "loss": 1.1417,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019787752146876127,
+      "loss": 1.0941,
+      "step": 1145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019785901063911016,
+      "loss": 1.0948,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001978404203134287,
+      "loss": 1.0978,
+      "step": 1155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019782175050681865,
+      "loss": 1.1043,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019780300123444663,
+      "loss": 1.128,
+      "step": 1165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001977841725115436,
+      "loss": 1.1261,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019776526435340516,
+      "loss": 1.0692,
+      "step": 1175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019774627677539137,
+      "loss": 1.0835,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019772720979292685,
+      "loss": 1.1175,
+      "step": 1185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019770806342150066,
+      "loss": 1.1298,
+      "step": 1190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001976888376766665,
+      "loss": 1.1589,
+      "step": 1195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001976695325740425,
+      "loss": 1.1263,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000197650148129311,
+      "loss": 1.227,
+      "step": 1205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019763068435821917,
+      "loss": 1.1143,
+      "step": 1210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019761114127657842,
+      "loss": 1.1257,
+      "step": 1215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019759151890026463,
+      "loss": 1.1668,
+      "step": 1220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001975718172452181,
+      "loss": 1.0921,
+      "step": 1225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019755203632744353,
+      "loss": 1.1107,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019753217616300995,
+      "loss": 1.0978,
+      "step": 1235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019751223676805083,
+      "loss": 1.118,
+      "step": 1240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000197492218158764,
+      "loss": 1.1295,
+      "step": 1245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974721203514117,
+      "loss": 1.1201,
+      "step": 1250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019745194336232038,
+      "loss": 1.0858,
+      "step": 1255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974316872078808,
+      "loss": 1.1888,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019741135190454818,
+      "loss": 1.13,
+      "step": 1265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019739093746884195,
+      "loss": 1.092,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019737044391734585,
+      "loss": 1.2147,
+      "step": 1275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973498712667078,
+      "loss": 1.2387,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973292195336401,
+      "loss": 1.0601,
+      "step": 1285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019730848873491925,
+      "loss": 1.0656,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019728767888738602,
+      "loss": 1.1525,
+      "step": 1295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019726679000794524,
+      "loss": 1.1181,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001972458221135661,
+      "loss": 1.043,
+      "step": 1305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001972247752212819,
+      "loss": 1.1091,
+      "step": 1310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001972036493481902,
+      "loss": 1.057,
+      "step": 1315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019718244451145256,
+      "loss": 1.0931,
+      "step": 1320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001971611607282949,
+      "loss": 1.14,
+      "step": 1325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019713979801600707,
+      "loss": 1.0144,
+      "step": 1330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019711835639194317,
+      "loss": 1.1032,
+      "step": 1335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019709683587352134,
+      "loss": 1.1165,
+      "step": 1340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019707523647822386,
+      "loss": 1.1322,
+      "step": 1345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000197053558223597,
+      "loss": 1.1617,
+      "step": 1350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019703180112725123,
+      "loss": 1.1173,
+      "step": 1355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019700996520686089,
+      "loss": 1.1341,
+      "step": 1360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019698805048016453,
+      "loss": 1.0231,
+      "step": 1365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019696605696496464,
+      "loss": 1.1765,
+      "step": 1370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019694398467912767,
+      "loss": 1.0667,
+      "step": 1375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969218336405841,
+      "loss": 1.182,
+      "step": 1380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019689960386732837,
+      "loss": 1.1106,
+      "step": 1385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019687729537741893,
+      "loss": 1.1577,
+      "step": 1390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019685490818897817,
+      "loss": 1.0872,
+      "step": 1395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968324423201923,
+      "loss": 1.0934,
+      "step": 1400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019680989778931155,
+      "loss": 1.0749,
+      "step": 1405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019678727461465014,
+      "loss": 1.1183,
+      "step": 1410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001967645728145859,
+      "loss": 1.1155,
+      "step": 1415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001967417924075608,
+      "loss": 1.0706,
+      "step": 1420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019671893341208053,
+      "loss": 1.1589,
+      "step": 1425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019669599584671467,
+      "loss": 1.0504,
+      "step": 1430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019667297973009664,
+      "loss": 1.1075,
+      "step": 1435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019664988508092356,
+      "loss": 1.1005,
+      "step": 1440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966267119179565,
+      "loss": 1.1036,
+      "step": 1445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019660346026002026,
+      "loss": 1.0551,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019658013012600342,
+      "loss": 1.0745,
+      "step": 1455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019655672153485817,
+      "loss": 1.0538,
+      "step": 1460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001965332345056006,
+      "loss": 1.1155,
+      "step": 1465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019650966905731043,
+      "loss": 1.1895,
+      "step": 1470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019648602520913112,
+      "loss": 1.1041,
+      "step": 1475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019646230298026984,
+      "loss": 1.0371,
+      "step": 1480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001964385023899974,
+      "loss": 1.1189,
+      "step": 1485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019641462345764825,
+      "loss": 1.1471,
+      "step": 1490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019639066620262053,
+      "loss": 1.1409,
+      "step": 1495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000196366630644376,
+      "loss": 1.0641,
+      "step": 1500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001963425168024399,
+      "loss": 1.1547,
+      "step": 1505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001963183246964013,
+      "loss": 1.0744,
+      "step": 1510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019629405434591269,
+      "loss": 1.1013,
+      "step": 1515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019626970577069013,
+      "loss": 1.1103,
+      "step": 1520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019624527899051323,
+      "loss": 1.1574,
+      "step": 1525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019622077402522522,
+      "loss": 1.0905,
+      "step": 1530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019619619089473276,
+      "loss": 1.0994,
+      "step": 1535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000196171529619006,
+      "loss": 1.1846,
+      "step": 1540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001961467902180786,
+      "loss": 1.1078,
+      "step": 1545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019612197271204763,
+      "loss": 1.1833,
+      "step": 1550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019609707712107372,
+      "loss": 1.1436,
+      "step": 1555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960721034653808,
+      "loss": 1.0994,
+      "step": 1560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019604705176525642,
+      "loss": 1.1771,
+      "step": 1565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019602192204105123,
+      "loss": 1.2116,
+      "step": 1570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019599671431317958,
+      "loss": 1.0557,
+      "step": 1575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019597142860211894,
+      "loss": 1.1689,
+      "step": 1580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001959460649284102,
+      "loss": 1.0702,
+      "step": 1585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001959206233126577,
+      "loss": 1.1284,
+      "step": 1590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019589510377552902,
+      "loss": 1.1074,
+      "step": 1595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019586950633775492,
+      "loss": 1.0759,
+      "step": 1600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001958438310201296,
+      "loss": 1.1689,
+      "step": 1605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019581807784351047,
+      "loss": 1.0666,
+      "step": 1610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001957922468288182,
+      "loss": 1.2082,
+      "step": 1615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001957663379970367,
+      "loss": 1.0637,
+      "step": 1620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019574035136921304,
+      "loss": 1.1707,
+      "step": 1625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019571428696645756,
+      "loss": 1.166,
+      "step": 1630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019568814480994373,
+      "loss": 1.1066,
+      "step": 1635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019566192492090817,
+      "loss": 1.1559,
+      "step": 1640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019563562732065071,
+      "loss": 1.1194,
+      "step": 1645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019560925203053432,
+      "loss": 1.1467,
+      "step": 1650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000195582799071985,
+      "loss": 1.1215,
+      "step": 1655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019555626846649184,
+      "loss": 1.1997,
+      "step": 1660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019552966023560713,
+      "loss": 1.2071,
+      "step": 1665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019550297440094604,
+      "loss": 1.0758,
+      "step": 1670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019547621098418697,
+      "loss": 1.1419,
+      "step": 1675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019544937000707122,
+      "loss": 1.0925,
+      "step": 1680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001954224514914031,
+      "loss": 1.0968,
+      "step": 1685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019539545545904996,
+      "loss": 1.1292,
+      "step": 1690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019536838193194217,
+      "loss": 1.0844,
+      "step": 1695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019534123093207286,
+      "loss": 1.0658,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019531400248149828,
+      "loss": 1.0474,
+      "step": 1705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019528669660233755,
+      "loss": 1.0425,
+      "step": 1710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019525931331677265,
+      "loss": 1.1484,
+      "step": 1715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019523185264704848,
+      "loss": 1.0311,
+      "step": 1720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019520431461547276,
+      "loss": 1.0861,
+      "step": 1725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019517669924441612,
+      "loss": 1.1524,
+      "step": 1730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019514900655631187,
+      "loss": 1.0902,
+      "step": 1735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019512123657365637,
+      "loss": 1.0677,
+      "step": 1740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019509338931900861,
+      "loss": 1.1203,
+      "step": 1745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019506546481499042,
+      "loss": 1.098,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019503746308428629,
+      "loss": 1.0712,
+      "step": 1755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001950093841496435,
+      "loss": 1.0832,
+      "step": 1760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019498122803387208,
+      "loss": 1.0931,
+      "step": 1765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019495299475984467,
+      "loss": 1.0802,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019492468435049673,
+      "loss": 1.1396,
+      "step": 1775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001948962968288262,
+      "loss": 1.1725,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019486783221789384,
+      "loss": 1.0901,
+      "step": 1785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019483929054082296,
+      "loss": 1.0767,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019481067182079934,
+      "loss": 1.1499,
+      "step": 1795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001947819760810716,
+      "loss": 1.0896,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019475320334495075,
+      "loss": 1.0615,
+      "step": 1805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001947243536358104,
+      "loss": 1.0938,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019469542697708674,
+      "loss": 1.1678,
+      "step": 1815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001946664233922784,
+      "loss": 1.0664,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019463734290494644,
+      "loss": 1.1349,
+      "step": 1825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019460818553871458,
+      "loss": 1.1067,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019457895131726888,
+      "loss": 1.1333,
+      "step": 1835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019454964026435775,
+      "loss": 1.0924,
+      "step": 1840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019452025240379215,
+      "loss": 1.0811,
+      "step": 1845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019449078775944537,
+      "loss": 1.1043,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001944612463552531,
+      "loss": 1.1101,
+      "step": 1855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019443162821521342,
+      "loss": 1.218,
+      "step": 1860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019440193336338674,
+      "loss": 1.1898,
+      "step": 1865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019437216182389565,
+      "loss": 1.0866,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019434231362092515,
+      "loss": 1.0558,
+      "step": 1875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019431238877872257,
+      "loss": 1.1855,
+      "step": 1880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019428238732159737,
+      "loss": 1.099,
+      "step": 1885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001942523092739214,
+      "loss": 1.165,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019422215466012856,
+      "loss": 1.1275,
+      "step": 1895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019419192350471508,
+      "loss": 1.1306,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019416161583223928,
+      "loss": 1.1622,
+      "step": 1905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019413123166732172,
+      "loss": 1.1309,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019410077103464501,
+      "loss": 1.1474,
+      "step": 1915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019407023395895398,
+      "loss": 1.132,
+      "step": 1920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940396204650555,
+      "loss": 1.0644,
+      "step": 1925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940089305778185,
+      "loss": 1.142,
+      "step": 1930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019397816432217398,
+      "loss": 1.0239,
+      "step": 1935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019394732172311503,
+      "loss": 1.0686,
+      "step": 1940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019391640280569668,
+      "loss": 1.1765,
+      "step": 1945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019388540759503604,
+      "loss": 1.0821,
+      "step": 1950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019385433611631214,
+      "loss": 1.1488,
+      "step": 1955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019382318839476594,
+      "loss": 1.1158,
+      "step": 1960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019379196445570032,
+      "loss": 1.099,
+      "step": 1965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001937606643244803,
+      "loss": 1.1725,
+      "step": 1970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019372928802653252,
+      "loss": 1.1067,
+      "step": 1975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019369783558734565,
+      "loss": 1.1181,
+      "step": 1980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001936663070324701,
+      "loss": 1.1236,
+      "step": 1985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019363470238751825,
+      "loss": 1.0417,
+      "step": 1990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001936030216781641,
+      "loss": 1.1227,
+      "step": 1995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019357126493014368,
+      "loss": 1.1774,
+      "step": 2000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019353943216925465,
+      "loss": 1.1386,
+      "step": 2005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019350752342135638,
+      "loss": 1.1066,
+      "step": 2010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019347553871237001,
+      "loss": 1.1344,
+      "step": 2015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001934434780682785,
+      "loss": 1.1089,
+      "step": 2020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001934113415151264,
+      "loss": 1.1433,
+      "step": 2025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019337912907901978,
+      "loss": 1.1276,
+      "step": 2030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019334684078612662,
+      "loss": 1.1633,
+      "step": 2035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019331447666267644,
+      "loss": 1.1863,
+      "step": 2040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019328203673496023,
+      "loss": 1.1657,
+      "step": 2045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001932495210293307,
+      "loss": 1.1807,
+      "step": 2050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019321692957220208,
+      "loss": 1.1703,
+      "step": 2055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001931842623900501,
+      "loss": 1.1008,
+      "step": 2060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019315151950941202,
+      "loss": 1.1054,
+      "step": 2065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019311870095688665,
+      "loss": 1.0917,
+      "step": 2070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001930858067591342,
+      "loss": 1.1134,
+      "step": 2075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019305283694287634,
+      "loss": 1.1339,
+      "step": 2080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001930197915348962,
+      "loss": 1.1417,
+      "step": 2085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019298667056203832,
+      "loss": 1.0491,
+      "step": 2090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019295347405120868,
+      "loss": 1.0736,
+      "step": 2095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019292020202937442,
+      "loss": 1.1395,
+      "step": 2100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019288685452356424,
+      "loss": 1.0608,
+      "step": 2105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019285343156086803,
+      "loss": 1.1576,
+      "step": 2110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001928199331684371,
+      "loss": 1.1131,
+      "step": 2115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019278635937348393,
+      "loss": 1.0297,
+      "step": 2120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019275271020328225,
+      "loss": 1.1325,
+      "step": 2125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000192718985685167,
+      "loss": 1.1304,
+      "step": 2130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001926851858465345,
+      "loss": 1.08,
+      "step": 2135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019265131071484213,
+      "loss": 1.0682,
+      "step": 2140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019261736031760837,
+      "loss": 1.1576,
+      "step": 2145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019258333468241296,
+      "loss": 1.128,
+      "step": 2150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019254923383689665,
+      "loss": 1.05,
+      "step": 2155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019251505780876148,
+      "loss": 1.1599,
+      "step": 2160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019248080662577036,
+      "loss": 1.1629,
+      "step": 2165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001924464803157473,
+      "loss": 1.1082,
+      "step": 2170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001924120789065774,
+      "loss": 1.132,
+      "step": 2175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001923776024262068,
+      "loss": 1.0797,
+      "step": 2180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001923430509026425,
+      "loss": 1.1171,
+      "step": 2185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019230842436395248,
+      "loss": 1.1678,
+      "step": 2190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019227372283826583,
+      "loss": 1.1186,
+      "step": 2195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019223894635377232,
+      "loss": 1.2379,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019220409493872275,
+      "loss": 1.1325,
+      "step": 2205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001921691686214288,
+      "loss": 1.0757,
+      "step": 2210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019213416743026285,
+      "loss": 1.2045,
+      "step": 2215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019209909139365834,
+      "loss": 1.1018,
+      "step": 2220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001920639405401093,
+      "loss": 1.123,
+      "step": 2225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001920287148981707,
+      "loss": 1.1013,
+      "step": 2230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001919934144964581,
+      "loss": 1.1402,
+      "step": 2235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019195803936364791,
+      "loss": 1.1148,
+      "step": 2240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019192258952847717,
+      "loss": 1.1465,
+      "step": 2245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019188706501974373,
+      "loss": 1.1607,
+      "step": 2250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019185146586630594,
+      "loss": 1.1231,
+      "step": 2255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019181579209708288,
+      "loss": 1.1591,
+      "step": 2260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001917800437410543,
+      "loss": 1.1014,
+      "step": 2265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019174422082726043,
+      "loss": 1.236,
+      "step": 2270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001917083233848021,
+      "loss": 1.1514,
+      "step": 2275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019167235144284072,
+      "loss": 1.0706,
+      "step": 2280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001916363050305982,
+      "loss": 1.0455,
+      "step": 2285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019160018417735693,
+      "loss": 1.1022,
+      "step": 2290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019156398891245974,
+      "loss": 1.1167,
+      "step": 2295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019152771926531,
+      "loss": 1.1118,
+      "step": 2300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019149137526537142,
+      "loss": 1.0846,
+      "step": 2305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001914549569421682,
+      "loss": 1.0916,
+      "step": 2310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001914184643252848,
+      "loss": 1.1376,
+      "step": 2315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019138189744436613,
+      "loss": 1.1085,
+      "step": 2320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019134525632911735,
+      "loss": 1.1026,
+      "step": 2325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019130854100930404,
+      "loss": 1.0874,
+      "step": 2330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019127175151475195,
+      "loss": 1.1252,
+      "step": 2335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019123488787534714,
+      "loss": 1.14,
+      "step": 2340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019119795012103588,
+      "loss": 1.1472,
+      "step": 2345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019116093828182464,
+      "loss": 1.1675,
+      "step": 2350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019112385238778012,
+      "loss": 1.1193,
+      "step": 2355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019108669246902915,
+      "loss": 1.139,
+      "step": 2360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019104945855575862,
+      "loss": 1.1458,
+      "step": 2365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019101215067821565,
+      "loss": 1.1479,
+      "step": 2370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019097476886670737,
+      "loss": 1.107,
+      "step": 2375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000190937313151601,
+      "loss": 1.09,
+      "step": 2380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019089978356332386,
+      "loss": 1.2123,
+      "step": 2385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019086218013236313,
+      "loss": 1.166,
+      "step": 2390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019082450288926615,
+      "loss": 1.1012,
+      "step": 2395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019078675186464002,
+      "loss": 1.0627,
+      "step": 2400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001907489270891519,
+      "loss": 1.1408,
+      "step": 2405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001907110285935289,
+      "loss": 1.0716,
+      "step": 2410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019067305640855793,
+      "loss": 1.0738,
+      "step": 2415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001906350105650858,
+      "loss": 1.1093,
+      "step": 2420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019059689109401915,
+      "loss": 1.1295,
+      "step": 2425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019055869802632447,
+      "loss": 1.1377,
+      "step": 2430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019052043139302795,
+      "loss": 1.1059,
+      "step": 2435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019048209122521562,
+      "loss": 1.107,
+      "step": 2440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001904436775540333,
+      "loss": 1.1829,
+      "step": 2445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001904051904106863,
+      "loss": 1.1255,
+      "step": 2450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001903666298264398,
+      "loss": 1.1854,
+      "step": 2455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019032799583261866,
+      "loss": 1.1515,
+      "step": 2460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019028928846060721,
+      "loss": 1.0586,
+      "step": 2465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019025050774184953,
+      "loss": 1.1042,
+      "step": 2470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019021165370784922,
+      "loss": 1.066,
+      "step": 2475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019017272639016954,
+      "loss": 1.1116,
+      "step": 2480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019013372582043312,
+      "loss": 1.1261,
+      "step": 2485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019009465203032223,
+      "loss": 1.1608,
+      "step": 2490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019005550505157848,
+      "loss": 1.1959,
+      "step": 2495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019001628491600314,
+      "loss": 1.1328,
+      "step": 2500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018997699165545676,
+      "loss": 1.133,
+      "step": 2505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018993762530185925,
+      "loss": 1.1307,
+      "step": 2510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018989818588719003,
+      "loss": 1.0921,
+      "step": 2515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018985867344348782,
+      "loss": 1.1079,
+      "step": 2520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018981908800285065,
+      "loss": 1.1946,
+      "step": 2525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018977942959743583,
+      "loss": 1.0337,
+      "step": 2530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018973969825946007,
+      "loss": 1.1972,
+      "step": 2535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018969989402119913,
+      "loss": 1.0439,
+      "step": 2540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018966001691498807,
+      "loss": 1.1402,
+      "step": 2545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018962006697322123,
+      "loss": 1.1336,
+      "step": 2550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001895800442283521,
+      "loss": 1.1067,
+      "step": 2555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001895399487128932,
+      "loss": 1.1201,
+      "step": 2560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001894997804594162,
+      "loss": 1.1701,
+      "step": 2565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018945953950055187,
+      "loss": 1.1359,
+      "step": 2570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018941922586899015,
+      "loss": 1.1092,
+      "step": 2575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018937883959747992,
+      "loss": 1.0772,
+      "step": 2580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018933838071882904,
+      "loss": 1.1807,
+      "step": 2585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018929784926590435,
+      "loss": 1.1507,
+      "step": 2590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018925724527163177,
+      "loss": 1.0999,
+      "step": 2595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000189216568768996,
+      "loss": 1.1505,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018917581979104072,
+      "loss": 1.1033,
+      "step": 2605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018913499837086852,
+      "loss": 1.0842,
+      "step": 2610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018909410454164073,
+      "loss": 1.0672,
+      "step": 2615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001890531383365776,
+      "loss": 1.1222,
+      "step": 2620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001890120997889581,
+      "loss": 1.1143,
+      "step": 2625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018897098893212,
+      "loss": 1.1198,
+      "step": 2630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018892980579945985,
+      "loss": 1.2037,
+      "step": 2635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001888885504244329,
+      "loss": 1.1384,
+      "step": 2640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018884722284055307,
+      "loss": 1.1693,
+      "step": 2645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018880582308139284,
+      "loss": 1.1019,
+      "step": 2650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018876435118058348,
+      "loss": 1.0825,
+      "step": 2655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018872280717181477,
+      "loss": 1.0906,
+      "step": 2660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018868119108883516,
+      "loss": 1.1341,
+      "step": 2665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018863950296545148,
+      "loss": 1.0723,
+      "step": 2670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018859774283552932,
+      "loss": 1.1234,
+      "step": 2675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001885559107329926,
+      "loss": 1.1248,
+      "step": 2680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018851400669182367,
+      "loss": 1.0689,
+      "step": 2685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018847203074606335,
+      "loss": 1.1267,
+      "step": 2690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018842998292981108,
+      "loss": 1.1193,
+      "step": 2695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018838786327722443,
+      "loss": 1.0327,
+      "step": 2700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001883456718225194,
+      "loss": 1.1111,
+      "step": 2705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018830340859997036,
+      "loss": 1.1441,
+      "step": 2710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018826107364390993,
+      "loss": 1.1243,
+      "step": 2715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000188218666988729,
+      "loss": 1.0446,
+      "step": 2720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001881761886688768,
+      "loss": 1.1997,
+      "step": 2725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018813363871886065,
+      "loss": 1.1008,
+      "step": 2730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018809101717324618,
+      "loss": 1.1542,
+      "step": 2735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018804832406665707,
+      "loss": 1.2327,
+      "step": 2740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018800555943377514,
+      "loss": 1.1403,
+      "step": 2745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001879627233093404,
+      "loss": 1.1483,
+      "step": 2750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018791981572815087,
+      "loss": 1.122,
+      "step": 2755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018787683672506265,
+      "loss": 1.0624,
+      "step": 2760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001878337863349899,
+      "loss": 1.0879,
+      "step": 2765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018779066459290468,
+      "loss": 1.1375,
+      "step": 2770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018774747153383703,
+      "loss": 1.0634,
+      "step": 2775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000187704207192875,
+      "loss": 1.0975,
+      "step": 2780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018766087160516446,
+      "loss": 1.1125,
+      "step": 2785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018761746480590915,
+      "loss": 1.0661,
+      "step": 2790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001875739868303708,
+      "loss": 1.1257,
+      "step": 2795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018753043771386877,
+      "loss": 1.1633,
+      "step": 2800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018748681749178035,
+      "loss": 1.1008,
+      "step": 2805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001874431261995406,
+      "loss": 1.1241,
+      "step": 2810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018739936387264207,
+      "loss": 1.1435,
+      "step": 2815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018735553054663538,
+      "loss": 1.1364,
+      "step": 2820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001873116262571286,
+      "loss": 1.0481,
+      "step": 2825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018726765103978745,
+      "loss": 1.072,
+      "step": 2830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018722360493033538,
+      "loss": 1.1007,
+      "step": 2835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001871794879645533,
+      "loss": 1.094,
+      "step": 2840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001871353001782798,
+      "loss": 1.1331,
+      "step": 2845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018709104160741097,
+      "loss": 1.0766,
+      "step": 2850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018704671228790036,
+      "loss": 1.1216,
+      "step": 2855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018700231225575892,
+      "loss": 1.1281,
+      "step": 2860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018695784154705523,
+      "loss": 1.1036,
+      "step": 2865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001869133001979152,
+      "loss": 1.0473,
+      "step": 2870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018686868824452208,
+      "loss": 1.0983,
+      "step": 2875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018682400572311652,
+      "loss": 1.1571,
+      "step": 2880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018677925266999645,
+      "loss": 1.2094,
+      "step": 2885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001867344291215172,
+      "loss": 1.157,
+      "step": 2890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018668953511409125,
+      "loss": 1.1718,
+      "step": 2895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018664457068418835,
+      "loss": 1.1758,
+      "step": 2900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018659953586833557,
+      "loss": 1.1578,
+      "step": 2905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018655443070311696,
+      "loss": 1.1511,
+      "step": 2910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018650925522517388,
+      "loss": 1.1209,
+      "step": 2915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018646400947120473,
+      "loss": 1.1243,
+      "step": 2920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000186418693477965,
+      "loss": 1.1156,
+      "step": 2925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018637330728226727,
+      "loss": 1.1021,
+      "step": 2930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001863278509209812,
+      "loss": 1.1547,
+      "step": 2935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001862823244310333,
+      "loss": 1.1437,
+      "step": 2940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018623672784940723,
+      "loss": 1.1695,
+      "step": 2945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018619106121314343,
+      "loss": 1.1239,
+      "step": 2950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018614532455933934,
+      "loss": 1.1569,
+      "step": 2955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018609951792514923,
+      "loss": 1.0326,
+      "step": 2960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001860536413477842,
+      "loss": 1.1536,
+      "step": 2965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001860076948645123,
+      "loss": 1.0572,
+      "step": 2970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018596167851265823,
+      "loss": 1.1142,
+      "step": 2975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018591559232960354,
+      "loss": 1.1039,
+      "step": 2980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001858694363527864,
+      "loss": 1.1157,
+      "step": 2985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018582321061970177,
+      "loss": 1.1308,
+      "step": 2990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018577691516790125,
+      "loss": 1.1011,
+      "step": 2995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001857305500349931,
+      "loss": 1.1176,
+      "step": 3000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018568411525864218,
+      "loss": 1.1115,
+      "step": 3005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018563761087656979,
+      "loss": 1.0932,
+      "step": 3010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018559103692655405,
+      "loss": 1.1332,
+      "step": 3015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018554439344642932,
+      "loss": 1.1191,
+      "step": 3020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001854976804740866,
+      "loss": 1.1141,
+      "step": 3025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018545089804747325,
+      "loss": 1.1157,
+      "step": 3030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018540404620459315,
+      "loss": 1.0612,
+      "step": 3035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001853571249835065,
+      "loss": 1.1196,
+      "step": 3040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001853101344223299,
+      "loss": 1.1427,
+      "step": 3045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018526307455923622,
+      "loss": 1.107,
+      "step": 3050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018521594543245465,
+      "loss": 1.0782,
+      "step": 3055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001851687470802708,
+      "loss": 1.1291,
+      "step": 3060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018512147954102622,
+      "loss": 1.1437,
+      "step": 3065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018507414285311893,
+      "loss": 1.1083,
+      "step": 3070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001850267370550029,
+      "loss": 1.1624,
+      "step": 3075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018497926218518846,
+      "loss": 1.1455,
+      "step": 3080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001849317182822419,
+      "loss": 1.1393,
+      "step": 3085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018488410538478566,
+      "loss": 1.1823,
+      "step": 3090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001848364235314982,
+      "loss": 1.1268,
+      "step": 3095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000184788672761114,
+      "loss": 1.0425,
+      "step": 3100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018474085311242356,
+      "loss": 1.0912,
+      "step": 3105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018469296462427323,
+      "loss": 1.1366,
+      "step": 3110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018464500733556536,
+      "loss": 1.1343,
+      "step": 3115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018459698128525827,
+      "loss": 1.1269,
+      "step": 3120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018454888651236602,
+      "loss": 1.085,
+      "step": 3125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018450072305595844,
+      "loss": 1.2009,
+      "step": 3130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018445249095516128,
+      "loss": 1.0964,
+      "step": 3135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018440419024915607,
+      "loss": 1.1326,
+      "step": 3140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018435582097717993,
+      "loss": 1.1127,
+      "step": 3145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018430738317852592,
+      "loss": 1.1855,
+      "step": 3150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018425887689254246,
+      "loss": 1.1376,
+      "step": 3155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018421030215863383,
+      "loss": 1.0792,
+      "step": 3160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018416165901625984,
+      "loss": 1.1013,
+      "step": 3165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001841129475049359,
+      "loss": 1.1642,
+      "step": 3170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001840641676642329,
+      "loss": 1.0473,
+      "step": 3175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001840153195337773,
+      "loss": 1.1308,
+      "step": 3180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018396640315325102,
+      "loss": 1.0742,
+      "step": 3185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001839174185623914,
+      "loss": 1.1699,
+      "step": 3190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018386836580099112,
+      "loss": 1.0544,
+      "step": 3195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018381924490889841,
+      "loss": 1.2204,
+      "step": 3200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001837700559260167,
+      "loss": 1.1262,
+      "step": 3205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001837207988923049,
+      "loss": 1.1881,
+      "step": 3210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018367147384777704,
+      "loss": 1.1342,
+      "step": 3215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018362208083250246,
+      "loss": 1.1371,
+      "step": 3220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018357261988660568,
+      "loss": 1.1405,
+      "step": 3225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018352309105026647,
+      "loss": 1.1261,
+      "step": 3230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018347349436371966,
+      "loss": 1.106,
+      "step": 3235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001834238298672553,
+      "loss": 1.1044,
+      "step": 3240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018337409760121846,
+      "loss": 1.1062,
+      "step": 3245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018332429760600926,
+      "loss": 1.1224,
+      "step": 3250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018327442992208289,
+      "loss": 1.1594,
+      "step": 3255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001832244945899495,
+      "loss": 1.0388,
+      "step": 3260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001831744916501742,
+      "loss": 1.0519,
+      "step": 3265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018312442114337696,
+      "loss": 1.1716,
+      "step": 3270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018307428311023272,
+      "loss": 1.1017,
+      "step": 3275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018302407759147128,
+      "loss": 1.1298,
+      "step": 3280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001829738046278772,
+      "loss": 1.1033,
+      "step": 3285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018292346426028984,
+      "loss": 1.2006,
+      "step": 3290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001828730565296033,
+      "loss": 1.1079,
+      "step": 3295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018282258147676655,
+      "loss": 1.1674,
+      "step": 3300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000182772039142783,
+      "loss": 1.0647,
+      "step": 3305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018272142956871098,
+      "loss": 1.1953,
+      "step": 3310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018267075279566317,
+      "loss": 1.1535,
+      "step": 3315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001826200088648071,
+      "loss": 1.0722,
+      "step": 3320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001825691978173647,
+      "loss": 1.1739,
+      "step": 3325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018251831969461245,
+      "loss": 1.1158,
+      "step": 3330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018246737453788135,
+      "loss": 1.1133,
+      "step": 3335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018241636238855682,
+      "loss": 1.1238,
+      "step": 3340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018236528328807874,
+      "loss": 1.0619,
+      "step": 3345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018231413727794134,
+      "loss": 1.0751,
+      "step": 3350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001822629243996932,
+      "loss": 1.1201,
+      "step": 3355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018221164469493727,
+      "loss": 1.1791,
+      "step": 3360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018216029820533074,
+      "loss": 1.1038,
+      "step": 3365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001821088849725851,
+      "loss": 1.0982,
+      "step": 3370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018205740503846596,
+      "loss": 1.1717,
+      "step": 3375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018200585844479325,
+      "loss": 1.0518,
+      "step": 3380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018195424523344094,
+      "loss": 1.0946,
+      "step": 3385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018190256544633717,
+      "loss": 1.1263,
+      "step": 3390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001818508191254642,
+      "loss": 1.1184,
+      "step": 3395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018179900631285823,
+      "loss": 1.1946,
+      "step": 3400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018174712705060957,
+      "loss": 1.0804,
+      "step": 3405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018169518138086245,
+      "loss": 1.1199,
+      "step": 3410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001816431693458151,
+      "loss": 1.1402,
+      "step": 3415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018159109098771963,
+      "loss": 1.0885,
+      "step": 3420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018153894634888204,
+      "loss": 1.1906,
+      "step": 3425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018148673547166216,
+      "loss": 1.1324,
+      "step": 3430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018143445839847368,
+      "loss": 1.1636,
+      "step": 3435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018138211517178395,
+      "loss": 1.1203,
+      "step": 3440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018132970583411418,
+      "loss": 1.133,
+      "step": 3445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018127723042803923,
+      "loss": 1.1441,
+      "step": 3450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001812246889961876,
+      "loss": 1.1346,
+      "step": 3455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001811720815812415,
+      "loss": 1.0715,
+      "step": 3460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018111940822593668,
+      "loss": 1.1466,
+      "step": 3465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018106666897306254,
+      "loss": 1.0836,
+      "step": 3470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018101386386546188,
+      "loss": 1.2009,
+      "step": 3475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001809609929460311,
+      "loss": 1.079,
+      "step": 3480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018090805625772002,
+      "loss": 1.0886,
+      "step": 3485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018085505384353193,
+      "loss": 1.0999,
+      "step": 3490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001808019857465235,
+      "loss": 1.0753,
+      "step": 3495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018074885200980472,
+      "loss": 1.0931,
+      "step": 3500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018069565267653888,
+      "loss": 1.1192,
+      "step": 3505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001806423877899426,
+      "loss": 1.1133,
+      "step": 3510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018058905739328583,
+      "loss": 1.0897,
+      "step": 3515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001805356615298916,
+      "loss": 1.1205,
+      "step": 3520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018048220024313618,
+      "loss": 1.0214,
+      "step": 3525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000180428673576449,
+      "loss": 1.1051,
+      "step": 3530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018037508157331267,
+      "loss": 1.0827,
+      "step": 3535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018032142427726268,
+      "loss": 1.1091,
+      "step": 3540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018026770173188778,
+      "loss": 1.1486,
+      "step": 3545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018021391398082953,
+      "loss": 1.1414,
+      "step": 3550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018016006106778265,
+      "loss": 1.0697,
+      "step": 3555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001801061430364946,
+      "loss": 1.1202,
+      "step": 3560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001800521599307659,
+      "loss": 1.0858,
+      "step": 3565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017999811179444984,
+      "loss": 1.0955,
+      "step": 3570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001799439986714526,
+      "loss": 1.1031,
+      "step": 3575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017988982060573313,
+      "loss": 1.0493,
+      "step": 3580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017983557764130308,
+      "loss": 1.1149,
+      "step": 3585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001797812698222269,
+      "loss": 1.1113,
+      "step": 3590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017972689719262174,
+      "loss": 1.0224,
+      "step": 3595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017967245979665723,
+      "loss": 1.0629,
+      "step": 3600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017961795767855584,
+      "loss": 1.0922,
+      "step": 3605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017956339088259248,
+      "loss": 1.0676,
+      "step": 3610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017950875945309459,
+      "loss": 1.1144,
+      "step": 3615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017945406343444225,
+      "loss": 1.0925,
+      "step": 3620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017939930287106778,
+      "loss": 1.1248,
+      "step": 3625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017934447780745623,
+      "loss": 1.1524,
+      "step": 3630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001792895882881448,
+      "loss": 1.1551,
+      "step": 3635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017923463435772316,
+      "loss": 1.1304,
+      "step": 3640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017917961606083327,
+      "loss": 1.1746,
+      "step": 3645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017912453344216936,
+      "loss": 1.1166,
+      "step": 3650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000179069386546478,
+      "loss": 1.1669,
+      "step": 3655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001790141754185579,
+      "loss": 1.079,
+      "step": 3660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017895890010325991,
+      "loss": 1.1902,
+      "step": 3665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001789035606454872,
+      "loss": 1.1527,
+      "step": 3670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017884815709019484,
+      "loss": 1.1386,
+      "step": 3675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001787926894823901,
+      "loss": 1.1751,
+      "step": 3680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001787371578671322,
+      "loss": 1.0432,
+      "step": 3685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017868156228953243,
+      "loss": 1.0398,
+      "step": 3690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000178625902794754,
+      "loss": 1.1049,
+      "step": 3695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017857017942801207,
+      "loss": 1.0748,
+      "step": 3700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001785143922345736,
+      "loss": 1.1651,
+      "step": 3705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001784585412597575,
+      "loss": 1.1152,
+      "step": 3710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001784026265489345,
+      "loss": 1.0759,
+      "step": 3715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000178346648147527,
+      "loss": 1.1675,
+      "step": 3720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017829060610100925,
+      "loss": 1.1433,
+      "step": 3725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001782345004549072,
+      "loss": 1.097,
+      "step": 3730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017817833125479836,
+      "loss": 1.1473,
+      "step": 3735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00017812209854631194,
+      "loss": 1.1478,
+      "step": 3740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017806580237512867,
+      "loss": 1.0655,
+      "step": 3745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000178009442786981,
+      "loss": 1.1638,
+      "step": 3750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001779530198276528,
+      "loss": 1.1192,
+      "step": 3755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001778965335429793,
+      "loss": 1.1106,
+      "step": 3760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001778399839788474,
+      "loss": 1.1277,
+      "step": 3765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017778337118119517,
+      "loss": 1.0939,
+      "step": 3770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001777266951960123,
+      "loss": 1.1052,
+      "step": 3775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017766995606933956,
+      "loss": 1.0946,
+      "step": 3780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001776131538472692,
+      "loss": 1.075,
+      "step": 3785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017755628857594465,
+      "loss": 1.0893,
+      "step": 3790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001774993603015606,
+      "loss": 1.1242,
+      "step": 3795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001774423690703628,
+      "loss": 1.0987,
+      "step": 3800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001773853149286483,
+      "loss": 1.1195,
+      "step": 3805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017732819792276515,
+      "loss": 1.2353,
+      "step": 3810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017727101809911257,
+      "loss": 1.11,
+      "step": 3815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017721377550414065,
+      "loss": 1.0743,
+      "step": 3820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017715647018435064,
+      "loss": 1.0919,
+      "step": 3825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017709910218629462,
+      "loss": 1.1631,
+      "step": 3830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017704167155657564,
+      "loss": 1.1065,
+      "step": 3835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017698417834184762,
+      "loss": 1.1687,
+      "step": 3840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017692662258881537,
+      "loss": 1.1229,
+      "step": 3845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017686900434423452,
+      "loss": 1.1302,
+      "step": 3850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017681132365491122,
+      "loss": 1.1169,
+      "step": 3855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017675358056770267,
+      "loss": 1.1913,
+      "step": 3860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001766957751295166,
+      "loss": 1.1608,
+      "step": 3865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017663790738731142,
+      "loss": 1.0848,
+      "step": 3870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001765799773880961,
+      "loss": 1.1747,
+      "step": 3875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017652198517893026,
+      "loss": 1.1902,
+      "step": 3880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017646393080692404,
+      "loss": 1.0914,
+      "step": 3885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017640581431923807,
+      "loss": 1.2147,
+      "step": 3890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017634763576308344,
+      "loss": 1.1117,
+      "step": 3895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017628939518572157,
+      "loss": 1.1725,
+      "step": 3900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017623109263446442,
+      "loss": 1.1705,
+      "step": 3905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001761727281566743,
+      "loss": 1.1261,
+      "step": 3910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001761143017997636,
+      "loss": 1.1658,
+      "step": 3915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001760558136111952,
+      "loss": 1.1473,
+      "step": 3920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017599726363848214,
+      "loss": 1.121,
+      "step": 3925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017593865192918766,
+      "loss": 1.1463,
+      "step": 3930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001758799785309251,
+      "loss": 1.0691,
+      "step": 3935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000175821243491358,
+      "loss": 1.061,
+      "step": 3940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017576244685819993,
+      "loss": 1.1438,
+      "step": 3945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001757035886792144,
+      "loss": 1.1461,
+      "step": 3950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001756446690022151,
+      "loss": 1.047,
+      "step": 3955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017558568787506555,
+      "loss": 1.2109,
+      "step": 3960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017552664534567923,
+      "loss": 1.1237,
+      "step": 3965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017546754146201954,
+      "loss": 1.0917,
+      "step": 3970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001754083762720996,
+      "loss": 1.0407,
+      "step": 3975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001753491498239825,
+      "loss": 1.0925,
+      "step": 3980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000175289862165781,
+      "loss": 1.1538,
+      "step": 3985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017523051334565753,
+      "loss": 1.0669,
+      "step": 3990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001751711034118243,
+      "loss": 1.0704,
+      "step": 3995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017511163241254308,
+      "loss": 1.1259,
+      "step": 4000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001750521003961253,
+      "loss": 1.1539,
+      "step": 4005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000174992507410932,
+      "loss": 1.1291,
+      "step": 4010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017493285350537367,
+      "loss": 1.2024,
+      "step": 4015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017487313872791035,
+      "loss": 1.0821,
+      "step": 4020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017481336312705144,
+      "loss": 1.1142,
+      "step": 4025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017475352675135595,
+      "loss": 1.0999,
+      "step": 4030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000174693629649432,
+      "loss": 1.1705,
+      "step": 4035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001746336718699372,
+      "loss": 1.0969,
+      "step": 4040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017457365346157836,
+      "loss": 1.09,
+      "step": 4045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017451357447311163,
+      "loss": 1.168,
+      "step": 4050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001744534349533424,
+      "loss": 1.1579,
+      "step": 4055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017439323495112508,
+      "loss": 1.1576,
+      "step": 4060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017433297451536331,
+      "loss": 1.1181,
+      "step": 4065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001742726536950099,
+      "loss": 1.155,
+      "step": 4070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017421227253906654,
+      "loss": 1.1496,
+      "step": 4075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001741518310965841,
+      "loss": 1.1455,
+      "step": 4080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001740913294166623,
+      "loss": 1.0682,
+      "step": 4085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001740307675484498,
+      "loss": 1.0999,
+      "step": 4090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001739701455411443,
+      "loss": 1.1136,
+      "step": 4095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017390946344399223,
+      "loss": 1.0865,
+      "step": 4100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017384872130628875,
+      "loss": 1.1859,
+      "step": 4105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000173787919177378,
+      "loss": 1.1327,
+      "step": 4110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017372705710665275,
+      "loss": 1.0803,
+      "step": 4115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017366613514355442,
+      "loss": 1.1387,
+      "step": 4120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017360515333757312,
+      "loss": 1.1616,
+      "step": 4125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017354411173824762,
+      "loss": 1.1676,
+      "step": 4130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001734830103951653,
+      "loss": 1.0807,
+      "step": 4135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017342184935796186,
+      "loss": 1.1346,
+      "step": 4140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017336062867632176,
+      "loss": 1.1262,
+      "step": 4145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017329934839997765,
+      "loss": 1.18,
+      "step": 4150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017323800857871083,
+      "loss": 1.1422,
+      "step": 4155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017317660926235084,
+      "loss": 1.1332,
+      "step": 4160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001731151505007756,
+      "loss": 1.1504,
+      "step": 4165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017305363234391123,
+      "loss": 1.0972,
+      "step": 4170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017299205484173224,
+      "loss": 1.0445,
+      "step": 4175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017293041804426125,
+      "loss": 1.1446,
+      "step": 4180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001728687220015691,
+      "loss": 1.1206,
+      "step": 4185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001728069667637747,
+      "loss": 1.1307,
+      "step": 4190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001727451523810451,
+      "loss": 1.1246,
+      "step": 4195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017268327890359542,
+      "loss": 1.0155,
+      "step": 4200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017262134638168868,
+      "loss": 1.1233,
+      "step": 4205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017255935486563598,
+      "loss": 1.146,
+      "step": 4210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017249730440579628,
+      "loss": 1.1842,
+      "step": 4215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017243519505257644,
+      "loss": 1.1238,
+      "step": 4220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017237302685643109,
+      "loss": 1.0993,
+      "step": 4225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017231079986786287,
+      "loss": 1.2037,
+      "step": 4230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017224851413742194,
+      "loss": 1.1127,
+      "step": 4235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017218616971570634,
+      "loss": 1.1249,
+      "step": 4240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017212376665336166,
+      "loss": 1.1526,
+      "step": 4245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017206130500108124,
+      "loss": 1.0851,
+      "step": 4250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017199878480960595,
+      "loss": 1.1375,
+      "step": 4255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017193620612972427,
+      "loss": 1.1092,
+      "step": 4260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017187356901227218,
+      "loss": 1.1143,
+      "step": 4265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017181087350813306,
+      "loss": 1.1178,
+      "step": 4270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017174811966823777,
+      "loss": 1.052,
+      "step": 4275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017168530754356452,
+      "loss": 1.1377,
+      "step": 4280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000171622437185139,
+      "loss": 1.1928,
+      "step": 4285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017155950864403407,
+      "loss": 1.1823,
+      "step": 4290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017149652197136987,
+      "loss": 1.1616,
+      "step": 4295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017143347721831383,
+      "loss": 1.1134,
+      "step": 4300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017137037443608054,
+      "loss": 1.1108,
+      "step": 4305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001713072136759317,
+      "loss": 1.1527,
+      "step": 4310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017124399498917605,
+      "loss": 1.1626,
+      "step": 4315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001711807184271696,
+      "loss": 1.1243,
+      "step": 4320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001711173840413152,
+      "loss": 1.1742,
+      "step": 4325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001710539918830626,
+      "loss": 1.1432,
+      "step": 4330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017099054200390867,
+      "loss": 1.1149,
+      "step": 4335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001709270344553971,
+      "loss": 1.12,
+      "step": 4340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017086346928911838,
+      "loss": 1.1046,
+      "step": 4345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001707998465567099,
+      "loss": 1.1193,
+      "step": 4350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001707361663098557,
+      "loss": 1.0834,
+      "step": 4355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017067242860028663,
+      "loss": 1.1611,
+      "step": 4360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017060863347978015,
+      "loss": 1.1091,
+      "step": 4365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017054478100016044,
+      "loss": 1.132,
+      "step": 4370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017048087121329823,
+      "loss": 1.1149,
+      "step": 4375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017041690417111084,
+      "loss": 1.1488,
+      "step": 4380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000170352879925562,
+      "loss": 1.0606,
+      "step": 4385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017028879852866197,
+      "loss": 1.0512,
+      "step": 4390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017022466003246747,
+      "loss": 1.1841,
+      "step": 4395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017016046448908165,
+      "loss": 1.1139,
+      "step": 4400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017009621195065387,
+      "loss": 1.1809,
+      "step": 4405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001700319024693799,
+      "loss": 1.0092,
+      "step": 4410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016996753609750163,
+      "loss": 1.0473,
+      "step": 4415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016990311288730733,
+      "loss": 1.1673,
+      "step": 4420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001698386328911313,
+      "loss": 1.1659,
+      "step": 4425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016977409616135408,
+      "loss": 1.118,
+      "step": 4430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016970950275040227,
+      "loss": 1.1553,
+      "step": 4435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016964485271074844,
+      "loss": 1.1502,
+      "step": 4440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016958014609491133,
+      "loss": 1.1436,
+      "step": 4445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016951538295545545,
+      "loss": 1.1698,
+      "step": 4450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016945056334499134,
+      "loss": 1.09,
+      "step": 4455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016938568731617536,
+      "loss": 1.2259,
+      "step": 4460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016932075492170974,
+      "loss": 1.1838,
+      "step": 4465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016925576621434248,
+      "loss": 1.108,
+      "step": 4470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016919072124686732,
+      "loss": 1.0844,
+      "step": 4475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016912562007212378,
+      "loss": 1.101,
+      "step": 4480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016906046274299687,
+      "loss": 1.1597,
+      "step": 4485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016899524931241746,
+      "loss": 1.1253,
+      "step": 4490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016892997983336173,
+      "loss": 1.1217,
+      "step": 4495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016886465435885154,
+      "loss": 1.1203,
+      "step": 4500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016879927294195427,
+      "loss": 1.1046,
+      "step": 4505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016873383563578267,
+      "loss": 1.1332,
+      "step": 4510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016866834249349487,
+      "loss": 1.1025,
+      "step": 4515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016860279356829445,
+      "loss": 1.1143,
+      "step": 4520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016853718891343027,
+      "loss": 1.1672,
+      "step": 4525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016847152858219638,
+      "loss": 1.0842,
+      "step": 4530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001684058126279321,
+      "loss": 1.1651,
+      "step": 4535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016834004110402207,
+      "loss": 1.1487,
+      "step": 4540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016827421406389592,
+      "loss": 1.046,
+      "step": 4545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001682083315610284,
+      "loss": 1.1106,
+      "step": 4550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016814239364893935,
+      "loss": 1.1432,
+      "step": 4555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016807640038119363,
+      "loss": 1.1026,
+      "step": 4560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016801035181140101,
+      "loss": 1.1978,
+      "step": 4565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001679442479932163,
+      "loss": 1.0761,
+      "step": 4570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016787808898033901,
+      "loss": 1.1077,
+      "step": 4575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016781187482651364,
+      "loss": 1.0634,
+      "step": 4580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016774560558552946,
+      "loss": 1.1863,
+      "step": 4585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016767928131122047,
+      "loss": 1.1161,
+      "step": 4590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016761290205746535,
+      "loss": 1.0533,
+      "step": 4595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016754646787818747,
+      "loss": 1.1095,
+      "step": 4600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016747997882735478,
+      "loss": 1.136,
+      "step": 4605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001674134349589799,
+      "loss": 1.0895,
+      "step": 4610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016734683632711986,
+      "loss": 1.1445,
+      "step": 4615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016728018298587622,
+      "loss": 1.1301,
+      "step": 4620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016721347498939513,
+      "loss": 1.1773,
+      "step": 4625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016714671239186692,
+      "loss": 1.1557,
+      "step": 4630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016707989524752636,
+      "loss": 1.1129,
+      "step": 4635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016701302361065253,
+      "loss": 1.1682,
+      "step": 4640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016694609753556885,
+      "loss": 1.0927,
+      "step": 4645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016687911707664275,
+      "loss": 1.1254,
+      "step": 4650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016681208228828612,
+      "loss": 1.0745,
+      "step": 4655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001667449932249548,
+      "loss": 1.1282,
+      "step": 4660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016667784994114882,
+      "loss": 1.147,
+      "step": 4665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016661065249141217,
+      "loss": 1.0956,
+      "step": 4670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001665434009303329,
+      "loss": 1.174,
+      "step": 4675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016647609531254294,
+      "loss": 1.1663,
+      "step": 4680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016640873569271825,
+      "loss": 1.155,
+      "step": 4685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016634132212557862,
+      "loss": 1.1717,
+      "step": 4690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016627385466588755,
+      "loss": 1.0635,
+      "step": 4695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016620633336845254,
+      "loss": 1.097,
+      "step": 4700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001661387582881246,
+      "loss": 1.0957,
+      "step": 4705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016607112947979852,
+      "loss": 1.1087,
+      "step": 4710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016600344699841277,
+      "loss": 1.1274,
+      "step": 4715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016593571089894938,
+      "loss": 1.1693,
+      "step": 4720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000165867921236434,
+      "loss": 1.0656,
+      "step": 4725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001658000780659357,
+      "loss": 1.143,
+      "step": 4730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016573218144256708,
+      "loss": 1.13,
+      "step": 4735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001656642314214841,
+      "loss": 1.1204,
+      "step": 4740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016559622805788622,
+      "loss": 1.113,
+      "step": 4745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016552817140701615,
+      "loss": 1.0681,
+      "step": 4750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016546006152415985,
+      "loss": 1.1235,
+      "step": 4755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001653918984646466,
+      "loss": 1.1158,
+      "step": 4760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016532368228384877,
+      "loss": 1.2022,
+      "step": 4765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016525541303718204,
+      "loss": 1.1223,
+      "step": 4770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001651870907801051,
+      "loss": 1.1205,
+      "step": 4775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001651187155681197,
+      "loss": 1.117,
+      "step": 4780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016505028745677063,
+      "loss": 1.1233,
+      "step": 4785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016498180650164566,
+      "loss": 1.1003,
+      "step": 4790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016491327275837547,
+      "loss": 1.1296,
+      "step": 4795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016484468628263362,
+      "loss": 1.1567,
+      "step": 4800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016477604713013657,
+      "loss": 1.0846,
+      "step": 4805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016470735535664346,
+      "loss": 1.1573,
+      "step": 4810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016463861101795627,
+      "loss": 1.1431,
+      "step": 4815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016456981416991965,
+      "loss": 1.0639,
+      "step": 4820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016450096486842085,
+      "loss": 1.0252,
+      "step": 4825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001644320631693898,
+      "loss": 1.074,
+      "step": 4830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000164363109128799,
+      "loss": 1.1785,
+      "step": 4835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001642941028026635,
+      "loss": 1.1214,
+      "step": 4840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016422504424704079,
+      "loss": 1.0599,
+      "step": 4845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001641559335180307,
+      "loss": 1.0839,
+      "step": 4850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001640867706717755,
+      "loss": 1.1294,
+      "step": 4855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016401755576445982,
+      "loss": 1.1001,
+      "step": 4860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016394828885231064,
+      "loss": 1.1101,
+      "step": 4865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016387896999159706,
+      "loss": 1.1239,
+      "step": 4870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016380959923863045,
+      "loss": 1.1619,
+      "step": 4875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016374017664976437,
+      "loss": 1.1858,
+      "step": 4880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001636707022813944,
+      "loss": 1.1114,
+      "step": 4885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016360117618995822,
+      "loss": 1.0545,
+      "step": 4890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016353159843193564,
+      "loss": 1.0748,
+      "step": 4895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001634619690638482,
+      "loss": 1.1413,
+      "step": 4900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016339228814225962,
+      "loss": 1.1231,
+      "step": 4905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016332255572377533,
+      "loss": 1.0801,
+      "step": 4910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016325277186504268,
+      "loss": 1.1022,
+      "step": 4915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016318293662275068,
+      "loss": 1.1441,
+      "step": 4920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016311305005363032,
+      "loss": 1.0301,
+      "step": 4925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016304311221445404,
+      "loss": 1.0874,
+      "step": 4930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016297312316203603,
+      "loss": 1.1444,
+      "step": 4935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016290308295323212,
+      "loss": 1.1124,
+      "step": 4940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016283299164493965,
+      "loss": 1.0805,
+      "step": 4945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016276284929409753,
+      "loss": 1.0188,
+      "step": 4950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016269265595768604,
+      "loss": 1.1557,
+      "step": 4955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016262241169272688,
+      "loss": 1.1399,
+      "step": 4960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016255211655628326,
+      "loss": 1.1706,
+      "step": 4965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016248177060545953,
+      "loss": 1.1087,
+      "step": 4970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016241137389740148,
+      "loss": 1.062,
+      "step": 4975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016234092648929614,
+      "loss": 1.1184,
+      "step": 4980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016227042843837148,
+      "loss": 1.0532,
+      "step": 4985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001621998798018969,
+      "loss": 1.1242,
+      "step": 4990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016212928063718277,
+      "loss": 1.1815,
+      "step": 4995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016205863100158044,
+      "loss": 1.1122,
+      "step": 5000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016198793095248239,
+      "loss": 1.064,
+      "step": 5005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016191718054732189,
+      "loss": 1.0185,
+      "step": 5010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016184637984357337,
+      "loss": 1.1035,
+      "step": 5015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016177552889875188,
+      "loss": 1.1197,
+      "step": 5020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016170462777041338,
+      "loss": 1.1555,
+      "step": 5025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016163367651615465,
+      "loss": 1.1192,
+      "step": 5030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016156267519361304,
+      "loss": 1.1748,
+      "step": 5035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016149162386046677,
+      "loss": 1.1335,
+      "step": 5040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016142052257443455,
+      "loss": 1.1156,
+      "step": 5045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016134937139327567,
+      "loss": 1.1977,
+      "step": 5050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016127817037479004,
+      "loss": 1.182,
+      "step": 5055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016120691957681796,
+      "loss": 1.1381,
+      "step": 5060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001611356190572402,
+      "loss": 1.1374,
+      "step": 5065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016106426887397787,
+      "loss": 1.1175,
+      "step": 5070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001609928690849926,
+      "loss": 1.0858,
+      "step": 5075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016092141974828614,
+      "loss": 1.0358,
+      "step": 5080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016084992092190052,
+      "loss": 1.1384,
+      "step": 5085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016077837266391807,
+      "loss": 1.1729,
+      "step": 5090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001607067750324612,
+      "loss": 1.1303,
+      "step": 5095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016063512808569243,
+      "loss": 1.078,
+      "step": 5100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016056343188181432,
+      "loss": 1.0699,
+      "step": 5105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001604916864790695,
+      "loss": 1.033,
+      "step": 5110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016041989193574054,
+      "loss": 1.1381,
+      "step": 5115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001603480483101499,
+      "loss": 1.0595,
+      "step": 5120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016027615566066,
+      "loss": 1.1194,
+      "step": 5125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016020421404567298,
+      "loss": 1.119,
+      "step": 5130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016013222352363082,
+      "loss": 1.1099,
+      "step": 5135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016006018415301517,
+      "loss": 1.1056,
+      "step": 5140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001599880959923475,
+      "loss": 1.1328,
+      "step": 5145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015991595910018873,
+      "loss": 1.0698,
+      "step": 5150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015984377353513951,
+      "loss": 1.2037,
+      "step": 5155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015977153935584,
+      "loss": 1.125,
+      "step": 5160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015969925662096977,
+      "loss": 1.1111,
+      "step": 5165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015962692538924793,
+      "loss": 1.1391,
+      "step": 5170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015955454571943292,
+      "loss": 1.1056,
+      "step": 5175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001594821176703226,
+      "loss": 1.1203,
+      "step": 5180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015940964130075404,
+      "loss": 1.0924,
+      "step": 5185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001593371166696037,
+      "loss": 1.0482,
+      "step": 5190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015926454383578708,
+      "loss": 1.1541,
+      "step": 5195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015919192285825896,
+      "loss": 1.1934,
+      "step": 5200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015911925379601317,
+      "loss": 1.122,
+      "step": 5205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001590465367080826,
+      "loss": 1.1165,
+      "step": 5210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015897377165353916,
+      "loss": 1.1322,
+      "step": 5215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015890095869149377,
+      "loss": 1.1082,
+      "step": 5220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001588280978810962,
+      "loss": 1.1623,
+      "step": 5225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015875518928153525,
+      "loss": 1.1047,
+      "step": 5230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015868223295203824,
+      "loss": 1.0627,
+      "step": 5235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015860922895187155,
+      "loss": 1.1736,
+      "step": 5240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015853617734034014,
+      "loss": 1.135,
+      "step": 5245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001584630781767877,
+      "loss": 1.1172,
+      "step": 5250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015838993152059654,
+      "loss": 1.1151,
+      "step": 5255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001583167374311875,
+      "loss": 1.0156,
+      "step": 5260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015824349596802004,
+      "loss": 1.1202,
+      "step": 5265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015817020719059206,
+      "loss": 1.1134,
+      "step": 5270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015809687115843983,
+      "loss": 1.0793,
+      "step": 5275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015802348793113815,
+      "loss": 1.1246,
+      "step": 5280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001579500575683,
+      "loss": 1.1475,
+      "step": 5285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001578765801295768,
+      "loss": 1.1283,
+      "step": 5290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001578030556746582,
+      "loss": 1.1768,
+      "step": 5295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015772948426327188,
+      "loss": 1.0936,
+      "step": 5300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001576558659551838,
+      "loss": 1.1402,
+      "step": 5305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015758220081019798,
+      "loss": 1.0974,
+      "step": 5310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015750848888815654,
+      "loss": 1.1174,
+      "step": 5315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015743473024893952,
+      "loss": 1.1125,
+      "step": 5320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000157360924952465,
+      "loss": 1.1624,
+      "step": 5325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015728707305868886,
+      "loss": 1.1312,
+      "step": 5330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015721317462760493,
+      "loss": 1.0624,
+      "step": 5335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015713922971924475,
+      "loss": 1.0681,
+      "step": 5340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015706523839367767,
+      "loss": 1.1555,
+      "step": 5345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001569912007110108,
+      "loss": 1.1092,
+      "step": 5350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015691711673138876,
+      "loss": 1.1266,
+      "step": 5355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015684298651499394,
+      "loss": 1.1312,
+      "step": 5360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015676881012204627,
+      "loss": 1.0635,
+      "step": 5365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015669458761280298,
+      "loss": 1.0268,
+      "step": 5370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015662031904755902,
+      "loss": 1.1884,
+      "step": 5375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015654600448664665,
+      "loss": 1.0921,
+      "step": 5380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015647164399043556,
+      "loss": 1.0646,
+      "step": 5385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015639723761933264,
+      "loss": 1.098,
+      "step": 5390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001563227854337821,
+      "loss": 1.1172,
+      "step": 5395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001562482874942654,
+      "loss": 1.1878,
+      "step": 5400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015617374386130113,
+      "loss": 1.104,
+      "step": 5405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000156099154595445,
+      "loss": 1.1744,
+      "step": 5410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015602451975728977,
+      "loss": 1.2066,
+      "step": 5415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015594983940746534,
+      "loss": 1.2062,
+      "step": 5420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015587511360663845,
+      "loss": 1.0693,
+      "step": 5425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015580034241551274,
+      "loss": 1.1161,
+      "step": 5430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015572552589482896,
+      "loss": 1.1196,
+      "step": 5435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015565066410536433,
+      "loss": 1.1331,
+      "step": 5440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001555757571079331,
+      "loss": 1.1623,
+      "step": 5445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015550080496338616,
+      "loss": 1.1009,
+      "step": 5450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015542580773261113,
+      "loss": 1.1545,
+      "step": 5455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015535076547653216,
+      "loss": 1.0772,
+      "step": 5460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015527567825611004,
+      "loss": 1.186,
+      "step": 5465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015520054613234207,
+      "loss": 1.0907,
+      "step": 5470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015512536916626203,
+      "loss": 1.095,
+      "step": 5475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015505014741894017,
+      "loss": 1.1298,
+      "step": 5480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000154974880951483,
+      "loss": 1.0656,
+      "step": 5485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015489956982503352,
+      "loss": 1.1046,
+      "step": 5490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015482421410077088,
+      "loss": 1.1659,
+      "step": 5495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015474881383991055,
+      "loss": 1.1071,
+      "step": 5500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015467336910370406,
+      "loss": 1.124,
+      "step": 5505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015459787995343919,
+      "loss": 1.1196,
+      "step": 5510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001545223464504398,
+      "loss": 1.107,
+      "step": 5515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015444676865606567,
+      "loss": 1.0854,
+      "step": 5520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015437114663171265,
+      "loss": 1.1515,
+      "step": 5525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015429548043881244,
+      "loss": 1.0814,
+      "step": 5530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015421977013883273,
+      "loss": 1.096,
+      "step": 5535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000154144015793277,
+      "loss": 1.1135,
+      "step": 5540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015406821746368445,
+      "loss": 1.0747,
+      "step": 5545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015399237521163012,
+      "loss": 1.1497,
+      "step": 5550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001539164890987246,
+      "loss": 1.0892,
+      "step": 5555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015384055918661423,
+      "loss": 1.1524,
+      "step": 5560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015376458553698083,
+      "loss": 1.0494,
+      "step": 5565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015368856821154186,
+      "loss": 1.1091,
+      "step": 5570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001536125072720501,
+      "loss": 1.1346,
+      "step": 5575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015353640278029397,
+      "loss": 1.134,
+      "step": 5580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015346025479809705,
+      "loss": 1.1815,
+      "step": 5585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015338406338731843,
+      "loss": 1.1357,
+      "step": 5590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001533078286098524,
+      "loss": 1.088,
+      "step": 5595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001532315505276285,
+      "loss": 1.1526,
+      "step": 5600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001531552292026114,
+      "loss": 1.066,
+      "step": 5605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001530788646968009,
+      "loss": 1.1866,
+      "step": 5610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015300245707223196,
+      "loss": 1.123,
+      "step": 5615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015292600639097453,
+      "loss": 1.1146,
+      "step": 5620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015284951271513355,
+      "loss": 1.1022,
+      "step": 5625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015277297610684883,
+      "loss": 1.1311,
+      "step": 5630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015269639662829506,
+      "loss": 1.12,
+      "step": 5635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015261977434168186,
+      "loss": 1.0553,
+      "step": 5640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015254310930925348,
+      "loss": 1.1149,
+      "step": 5645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000152466401593289,
+      "loss": 1.1002,
+      "step": 5650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015238965125610217,
+      "loss": 1.1554,
+      "step": 5655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015231285836004126,
+      "loss": 1.1515,
+      "step": 5660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001522360229674892,
+      "loss": 1.0392,
+      "step": 5665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001521591451408634,
+      "loss": 1.1207,
+      "step": 5670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015208222494261586,
+      "loss": 1.0967,
+      "step": 5675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015200526243523287,
+      "loss": 1.1501,
+      "step": 5680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001519282576812351,
+      "loss": 1.1713,
+      "step": 5685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001518512107431776,
+      "loss": 1.2065,
+      "step": 5690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015177412168364962,
+      "loss": 1.2021,
+      "step": 5695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015169699056527474,
+      "loss": 1.1082,
+      "step": 5700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015161981745071056,
+      "loss": 1.1254,
+      "step": 5705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001515426024026489,
+      "loss": 1.1249,
+      "step": 5710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015146534548381562,
+      "loss": 1.1814,
+      "step": 5715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015138804675697055,
+      "loss": 1.2072,
+      "step": 5720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015131070628490753,
+      "loss": 1.114,
+      "step": 5725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001512333241304543,
+      "loss": 1.0971,
+      "step": 5730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015115590035647247,
+      "loss": 1.1448,
+      "step": 5735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015107843502585748,
+      "loss": 1.1395,
+      "step": 5740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015100092820153848,
+      "loss": 1.0912,
+      "step": 5745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001509233799464783,
+      "loss": 1.1447,
+      "step": 5750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015084579032367354,
+      "loss": 1.1428,
+      "step": 5755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015076815939615428,
+      "loss": 1.1153,
+      "step": 5760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001506904872269843,
+      "loss": 1.0457,
+      "step": 5765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015061277387926072,
+      "loss": 1.1668,
+      "step": 5770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015053501941611418,
+      "loss": 1.0989,
+      "step": 5775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015045722390070874,
+      "loss": 1.0823,
+      "step": 5780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015037938739624184,
+      "loss": 1.1403,
+      "step": 5785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015030150996594413,
+      "loss": 1.1209,
+      "step": 5790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015022359167307956,
+      "loss": 1.1237,
+      "step": 5795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015014563258094527,
+      "loss": 1.1439,
+      "step": 5800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015006763275287154,
+      "loss": 1.0948,
+      "step": 5805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014998959225222177,
+      "loss": 1.1501,
+      "step": 5810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001499115111423923,
+      "loss": 1.1745,
+      "step": 5815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014983338948681255,
+      "loss": 1.1036,
+      "step": 5820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014975522734894488,
+      "loss": 1.1006,
+      "step": 5825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001496770247922845,
+      "loss": 1.0889,
+      "step": 5830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014959878188035945,
+      "loss": 1.1197,
+      "step": 5835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001495204986767306,
+      "loss": 1.1228,
+      "step": 5840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014944217524499154,
+      "loss": 1.1758,
+      "step": 5845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001493638116487685,
+      "loss": 1.1087,
+      "step": 5850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014928540795172032,
+      "loss": 1.0652,
+      "step": 5855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001492069642175385,
+      "loss": 1.1196,
+      "step": 5860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000149128480509947,
+      "loss": 1.1953,
+      "step": 5865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014904995689270223,
+      "loss": 1.1149,
+      "step": 5870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014897139342959318,
+      "loss": 1.1315,
+      "step": 5875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014889279018444102,
+      "loss": 1.0991,
+      "step": 5880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014881414722109929,
+      "loss": 1.0311,
+      "step": 5885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014873546460345376,
+      "loss": 1.1245,
+      "step": 5890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014865674239542256,
+      "loss": 1.0856,
+      "step": 5895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014857798066095584,
+      "loss": 1.126,
+      "step": 5900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014849917946403596,
+      "loss": 1.1151,
+      "step": 5905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001484203388686773,
+      "loss": 1.1623,
+      "step": 5910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014834145893892614,
+      "loss": 1.1678,
+      "step": 5915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001482625397388608,
+      "loss": 1.1427,
+      "step": 5920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001481835813325916,
+      "loss": 1.1141,
+      "step": 5925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014810458378426055,
+      "loss": 1.0961,
+      "step": 5930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014802554715804157,
+      "loss": 1.1634,
+      "step": 5935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014794647151814022,
+      "loss": 1.1194,
+      "step": 5940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001478673569287938,
+      "loss": 1.155,
+      "step": 5945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001477882034542713,
+      "loss": 1.1072,
+      "step": 5950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001477090111588732,
+      "loss": 1.078,
+      "step": 5955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014762978010693167,
+      "loss": 1.1677,
+      "step": 5960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014755051036281013,
+      "loss": 1.181,
+      "step": 5965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014747120199090375,
+      "loss": 1.1694,
+      "step": 5970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001473918550556388,
+      "loss": 1.1733,
+      "step": 5975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014731246962147292,
+      "loss": 1.114,
+      "step": 5980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014723304575289516,
+      "loss": 1.1161,
+      "step": 5985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014715358351442573,
+      "loss": 1.0877,
+      "step": 5990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014707408297061591,
+      "loss": 1.1886,
+      "step": 5995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001469945441860483,
+      "loss": 1.111,
+      "step": 6000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014691496722533633,
+      "loss": 1.1828,
+      "step": 6005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001468353521531246,
+      "loss": 1.0762,
+      "step": 6010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014675569903408868,
+      "loss": 1.1449,
+      "step": 6015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014667600793293492,
+      "loss": 1.1499,
+      "step": 6020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014659627891440064,
+      "loss": 1.0797,
+      "step": 6025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014651651204325384,
+      "loss": 1.1069,
+      "step": 6030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014643670738429348,
+      "loss": 1.1029,
+      "step": 6035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000146356865002349,
+      "loss": 1.1036,
+      "step": 6040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014627698496228063,
+      "loss": 1.1002,
+      "step": 6045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001461970673289791,
+      "loss": 1.0889,
+      "step": 6050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014611711216736572,
+      "loss": 1.0716,
+      "step": 6055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014603711954239227,
+      "loss": 1.1705,
+      "step": 6060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014595708951904102,
+      "loss": 1.0456,
+      "step": 6065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014587702216232453,
+      "loss": 1.0921,
+      "step": 6070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014579691753728572,
+      "loss": 1.1518,
+      "step": 6075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014571677570899782,
+      "loss": 1.0895,
+      "step": 6080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014563659674256428,
+      "loss": 1.0871,
+      "step": 6085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001455563807031187,
+      "loss": 1.163,
+      "step": 6090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014547612765582473,
+      "loss": 1.1374,
+      "step": 6095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014539583766587624,
+      "loss": 1.1316,
+      "step": 6100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014531551079849695,
+      "loss": 1.1371,
+      "step": 6105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014523514711894066,
+      "loss": 1.1409,
+      "step": 6110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014515474669249095,
+      "loss": 1.0427,
+      "step": 6115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014507430958446142,
+      "loss": 1.1107,
+      "step": 6120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014499383586019524,
+      "loss": 1.2219,
+      "step": 6125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001449133255850656,
+      "loss": 1.142,
+      "step": 6130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014483277882447508,
+      "loss": 1.0851,
+      "step": 6135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014475219564385623,
+      "loss": 1.1343,
+      "step": 6140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014467157610867094,
+      "loss": 1.1167,
+      "step": 6145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014459092028441069,
+      "loss": 1.16,
+      "step": 6150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014451022823659646,
+      "loss": 1.1601,
+      "step": 6155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014442950003077862,
+      "loss": 1.0655,
+      "step": 6160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014434873573253697,
+      "loss": 1.1391,
+      "step": 6165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014426793540748062,
+      "loss": 1.1291,
+      "step": 6170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014418709912124796,
+      "loss": 1.1682,
+      "step": 6175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001441062269395065,
+      "loss": 1.1188,
+      "step": 6180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014402531892795305,
+      "loss": 1.0427,
+      "step": 6185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014394437515231336,
+      "loss": 1.1258,
+      "step": 6190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014386339567834243,
+      "loss": 1.0512,
+      "step": 6195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014378238057182404,
+      "loss": 1.1841,
+      "step": 6200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014370132989857112,
+      "loss": 1.105,
+      "step": 6205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014362024372442534,
+      "loss": 1.1123,
+      "step": 6210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001435391221152573,
+      "loss": 1.1275,
+      "step": 6215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014345796513696638,
+      "loss": 1.0553,
+      "step": 6220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014337677285548059,
+      "loss": 1.1397,
+      "step": 6225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001432955453367568,
+      "loss": 1.0851,
+      "step": 6230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014321428264678026,
+      "loss": 1.1905,
+      "step": 6235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001431329848515651,
+      "loss": 1.0644,
+      "step": 6240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014305165201715372,
+      "loss": 1.1171,
+      "step": 6245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014297028420961713,
+      "loss": 1.1036,
+      "step": 6250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014288888149505458,
+      "loss": 1.1324,
+      "step": 6255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014280744393959386,
+      "loss": 1.1594,
+      "step": 6260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001427259716093909,
+      "loss": 1.1471,
+      "step": 6265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014264446457063013,
+      "loss": 1.103,
+      "step": 6270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014256292288952388,
+      "loss": 1.1004,
+      "step": 6275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014248134663231274,
+      "loss": 1.171,
+      "step": 6280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001423997358652655,
+      "loss": 1.1521,
+      "step": 6285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014231809065467883,
+      "loss": 1.1031,
+      "step": 6290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014223641106687746,
+      "loss": 1.1652,
+      "step": 6295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000142154697168214,
+      "loss": 1.1709,
+      "step": 6300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014207294902506898,
+      "loss": 1.0423,
+      "step": 6305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014199116670385068,
+      "loss": 1.1511,
+      "step": 6310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014190935027099522,
+      "loss": 1.1246,
+      "step": 6315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014182749979296634,
+      "loss": 1.0853,
+      "step": 6320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001417456153362556,
+      "loss": 1.069,
+      "step": 6325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014166369696738194,
+      "loss": 1.1839,
+      "step": 6330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001415817447528921,
+      "loss": 1.0938,
+      "step": 6335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014149975875936012,
+      "loss": 1.2167,
+      "step": 6340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014141773905338752,
+      "loss": 1.0523,
+      "step": 6345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014133568570160327,
+      "loss": 1.0932,
+      "step": 6350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014125359877066362,
+      "loss": 1.076,
+      "step": 6355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014117147832725207,
+      "loss": 1.1642,
+      "step": 6360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001410893244380794,
+      "loss": 1.1171,
+      "step": 6365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001410071371698836,
+      "loss": 1.1356,
+      "step": 6370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014092491658942958,
+      "loss": 1.2037,
+      "step": 6375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001408426627635096,
+      "loss": 1.0983,
+      "step": 6380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001407603757589427,
+      "loss": 1.0695,
+      "step": 6385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014067805564257503,
+      "loss": 1.0783,
+      "step": 6390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014059570248127942,
+      "loss": 1.1243,
+      "step": 6395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014051331634195582,
+      "loss": 1.1064,
+      "step": 6400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014043089729153075,
+      "loss": 1.0942,
+      "step": 6405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014034844539695753,
+      "loss": 1.0615,
+      "step": 6410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014026596072521622,
+      "loss": 1.1113,
+      "step": 6415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014018344334331345,
+      "loss": 1.0846,
+      "step": 6420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014010089331828248,
+      "loss": 1.1403,
+      "step": 6425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000140018310717183,
+      "loss": 1.1468,
+      "step": 6430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013993569560710118,
+      "loss": 1.167,
+      "step": 6435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001398530480551497,
+      "loss": 1.0327,
+      "step": 6440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013977036812846751,
+      "loss": 1.1259,
+      "step": 6445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013968765589421982,
+      "loss": 1.1125,
+      "step": 6450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013960491141959817,
+      "loss": 1.0698,
+      "step": 6455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013952213477182025,
+      "loss": 1.1647,
+      "step": 6460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013943932601812986,
+      "loss": 1.0381,
+      "step": 6465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013935648522579695,
+      "loss": 1.0488,
+      "step": 6470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001392736124621175,
+      "loss": 1.1777,
+      "step": 6475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013919070779441333,
+      "loss": 1.1648,
+      "step": 6480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013910777129003235,
+      "loss": 1.1034,
+      "step": 6485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013902480301634818,
+      "loss": 1.0858,
+      "step": 6490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001389418030407604,
+      "loss": 1.2123,
+      "step": 6495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001388587714306942,
+      "loss": 1.0523,
+      "step": 6500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013877570825360057,
+      "loss": 1.1354,
+      "step": 6505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001386926135769561,
+      "loss": 1.1546,
+      "step": 6510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013860948746826295,
+      "loss": 1.1443,
+      "step": 6515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001385263299950489,
+      "loss": 1.1738,
+      "step": 6520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013844314122486704,
+      "loss": 1.1277,
+      "step": 6525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001383599212252961,
+      "loss": 1.1733,
+      "step": 6530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013827667006393994,
+      "loss": 1.139,
+      "step": 6535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000138193387808428,
+      "loss": 1.1888,
+      "step": 6540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001381100745264148,
+      "loss": 1.1492,
+      "step": 6545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013802673028558005,
+      "loss": 1.0426,
+      "step": 6550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013794335515362877,
+      "loss": 1.0669,
+      "step": 6555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001378599491982909,
+      "loss": 1.1074,
+      "step": 6560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013777651248732148,
+      "loss": 1.1361,
+      "step": 6565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013769304508850063,
+      "loss": 1.1386,
+      "step": 6570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013760954706963324,
+      "loss": 1.1578,
+      "step": 6575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001375260184985491,
+      "loss": 1.0831,
+      "step": 6580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013744245944310305,
+      "loss": 1.1934,
+      "step": 6585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013735886997117439,
+      "loss": 1.0961,
+      "step": 6590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013727525015066726,
+      "loss": 1.0284,
+      "step": 6595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013719160004951045,
+      "loss": 1.1238,
+      "step": 6600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013710791973565734,
+      "loss": 1.1232,
+      "step": 6605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013702420927708586,
+      "loss": 1.1575,
+      "step": 6610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013694046874179844,
+      "loss": 1.065,
+      "step": 6615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013685669819782196,
+      "loss": 1.2215,
+      "step": 6620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013677289771320755,
+      "loss": 1.1194,
+      "step": 6625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001366890673560309,
+      "loss": 1.1326,
+      "step": 6630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013660520719439165,
+      "loss": 1.0888,
+      "step": 6635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013652131729641393,
+      "loss": 1.0788,
+      "step": 6640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001364373977302459,
+      "loss": 1.1553,
+      "step": 6645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013635344856405995,
+      "loss": 1.21,
+      "step": 6650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013626946986605224,
+      "loss": 1.202,
+      "step": 6655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013618546170444322,
+      "loss": 1.0331,
+      "step": 6660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013610142414747709,
+      "loss": 1.1465,
+      "step": 6665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000136017357263422,
+      "loss": 1.0962,
+      "step": 6670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013593326112056988,
+      "loss": 1.1186,
+      "step": 6675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001358491357872365,
+      "loss": 1.1885,
+      "step": 6680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013576498133176126,
+      "loss": 1.2042,
+      "step": 6685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001356807978225073,
+      "loss": 1.1453,
+      "step": 6690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001355965853278613,
+      "loss": 1.054,
+      "step": 6695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001355123439162335,
+      "loss": 1.0556,
+      "step": 6700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013542807365605764,
+      "loss": 1.1027,
+      "step": 6705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013534377461579093,
+      "loss": 1.1701,
+      "step": 6710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013525944686391386,
+      "loss": 1.0448,
+      "step": 6715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001351750904689303,
+      "loss": 1.1077,
+      "step": 6720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013509070549936746,
+      "loss": 1.1626,
+      "step": 6725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013500629202377565,
+      "loss": 1.1037,
+      "step": 6730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001349218501107284,
+      "loss": 1.0966,
+      "step": 6735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013483737982882235,
+      "loss": 1.1237,
+      "step": 6740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001347528812466771,
+      "loss": 1.1236,
+      "step": 6745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001346683544329354,
+      "loss": 1.095,
+      "step": 6750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013458379945626274,
+      "loss": 1.0883,
+      "step": 6755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013449921638534763,
+      "loss": 1.1111,
+      "step": 6760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013441460528890127,
+      "loss": 1.152,
+      "step": 6765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013432996623565782,
+      "loss": 1.1415,
+      "step": 6770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013424529929437397,
+      "loss": 1.1539,
+      "step": 6775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001341606045338292,
+      "loss": 1.1401,
+      "step": 6780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001340758820228255,
+      "loss": 1.061,
+      "step": 6785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013399113183018744,
+      "loss": 1.1434,
+      "step": 6790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000133906354024762,
+      "loss": 1.114,
+      "step": 6795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013382154867541874,
+      "loss": 1.0926,
+      "step": 6800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013373671585104952,
+      "loss": 1.1791,
+      "step": 6805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013365185562056844,
+      "loss": 1.1145,
+      "step": 6810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013356696805291198,
+      "loss": 1.1149,
+      "step": 6815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013348205321703875,
+      "loss": 1.1729,
+      "step": 6820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001333971111819296,
+      "loss": 1.0368,
+      "step": 6825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013331214201658736,
+      "loss": 1.0923,
+      "step": 6830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013322714579003703,
+      "loss": 1.0915,
+      "step": 6835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013314212257132542,
+      "loss": 1.1232,
+      "step": 6840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001330570724295215,
+      "loss": 1.0825,
+      "step": 6845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001329719954337159,
+      "loss": 1.1598,
+      "step": 6850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001328868916530212,
+      "loss": 1.1328,
+      "step": 6855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013280176115657163,
+      "loss": 1.1662,
+      "step": 6860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013271660401352318,
+      "loss": 1.0361,
+      "step": 6865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013263142029305348,
+      "loss": 1.1472,
+      "step": 6870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013254621006436178,
+      "loss": 1.1309,
+      "step": 6875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013246097339666884,
+      "loss": 1.1358,
+      "step": 6880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013237571035921693,
+      "loss": 1.0482,
+      "step": 6885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001322904210212696,
+      "loss": 1.1341,
+      "step": 6890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013220510545211198,
+      "loss": 1.0845,
+      "step": 6895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013211976372105036,
+      "loss": 1.1466,
+      "step": 6900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001320343958974123,
+      "loss": 1.1789,
+      "step": 6905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013194900205054666,
+      "loss": 1.134,
+      "step": 6910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013186358224982325,
+      "loss": 1.1166,
+      "step": 6915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013177813656463312,
+      "loss": 1.1139,
+      "step": 6920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001316926650643883,
+      "loss": 1.1691,
+      "step": 6925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013160716781852186,
+      "loss": 1.1846,
+      "step": 6930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013152164489648767,
+      "loss": 1.103,
+      "step": 6935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013143609636776053,
+      "loss": 1.1676,
+      "step": 6940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013135052230183594,
+      "loss": 1.1054,
+      "step": 6945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001312649227682303,
+      "loss": 1.1226,
+      "step": 6950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001311792978364806,
+      "loss": 1.0258,
+      "step": 6955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013109364757614441,
+      "loss": 1.1132,
+      "step": 6960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013100797205680008,
+      "loss": 1.0548,
+      "step": 6965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013092227134804628,
+      "loss": 1.0858,
+      "step": 6970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013083654551950224,
+      "loss": 1.1066,
+      "step": 6975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013075079464080745,
+      "loss": 1.1426,
+      "step": 6980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000130665018781622,
+      "loss": 1.1513,
+      "step": 6985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013057921801162614,
+      "loss": 1.1525,
+      "step": 6990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013049339240052028,
+      "loss": 1.074,
+      "step": 6995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013040754201802508,
+      "loss": 1.1259,
+      "step": 7000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013032166693388144,
+      "loss": 1.0973,
+      "step": 7005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013023576721785013,
+      "loss": 1.0996,
+      "step": 7010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000130149842939712,
+      "loss": 1.0364,
+      "step": 7015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013006389416926793,
+      "loss": 1.1164,
+      "step": 7020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001299779209763386,
+      "loss": 1.0655,
+      "step": 7025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012989192343076453,
+      "loss": 1.0597,
+      "step": 7030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001298059016024061,
+      "loss": 1.0968,
+      "step": 7035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001297198555611434,
+      "loss": 1.1435,
+      "step": 7040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012963378537687612,
+      "loss": 1.0785,
+      "step": 7045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012954769111952363,
+      "loss": 1.1942,
+      "step": 7050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012946157285902474,
+      "loss": 1.1062,
+      "step": 7055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012937543066533804,
+      "loss": 1.0923,
+      "step": 7060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012928926460844124,
+      "loss": 1.1107,
+      "step": 7065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001292030747583316,
+      "loss": 1.1233,
+      "step": 7070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001291168611850256,
+      "loss": 1.1603,
+      "step": 7075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001290306239585592,
+      "loss": 1.0632,
+      "step": 7080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012894436314898734,
+      "loss": 1.1127,
+      "step": 7085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012885807882638428,
+      "loss": 1.139,
+      "step": 7090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001287717710608433,
+      "loss": 1.0544,
+      "step": 7095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012868543992247672,
+      "loss": 1.1379,
+      "step": 7100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001285990854814159,
+      "loss": 1.1058,
+      "step": 7105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012851270780781112,
+      "loss": 1.1533,
+      "step": 7110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012842630697183145,
+      "loss": 1.1677,
+      "step": 7115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001283398830436649,
+      "loss": 1.1072,
+      "step": 7120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012825343609351822,
+      "loss": 1.1045,
+      "step": 7125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012816696619161668,
+      "loss": 1.1754,
+      "step": 7130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012808047340820452,
+      "loss": 1.1206,
+      "step": 7135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012799395781354426,
+      "loss": 1.1517,
+      "step": 7140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001279074194779171,
+      "loss": 1.1024,
+      "step": 7145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012782085847162273,
+      "loss": 1.1469,
+      "step": 7150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012773427486497912,
+      "loss": 1.0741,
+      "step": 7155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012764766872832285,
+      "loss": 1.1489,
+      "step": 7160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012756104013200847,
+      "loss": 1.0851,
+      "step": 7165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012747438914640906,
+      "loss": 1.0699,
+      "step": 7170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001273877158419158,
+      "loss": 1.0624,
+      "step": 7175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012730102028893792,
+      "loss": 1.1018,
+      "step": 7180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001272143025579028,
+      "loss": 1.0596,
+      "step": 7185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012712756271925584,
+      "loss": 1.1285,
+      "step": 7190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012704080084346037,
+      "loss": 1.0589,
+      "step": 7195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012695401700099765,
+      "loss": 1.0736,
+      "step": 7200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012686721126236672,
+      "loss": 1.035,
+      "step": 7205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012678038369808448,
+      "loss": 1.141,
+      "step": 7210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012669353437868555,
+      "loss": 1.1144,
+      "step": 7215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012660666337472223,
+      "loss": 1.1234,
+      "step": 7220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012651977075676436,
+      "loss": 1.1572,
+      "step": 7225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012643285659539943,
+      "loss": 1.1423,
+      "step": 7230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001263459209612324,
+      "loss": 1.1231,
+      "step": 7235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001262589639248856,
+      "loss": 1.1774,
+      "step": 7240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012617198555699887,
+      "loss": 1.0522,
+      "step": 7245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012608498592822932,
+      "loss": 1.0903,
+      "step": 7250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012599796510925134,
+      "loss": 1.0749,
+      "step": 7255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001259109231707565,
+      "loss": 1.1955,
+      "step": 7260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012582386018345358,
+      "loss": 1.0992,
+      "step": 7265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012573677621806838,
+      "loss": 1.1025,
+      "step": 7270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012564967134534385,
+      "loss": 1.1453,
+      "step": 7275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001255625456360398,
+      "loss": 1.1364,
+      "step": 7280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012547539916093307,
+      "loss": 1.1715,
+      "step": 7285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012538823199081733,
+      "loss": 1.035,
+      "step": 7290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012530104419650312,
+      "loss": 1.0472,
+      "step": 7295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012521383584881767,
+      "loss": 1.0919,
+      "step": 7300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012512660701860485,
+      "loss": 1.0935,
+      "step": 7305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012503935777672526,
+      "loss": 1.1187,
+      "step": 7310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000124952088194056,
+      "loss": 1.0953,
+      "step": 7315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012486479834149087,
+      "loss": 1.1204,
+      "step": 7320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012477748828993994,
+      "loss": 1.1504,
+      "step": 7325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001246901581103298,
+      "loss": 1.1013,
+      "step": 7330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012460280787360327,
+      "loss": 1.0877,
+      "step": 7335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012451543765071968,
+      "loss": 1.0486,
+      "step": 7340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012442804751265437,
+      "loss": 1.015,
+      "step": 7345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012434063753039902,
+      "loss": 1.211,
+      "step": 7350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012425320777496136,
+      "loss": 1.0825,
+      "step": 7355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012416575831736512,
+      "loss": 1.1927,
+      "step": 7360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012407828922865016,
+      "loss": 1.1035,
+      "step": 7365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012399080057987218,
+      "loss": 1.064,
+      "step": 7370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001239032924421029,
+      "loss": 1.1207,
+      "step": 7375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012381576488642969,
+      "loss": 1.0704,
+      "step": 7380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012372821798395587,
+      "loss": 1.0769,
+      "step": 7385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012364065180580033,
+      "loss": 1.1172,
+      "step": 7390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012355306642309783,
+      "loss": 1.1979,
+      "step": 7395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012346546190699845,
+      "loss": 1.1797,
+      "step": 7400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012337783832866802,
+      "loss": 1.1502,
+      "step": 7405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012329019575928773,
+      "loss": 1.141,
+      "step": 7410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012320253427005435,
+      "loss": 1.1396,
+      "step": 7415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001231148539321798,
+      "loss": 1.1164,
+      "step": 7420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012302715481689154,
+      "loss": 1.1889,
+      "step": 7425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001229394369954322,
+      "loss": 1.165,
+      "step": 7430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001228517005390595,
+      "loss": 1.1114,
+      "step": 7435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012276394551904646,
+      "loss": 1.1685,
+      "step": 7440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000122676172006681,
+      "loss": 1.1438,
+      "step": 7445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012258838007326634,
+      "loss": 1.0657,
+      "step": 7450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012250056979012033,
+      "loss": 1.1096,
+      "step": 7455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012241274122857601,
+      "loss": 1.2047,
+      "step": 7460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012232489445998106,
+      "loss": 1.1617,
+      "step": 7465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001222370295556981,
+      "loss": 1.1167,
+      "step": 7470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012214914658710438,
+      "loss": 1.2007,
+      "step": 7475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012206124562559188,
+      "loss": 1.111,
+      "step": 7480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012197332674256717,
+      "loss": 1.0913,
+      "step": 7485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012188539000945141,
+      "loss": 1.0143,
+      "step": 7490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012179743549768025,
+      "loss": 1.1422,
+      "step": 7495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012170946327870377,
+      "loss": 1.0883,
+      "step": 7500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012162147342398645,
+      "loss": 1.0899,
+      "step": 7505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012153346600500702,
+      "loss": 1.1782,
+      "step": 7510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012144544109325861,
+      "loss": 1.1643,
+      "step": 7515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012135739876024845,
+      "loss": 1.0894,
+      "step": 7520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012126933907749802,
+      "loss": 1.0882,
+      "step": 7525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012118126211654275,
+      "loss": 1.0165,
+      "step": 7530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012109316794893234,
+      "loss": 1.1583,
+      "step": 7535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012100505664623017,
+      "loss": 1.1101,
+      "step": 7540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012091692828001379,
+      "loss": 1.0364,
+      "step": 7545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012082878292187446,
+      "loss": 1.1058,
+      "step": 7550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012074062064341734,
+      "loss": 1.087,
+      "step": 7555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012065244151626128,
+      "loss": 1.1542,
+      "step": 7560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012056424561203881,
+      "loss": 1.0439,
+      "step": 7565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012047603300239615,
+      "loss": 1.1374,
+      "step": 7570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012038780375899301,
+      "loss": 1.144,
+      "step": 7575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001202995579535027,
+      "loss": 1.0997,
+      "step": 7580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001202112956576119,
+      "loss": 1.0783,
+      "step": 7585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012012301694302073,
+      "loss": 1.2033,
+      "step": 7590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012003472188144268,
+      "loss": 1.1214,
+      "step": 7595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011994641054460442,
+      "loss": 1.0755,
+      "step": 7600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011985808300424591,
+      "loss": 1.1237,
+      "step": 7605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011976973933212029,
+      "loss": 1.0691,
+      "step": 7610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001196813795999938,
+      "loss": 0.9849,
+      "step": 7615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011959300387964566,
+      "loss": 1.1364,
+      "step": 7620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011950461224286811,
+      "loss": 1.168,
+      "step": 7625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011941620476146639,
+      "loss": 1.069,
+      "step": 7630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011932778150725853,
+      "loss": 1.1551,
+      "step": 7635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001192393425520754,
+      "loss": 1.2237,
+      "step": 7640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011915088796776064,
+      "loss": 1.09,
+      "step": 7645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011906241782617053,
+      "loss": 1.028,
+      "step": 7650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011897393219917407,
+      "loss": 1.1026,
+      "step": 7655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011888543115865275,
+      "loss": 1.2062,
+      "step": 7660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011879691477650066,
+      "loss": 1.074,
+      "step": 7665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001187083831246243,
+      "loss": 1.0966,
+      "step": 7670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011861983627494263,
+      "loss": 1.1676,
+      "step": 7675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011853127429938691,
+      "loss": 1.1287,
+      "step": 7680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011844269726990065,
+      "loss": 1.1584,
+      "step": 7685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011835410525843971,
+      "loss": 1.0942,
+      "step": 7690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000118265498336972,
+      "loss": 1.0862,
+      "step": 7695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011817687657747764,
+      "loss": 1.1383,
+      "step": 7700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001180882400519487,
+      "loss": 1.2052,
+      "step": 7705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011799958883238936,
+      "loss": 1.0728,
+      "step": 7710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011791092299081557,
+      "loss": 1.0966,
+      "step": 7715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011782224259925539,
+      "loss": 1.0579,
+      "step": 7720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011773354772974851,
+      "loss": 1.1445,
+      "step": 7725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011764483845434648,
+      "loss": 1.0766,
+      "step": 7730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011755611484511243,
+      "loss": 1.0613,
+      "step": 7735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011746737697412134,
+      "loss": 1.1561,
+      "step": 7740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011737862491345958,
+      "loss": 1.1509,
+      "step": 7745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011728985873522515,
+      "loss": 1.1618,
+      "step": 7750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011720107851152748,
+      "loss": 1.1325,
+      "step": 7755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011711228431448747,
+      "loss": 1.0781,
+      "step": 7760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011702347621623727,
+      "loss": 1.1156,
+      "step": 7765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011693465428892038,
+      "loss": 1.0532,
+      "step": 7770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011684581860469158,
+      "loss": 1.1063,
+      "step": 7775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011675696923571672,
+      "loss": 1.1312,
+      "step": 7780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011666810625417287,
+      "loss": 1.0498,
+      "step": 7785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001165792297322481,
+      "loss": 1.0892,
+      "step": 7790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011649033974214147,
+      "loss": 1.1588,
+      "step": 7795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011640143635606302,
+      "loss": 1.1111,
+      "step": 7800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011631251964623364,
+      "loss": 1.0888,
+      "step": 7805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011622358968488508,
+      "loss": 1.0872,
+      "step": 7810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011613464654425983,
+      "loss": 1.1512,
+      "step": 7815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011604569029661104,
+      "loss": 1.0845,
+      "step": 7820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011595672101420262,
+      "loss": 1.1367,
+      "step": 7825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011586773876930899,
+      "loss": 1.1181,
+      "step": 7830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011577874363421512,
+      "loss": 1.0756,
+      "step": 7835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011568973568121645,
+      "loss": 1.1412,
+      "step": 7840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011560071498261881,
+      "loss": 1.0736,
+      "step": 7845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011551168161073839,
+      "loss": 1.1778,
+      "step": 7850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011542263563790174,
+      "loss": 1.0603,
+      "step": 7855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011533357713644553,
+      "loss": 1.1548,
+      "step": 7860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011524450617871672,
+      "loss": 1.1176,
+      "step": 7865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011515542283707233,
+      "loss": 1.1566,
+      "step": 7870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011506632718387944,
+      "loss": 1.154,
+      "step": 7875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011497721929151517,
+      "loss": 1.1619,
+      "step": 7880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011488809923236655,
+      "loss": 1.1674,
+      "step": 7885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011479896707883044,
+      "loss": 1.1822,
+      "step": 7890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001147098229033137,
+      "loss": 1.0471,
+      "step": 7895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011462066677823272,
+      "loss": 1.0724,
+      "step": 7900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011453149877601381,
+      "loss": 1.0871,
+      "step": 7905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011444231896909281,
+      "loss": 1.1505,
+      "step": 7910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011435312742991515,
+      "loss": 1.0695,
+      "step": 7915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011426392423093583,
+      "loss": 1.1652,
+      "step": 7920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011417470944461934,
+      "loss": 1.116,
+      "step": 7925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011408548314343948,
+      "loss": 1.1355,
+      "step": 7930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011399624539987953,
+      "loss": 1.1658,
+      "step": 7935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011390699628643199,
+      "loss": 1.1076,
+      "step": 7940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011381773587559867,
+      "loss": 1.1546,
+      "step": 7945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011372846423989039,
+      "loss": 1.1903,
+      "step": 7950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011363918145182732,
+      "loss": 1.0247,
+      "step": 7955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011354988758393852,
+      "loss": 1.1272,
+      "step": 7960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011346058270876213,
+      "loss": 1.0561,
+      "step": 7965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001133712668988452,
+      "loss": 1.1478,
+      "step": 7970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011328194022674359,
+      "loss": 1.1178,
+      "step": 7975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001131926027650222,
+      "loss": 1.1229,
+      "step": 7980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011310325458625445,
+      "loss": 1.0693,
+      "step": 7985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011301389576302265,
+      "loss": 1.1717,
+      "step": 7990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001129245263679176,
+      "loss": 1.1281,
+      "step": 7995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011283514647353882,
+      "loss": 1.1596,
+      "step": 8000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011274575615249428,
+      "loss": 1.1106,
+      "step": 8005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011265635547740047,
+      "loss": 1.0998,
+      "step": 8010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011256694452088229,
+      "loss": 1.1486,
+      "step": 8015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011247752335557291,
+      "loss": 1.2001,
+      "step": 8020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011238809205411388,
+      "loss": 1.1696,
+      "step": 8025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011229865068915498,
+      "loss": 1.0776,
+      "step": 8030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011220919933335415,
+      "loss": 1.1891,
+      "step": 8035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011211973805937738,
+      "loss": 1.118,
+      "step": 8040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011203026693989887,
+      "loss": 1.0707,
+      "step": 8045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011194078604760059,
+      "loss": 1.1028,
+      "step": 8050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011185129545517271,
+      "loss": 1.0917,
+      "step": 8055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011176179523531304,
+      "loss": 1.0812,
+      "step": 8060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011167228546072745,
+      "loss": 1.0958,
+      "step": 8065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011158276620412938,
+      "loss": 1.1227,
+      "step": 8070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011149323753824004,
+      "loss": 1.1266,
+      "step": 8075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011140369953578825,
+      "loss": 1.1608,
+      "step": 8080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011131415226951053,
+      "loss": 1.1437,
+      "step": 8085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011122459581215079,
+      "loss": 1.1049,
+      "step": 8090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011113503023646044,
+      "loss": 1.058,
+      "step": 8095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011104545561519839,
+      "loss": 1.1221,
+      "step": 8100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011095587202113073,
+      "loss": 1.1238,
+      "step": 8105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011086627952703098,
+      "loss": 1.0795,
+      "step": 8110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011077667820567983,
+      "loss": 1.0916,
+      "step": 8115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011068706812986519,
+      "loss": 1.1748,
+      "step": 8120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000110597449372382,
+      "loss": 1.0677,
+      "step": 8125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011050782200603235,
+      "loss": 1.0826,
+      "step": 8130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011041818610362524,
+      "loss": 1.1312,
+      "step": 8135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011032854173797663,
+      "loss": 1.1722,
+      "step": 8140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011023888898190938,
+      "loss": 1.1446,
+      "step": 8145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011014922790825312,
+      "loss": 1.0922,
+      "step": 8150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011005955858984429,
+      "loss": 1.1422,
+      "step": 8155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010996988109952601,
+      "loss": 1.1008,
+      "step": 8160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010988019551014796,
+      "loss": 1.1327,
+      "step": 8165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010979050189456655,
+      "loss": 1.1599,
+      "step": 8170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010970080032564465,
+      "loss": 1.1243,
+      "step": 8175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001096110908762515,
+      "loss": 1.0868,
+      "step": 8180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010952137361926283,
+      "loss": 1.1039,
+      "step": 8185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010943164862756068,
+      "loss": 1.1166,
+      "step": 8190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010934191597403346,
+      "loss": 1.1886,
+      "step": 8195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010925217573157567,
+      "loss": 1.1081,
+      "step": 8200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010916242797308802,
+      "loss": 1.0353,
+      "step": 8205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010907267277147739,
+      "loss": 1.0747,
+      "step": 8210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010898291019965666,
+      "loss": 1.1725,
+      "step": 8215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010889314033054463,
+      "loss": 1.1889,
+      "step": 8220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010880336323706617,
+      "loss": 1.083,
+      "step": 8225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010871357899215189,
+      "loss": 1.0201,
+      "step": 8230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010862378766873829,
+      "loss": 1.1199,
+      "step": 8235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010853398933976762,
+      "loss": 1.0812,
+      "step": 8240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010844418407818777,
+      "loss": 1.0576,
+      "step": 8245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001083543719569523,
+      "loss": 1.1362,
+      "step": 8250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010826455304902028,
+      "loss": 1.1578,
+      "step": 8255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010817472742735639,
+      "loss": 1.1307,
+      "step": 8260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010808489516493067,
+      "loss": 1.0221,
+      "step": 8265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010799505633471865,
+      "loss": 1.084,
+      "step": 8270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010790521100970109,
+      "loss": 1.2108,
+      "step": 8275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010781535926286415,
+      "loss": 1.0533,
+      "step": 8280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010772550116719907,
+      "loss": 1.0747,
+      "step": 8285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010763563679570236,
+      "loss": 1.0919,
+      "step": 8290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010754576622137557,
+      "loss": 1.1159,
+      "step": 8295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010745588951722532,
+      "loss": 1.0903,
+      "step": 8300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00010736600675626317,
+      "loss": 1.2006,
+      "step": 8305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001072761180115056,
+      "loss": 1.0875,
+      "step": 8310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010718622335597404,
+      "loss": 1.0446,
+      "step": 8315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001070963228626946,
+      "loss": 1.1178,
+      "step": 8320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001070064166046982,
+      "loss": 1.1285,
+      "step": 8325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010691650465502045,
+      "loss": 1.1103,
+      "step": 8330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010682658708670152,
+      "loss": 1.0727,
+      "step": 8335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001067366639727862,
+      "loss": 1.0388,
+      "step": 8340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010664673538632383,
+      "loss": 1.068,
+      "step": 8345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010655680140036806,
+      "loss": 1.0831,
+      "step": 8350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010646686208797709,
+      "loss": 1.1118,
+      "step": 8355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010637691752221326,
+      "loss": 1.0489,
+      "step": 8360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010628696777614337,
+      "loss": 1.2107,
+      "step": 8365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010619701292283831,
+      "loss": 1.1086,
+      "step": 8370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010610705303537314,
+      "loss": 1.1162,
+      "step": 8375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000106017088186827,
+      "loss": 1.1369,
+      "step": 8380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010592711845028311,
+      "loss": 1.0564,
+      "step": 8385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010583714389882863,
+      "loss": 1.0543,
+      "step": 8390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010574716460555456,
+      "loss": 1.0828,
+      "step": 8395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010565718064355594,
+      "loss": 1.1165,
+      "step": 8400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010556719208593139,
+      "loss": 1.0853,
+      "step": 8405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010547719900578339,
+      "loss": 1.1088,
+      "step": 8410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000105387201476218,
+      "loss": 1.181,
+      "step": 8415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010529719957034506,
+      "loss": 1.0139,
+      "step": 8420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010520719336127777,
+      "loss": 1.0587,
+      "step": 8425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010511718292213296,
+      "loss": 1.114,
+      "step": 8430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010502716832603086,
+      "loss": 1.0848,
+      "step": 8435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010493714964609504,
+      "loss": 1.0236,
+      "step": 8440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010484712695545241,
+      "loss": 1.0697,
+      "step": 8445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010475710032723319,
+      "loss": 1.139,
+      "step": 8450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010466706983457074,
+      "loss": 1.0962,
+      "step": 8455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010457703555060156,
+      "loss": 1.1206,
+      "step": 8460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010448699754846526,
+      "loss": 1.1129,
+      "step": 8465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010439695590130442,
+      "loss": 1.1232,
+      "step": 8470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010430691068226464,
+      "loss": 1.1069,
+      "step": 8475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010421686196449439,
+      "loss": 1.1377,
+      "step": 8480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010412680982114497,
+      "loss": 1.1074,
+      "step": 8485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010403675432537048,
+      "loss": 1.0847,
+      "step": 8490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001039466955503278,
+      "loss": 1.1333,
+      "step": 8495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010385663356917636,
+      "loss": 1.126,
+      "step": 8500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010376656845507828,
+      "loss": 1.1479,
+      "step": 8505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001036765002811982,
+      "loss": 1.1399,
+      "step": 8510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001035864291207032,
+      "loss": 1.0221,
+      "step": 8515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001034963550467629,
+      "loss": 1.0601,
+      "step": 8520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010340627813254912,
+      "loss": 1.1176,
+      "step": 8525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001033161984512362,
+      "loss": 1.0856,
+      "step": 8530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010322611607600051,
+      "loss": 1.1335,
+      "step": 8535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010313603108002078,
+      "loss": 1.2108,
+      "step": 8540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010304594353647775,
+      "loss": 1.14,
+      "step": 8545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010295585351855434,
+      "loss": 1.1015,
+      "step": 8550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010286576109943532,
+      "loss": 1.1345,
+      "step": 8555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010277566635230758,
+      "loss": 1.198,
+      "step": 8560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010268556935035981,
+      "loss": 1.1173,
+      "step": 8565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010259547016678258,
+      "loss": 1.0916,
+      "step": 8570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001025053688747682,
+      "loss": 1.0415,
+      "step": 8575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010241526554751064,
+      "loss": 1.0853,
+      "step": 8580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010232516025820567,
+      "loss": 1.1686,
+      "step": 8585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010223505308005045,
+      "loss": 1.1566,
+      "step": 8590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010214494408624389,
+      "loss": 1.1284,
+      "step": 8595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010205483334998624,
+      "loss": 1.0405,
+      "step": 8600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010196472094447919,
+      "loss": 1.1414,
+      "step": 8605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010187460694292579,
+      "loss": 1.0284,
+      "step": 8610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010178449141853041,
+      "loss": 1.0877,
+      "step": 8615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010169437444449861,
+      "loss": 1.1489,
+      "step": 8620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001016042560940372,
+      "loss": 1.1652,
+      "step": 8625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010151413644035399,
+      "loss": 1.1412,
+      "step": 8630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010142401555665801,
+      "loss": 1.1032,
+      "step": 8635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010133389351615913,
+      "loss": 1.1724,
+      "step": 8640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010124377039206829,
+      "loss": 1.1178,
+      "step": 8645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010115364625759724,
+      "loss": 1.1241,
+      "step": 8650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010106352118595853,
+      "loss": 1.1038,
+      "step": 8655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010097339525036557,
+      "loss": 1.1397,
+      "step": 8660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010088326852403228,
+      "loss": 1.1775,
+      "step": 8665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010079314108017346,
+      "loss": 1.0928,
+      "step": 8670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010070301299200436,
+      "loss": 1.2399,
+      "step": 8675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010061288433274078,
+      "loss": 1.1488,
+      "step": 8680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010052275517559896,
+      "loss": 1.0828,
+      "step": 8685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010043262559379559,
+      "loss": 1.1745,
+      "step": 8690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001003424956605476,
+      "loss": 1.1657,
+      "step": 8695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001002523654490724,
+      "loss": 1.1242,
+      "step": 8700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010016223503258743,
+      "loss": 1.1512,
+      "step": 8705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010007210448431039,
+      "loss": 1.0913,
+      "step": 8710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.998197387745908e-05,
+      "loss": 1.1523,
+      "step": 8715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.989184328525132e-05,
+      "loss": 1.1509,
+      "step": 8720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.9801712780905e-05,
+      "loss": 1.11,
+      "step": 8725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.97115824376378e-05,
+      "loss": 1.0863,
+      "step": 8730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.962145232866739e-05,
+      "loss": 1.1648,
+      "step": 8735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.953132252721119e-05,
+      "loss": 1.1273,
+      "step": 8740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.944119310648642e-05,
+      "loss": 1.1452,
+      "step": 8745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.93510641397099e-05,
+      "loss": 1.158,
+      "step": 8750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.92609357000982e-05,
+      "loss": 1.1209,
+      "step": 8755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.917080786086734e-05,
+      "loss": 1.1773,
+      "step": 8760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.908068069523296e-05,
+      "loss": 1.153,
+      "step": 8765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.899055427641009e-05,
+      "loss": 1.0285,
+      "step": 8770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.890042867761317e-05,
+      "loss": 1.0746,
+      "step": 8775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.881030397205597e-05,
+      "loss": 1.0845,
+      "step": 8780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.872018023295152e-05,
+      "loss": 1.0393,
+      "step": 8785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.863005753351213e-05,
+      "loss": 1.004,
+      "step": 8790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.853993594694917e-05,
+      "loss": 1.1829,
+      "step": 8795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.84498155464732e-05,
+      "loss": 1.1199,
+      "step": 8800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.83596964052937e-05,
+      "loss": 1.0783,
+      "step": 8805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.826957859661926e-05,
+      "loss": 1.1083,
+      "step": 8810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.817946219365726e-05,
+      "loss": 1.086,
+      "step": 8815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.80893472696141e-05,
+      "loss": 1.057,
+      "step": 8820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.799923389769476e-05,
+      "loss": 1.0881,
+      "step": 8825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.790912215110317e-05,
+      "loss": 1.1337,
+      "step": 8830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.78190121030418e-05,
+      "loss": 1.1191,
+      "step": 8835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.772890382671183e-05,
+      "loss": 1.1182,
+      "step": 8840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.763879739531298e-05,
+      "loss": 1.1848,
+      "step": 8845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.754869288204335e-05,
+      "loss": 1.1397,
+      "step": 8850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.74585903600997e-05,
+      "loss": 1.1849,
+      "step": 8855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.736848990267696e-05,
+      "loss": 1.1042,
+      "step": 8860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.727839158296855e-05,
+      "loss": 1.1374,
+      "step": 8865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.718829547416604e-05,
+      "loss": 1.0807,
+      "step": 8870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.70982016494593e-05,
+      "loss": 1.1148,
+      "step": 8875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.700811018203617e-05,
+      "loss": 1.157,
+      "step": 8880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.691802114508287e-05,
+      "loss": 1.0911,
+      "step": 8885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.682793461178332e-05,
+      "loss": 1.095,
+      "step": 8890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.673785065531967e-05,
+      "loss": 1.0803,
+      "step": 8895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.664776934887174e-05,
+      "loss": 1.1453,
+      "step": 8900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.655769076561741e-05,
+      "loss": 1.1647,
+      "step": 8905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.646761497873227e-05,
+      "loss": 1.1342,
+      "step": 8910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.637754206138958e-05,
+      "loss": 1.1248,
+      "step": 8915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.628747208676034e-05,
+      "loss": 1.0698,
+      "step": 8920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.619740512801313e-05,
+      "loss": 1.0617,
+      "step": 8925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.610734125831408e-05,
+      "loss": 1.0975,
+      "step": 8930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.601728055082683e-05,
+      "loss": 1.0645,
+      "step": 8935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.592722307871245e-05,
+      "loss": 1.0875,
+      "step": 8940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.58371689151293e-05,
+      "loss": 1.1371,
+      "step": 8945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.574711813323325e-05,
+      "loss": 1.0814,
+      "step": 8950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.565707080617716e-05,
+      "loss": 1.1898,
+      "step": 8955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.55670270071113e-05,
+      "loss": 1.0478,
+      "step": 8960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.547698680918297e-05,
+      "loss": 1.0517,
+      "step": 8965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.538695028553656e-05,
+      "loss": 1.1166,
+      "step": 8970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.529691750931351e-05,
+      "loss": 1.136,
+      "step": 8975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.520688855365215e-05,
+      "loss": 1.1232,
+      "step": 8980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.511686349168777e-05,
+      "loss": 1.1879,
+      "step": 8985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.502684239655242e-05,
+      "loss": 1.1094,
+      "step": 8990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.493682534137506e-05,
+      "loss": 1.075,
+      "step": 8995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.484681239928122e-05,
+      "loss": 1.1481,
+      "step": 9000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.475680364339322e-05,
+      "loss": 1.0817,
+      "step": 9005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.466679914682984e-05,
+      "loss": 1.1945,
+      "step": 9010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.457679898270654e-05,
+      "loss": 1.0667,
+      "step": 9015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.448680322413513e-05,
+      "loss": 1.0896,
+      "step": 9020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.439681194422398e-05,
+      "loss": 1.0759,
+      "step": 9025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.430682521607768e-05,
+      "loss": 1.104,
+      "step": 9030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.421684311279724e-05,
+      "loss": 1.1019,
+      "step": 9035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.412686570747982e-05,
+      "loss": 1.1248,
+      "step": 9040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.403689307321886e-05,
+      "loss": 1.1526,
+      "step": 9045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.394692528310385e-05,
+      "loss": 1.1475,
+      "step": 9050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.385696241022034e-05,
+      "loss": 1.0939,
+      "step": 9055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.376700452764995e-05,
+      "loss": 1.1197,
+      "step": 9060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.367705170847017e-05,
+      "loss": 1.0349,
+      "step": 9065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.358710402575445e-05,
+      "loss": 1.1744,
+      "step": 9070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.349716155257198e-05,
+      "loss": 1.1197,
+      "step": 9075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.340722436198785e-05,
+      "loss": 1.137,
+      "step": 9080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.331729252706271e-05,
+      "loss": 1.1218,
+      "step": 9085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.322736612085298e-05,
+      "loss": 1.1465,
+      "step": 9090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.313744521641057e-05,
+      "loss": 1.1061,
+      "step": 9095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.3047529886783e-05,
+      "loss": 1.1017,
+      "step": 9100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.295762020501321e-05,
+      "loss": 1.129,
+      "step": 9105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.286771624413959e-05,
+      "loss": 1.1193,
+      "step": 9110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.277781807719585e-05,
+      "loss": 1.0572,
+      "step": 9115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.268792577721101e-05,
+      "loss": 1.1119,
+      "step": 9120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.259803941720935e-05,
+      "loss": 1.1245,
+      "step": 9125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.250815907021021e-05,
+      "loss": 1.1375,
+      "step": 9130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.241828480922824e-05,
+      "loss": 1.1915,
+      "step": 9135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.232841670727296e-05,
+      "loss": 1.2246,
+      "step": 9140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.223855483734902e-05,
+      "loss": 1.0714,
+      "step": 9145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.21486992724559e-05,
+      "loss": 1.1096,
+      "step": 9150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.205885008558804e-05,
+      "loss": 1.1561,
+      "step": 9155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.196900734973467e-05,
+      "loss": 1.1509,
+      "step": 9160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.187917113787978e-05,
+      "loss": 1.1323,
+      "step": 9165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.1789341523002e-05,
+      "loss": 1.2388,
+      "step": 9170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.169951857807473e-05,
+      "loss": 1.0981,
+      "step": 9175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.160970237606589e-05,
+      "loss": 1.1694,
+      "step": 9180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.151989298993788e-05,
+      "loss": 1.1463,
+      "step": 9185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.143009049264758e-05,
+      "loss": 1.1138,
+      "step": 9190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.13402949571463e-05,
+      "loss": 1.1081,
+      "step": 9195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.125050645637973e-05,
+      "loss": 1.1866,
+      "step": 9200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.116072506328772e-05,
+      "loss": 1.1086,
+      "step": 9205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.107095085080447e-05,
+      "loss": 1.1476,
+      "step": 9210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.098118389185831e-05,
+      "loss": 1.165,
+      "step": 9215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.08914242593717e-05,
+      "loss": 1.0506,
+      "step": 9220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.080167202626106e-05,
+      "loss": 1.1633,
+      "step": 9225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.071192726543688e-05,
+      "loss": 1.1322,
+      "step": 9230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.062219004980355e-05,
+      "loss": 1.0703,
+      "step": 9235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.053246045225935e-05,
+      "loss": 1.1341,
+      "step": 9240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.044273854569635e-05,
+      "loss": 1.1245,
+      "step": 9245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.035302440300037e-05,
+      "loss": 1.0769,
+      "step": 9250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.026331809705098e-05,
+      "loss": 1.1332,
+      "step": 9255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.017361970072127e-05,
+      "loss": 1.0415,
+      "step": 9260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.008392928687806e-05,
+      "loss": 1.0913,
+      "step": 9265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.999424692838153e-05,
+      "loss": 1.219,
+      "step": 9270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.99045726980854e-05,
+      "loss": 1.0861,
+      "step": 9275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.981490666883679e-05,
+      "loss": 1.2252,
+      "step": 9280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.972524891347613e-05,
+      "loss": 1.0959,
+      "step": 9285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.963559950483714e-05,
+      "loss": 1.1034,
+      "step": 9290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.954595851574678e-05,
+      "loss": 1.098,
+      "step": 9295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.94563260190251e-05,
+      "loss": 1.1551,
+      "step": 9300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.936670208748532e-05,
+      "loss": 1.1954,
+      "step": 9305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.927708679393374e-05,
+      "loss": 1.1415,
+      "step": 9310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.918748021116947e-05,
+      "loss": 1.0772,
+      "step": 9315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.909788241198477e-05,
+      "loss": 1.1076,
+      "step": 9320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.900829346916456e-05,
+      "loss": 1.0871,
+      "step": 9325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.891871345548674e-05,
+      "loss": 1.1126,
+      "step": 9330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.88291424437218e-05,
+      "loss": 1.0862,
+      "step": 9335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.873958050663306e-05,
+      "loss": 1.1689,
+      "step": 9340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.865002771697636e-05,
+      "loss": 1.1083,
+      "step": 9345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.856048414750016e-05,
+      "loss": 1.1504,
+      "step": 9350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.847094987094539e-05,
+      "loss": 1.0532,
+      "step": 9355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.838142496004555e-05,
+      "loss": 1.0392,
+      "step": 9360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.829190948752635e-05,
+      "loss": 1.1317,
+      "step": 9365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.820240352610594e-05,
+      "loss": 1.1114,
+      "step": 9370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.811290714849475e-05,
+      "loss": 1.1714,
+      "step": 9375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.802342042739538e-05,
+      "loss": 1.1133,
+      "step": 9380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.793394343550268e-05,
+      "loss": 1.1332,
+      "step": 9385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.784447624550344e-05,
+      "loss": 1.1295,
+      "step": 9390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.775501893007663e-05,
+      "loss": 1.1029,
+      "step": 9395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.766557156189304e-05,
+      "loss": 1.1589,
+      "step": 9400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.75761342136156e-05,
+      "loss": 1.1283,
+      "step": 9405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.748670695789893e-05,
+      "loss": 1.0744,
+      "step": 9410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.739728986738951e-05,
+      "loss": 1.1723,
+      "step": 9415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.730788301472553e-05,
+      "loss": 1.171,
+      "step": 9420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.72184864725369e-05,
+      "loss": 1.1209,
+      "step": 9425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.712910031344516e-05,
+      "loss": 1.0965,
+      "step": 9430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.70397246100634e-05,
+      "loss": 1.1513,
+      "step": 9435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.695035943499618e-05,
+      "loss": 1.1732,
+      "step": 9440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.686100486083955e-05,
+      "loss": 1.1174,
+      "step": 9445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.677166096018097e-05,
+      "loss": 1.1186,
+      "step": 9450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.668232780559915e-05,
+      "loss": 1.1368,
+      "step": 9455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.659300546966415e-05,
+      "loss": 1.076,
+      "step": 9460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.650369402493722e-05,
+      "loss": 1.131,
+      "step": 9465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.641439354397073e-05,
+      "loss": 1.1503,
+      "step": 9470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.632510409930815e-05,
+      "loss": 1.1177,
+      "step": 9475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.623582576348406e-05,
+      "loss": 1.0798,
+      "step": 9480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.614655860902388e-05,
+      "loss": 1.1277,
+      "step": 9485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.605730270844408e-05,
+      "loss": 1.1412,
+      "step": 9490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.596805813425189e-05,
+      "loss": 1.1169,
+      "step": 9495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.587882495894543e-05,
+      "loss": 1.1518,
+      "step": 9500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.578960325501345e-05,
+      "loss": 1.1035,
+      "step": 9505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.570039309493546e-05,
+      "loss": 1.1088,
+      "step": 9510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.561119455118163e-05,
+      "loss": 1.0848,
+      "step": 9515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.552200769621255e-05,
+      "loss": 1.1476,
+      "step": 9520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.543283260247948e-05,
+      "loss": 1.0809,
+      "step": 9525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.534366934242396e-05,
+      "loss": 1.1005,
+      "step": 9530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.525451798847809e-05,
+      "loss": 1.0762,
+      "step": 9535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.516537861306419e-05,
+      "loss": 1.1484,
+      "step": 9540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.507625128859484e-05,
+      "loss": 1.0923,
+      "step": 9545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.498713608747285e-05,
+      "loss": 1.1191,
+      "step": 9550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.489803308209125e-05,
+      "loss": 1.1145,
+      "step": 9555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.480894234483307e-05,
+      "loss": 1.1304,
+      "step": 9560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.471986394807142e-05,
+      "loss": 1.0836,
+      "step": 9565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.463079796416937e-05,
+      "loss": 1.1304,
+      "step": 9570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.454174446547995e-05,
+      "loss": 1.1259,
+      "step": 9575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.445270352434593e-05,
+      "loss": 1.1105,
+      "step": 9580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.43636752131e-05,
+      "loss": 1.1828,
+      "step": 9585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.427465960406458e-05,
+      "loss": 1.1204,
+      "step": 9590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.418565676955167e-05,
+      "loss": 1.1321,
+      "step": 9595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.409666678186303e-05,
+      "loss": 1.1252,
+      "step": 9600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.400768971328988e-05,
+      "loss": 1.1206,
+      "step": 9605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.391872563611298e-05,
+      "loss": 1.0638,
+      "step": 9610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.382977462260257e-05,
+      "loss": 1.0795,
+      "step": 9615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.37408367450182e-05,
+      "loss": 1.203,
+      "step": 9620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.365191207560878e-05,
+      "loss": 1.0837,
+      "step": 9625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.356300068661259e-05,
+      "loss": 1.1268,
+      "step": 9630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.347410265025695e-05,
+      "loss": 1.1756,
+      "step": 9635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.338521803875848e-05,
+      "loss": 1.0606,
+      "step": 9640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.329634692432279e-05,
+      "loss": 1.0965,
+      "step": 9645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.320748937914458e-05,
+      "loss": 1.1075,
+      "step": 9650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.311864547540753e-05,
+      "loss": 1.0918,
+      "step": 9655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.302981528528417e-05,
+      "loss": 1.1592,
+      "step": 9660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.294099888093603e-05,
+      "loss": 1.1859,
+      "step": 9665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.285219633451333e-05,
+      "loss": 1.1572,
+      "step": 9670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.276340771815502e-05,
+      "loss": 1.1321,
+      "step": 9675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.267463310398876e-05,
+      "loss": 1.0831,
+      "step": 9680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.25858725641309e-05,
+      "loss": 1.147,
+      "step": 9685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.249712617068629e-05,
+      "loss": 1.1761,
+      "step": 9690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.240839399574831e-05,
+      "loss": 1.1747,
+      "step": 9695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.231967611139874e-05,
+      "loss": 1.1002,
+      "step": 9700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.223097258970783e-05,
+      "loss": 1.0486,
+      "step": 9705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.21422835027341e-05,
+      "loss": 1.142,
+      "step": 9710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.205360892252438e-05,
+      "loss": 1.0942,
+      "step": 9715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.196494892111369e-05,
+      "loss": 1.1185,
+      "step": 9720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.187630357052522e-05,
+      "loss": 1.0845,
+      "step": 9725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.178767294277031e-05,
+      "loss": 1.1787,
+      "step": 9730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.169905710984823e-05,
+      "loss": 1.1407,
+      "step": 9735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.161045614374632e-05,
+      "loss": 1.0873,
+      "step": 9740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.152187011643978e-05,
+      "loss": 1.0734,
+      "step": 9745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.143329909989175e-05,
+      "loss": 1.1591,
+      "step": 9750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.134474316605311e-05,
+      "loss": 1.0711,
+      "step": 9755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.125620238686253e-05,
+      "loss": 1.0927,
+      "step": 9760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.11676768342463e-05,
+      "loss": 1.1376,
+      "step": 9765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.107916658011847e-05,
+      "loss": 1.1452,
+      "step": 9770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.09906716963805e-05,
+      "loss": 1.1001,
+      "step": 9775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.090219225492153e-05,
+      "loss": 1.0564,
+      "step": 9780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.081372832761798e-05,
+      "loss": 1.0939,
+      "step": 9785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.072527998633381e-05,
+      "loss": 1.1263,
+      "step": 9790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.063684730292028e-05,
+      "loss": 1.0566,
+      "step": 9795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.054843034921589e-05,
+      "loss": 1.1386,
+      "step": 9800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.04600291970464e-05,
+      "loss": 1.106,
+      "step": 9805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.037164391822472e-05,
+      "loss": 1.1235,
+      "step": 9810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.02832745845509e-05,
+      "loss": 1.0541,
+      "step": 9815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.019492126781195e-05,
+      "loss": 1.1787,
+      "step": 9820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.010658403978197e-05,
+      "loss": 1.0591,
+      "step": 9825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.001826297222194e-05,
+      "loss": 1.1217,
+      "step": 9830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.992995813687969e-05,
+      "loss": 1.0979,
+      "step": 9835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.984166960548991e-05,
+      "loss": 1.1028,
+      "step": 9840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.975339744977401e-05,
+      "loss": 1.1215,
+      "step": 9845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.966514174144015e-05,
+      "loss": 1.1067,
+      "step": 9850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.957690255218304e-05,
+      "loss": 1.0957,
+      "step": 9855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.948867995368408e-05,
+      "loss": 1.1645,
+      "step": 9860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.940047401761107e-05,
+      "loss": 1.154,
+      "step": 9865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.931228481561841e-05,
+      "loss": 1.1907,
+      "step": 9870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.922411241934678e-05,
+      "loss": 1.1839,
+      "step": 9875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.91359569004233e-05,
+      "loss": 1.1389,
+      "step": 9880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.904781833046129e-05,
+      "loss": 1.124,
+      "step": 9885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.895969678106041e-05,
+      "loss": 1.1307,
+      "step": 9890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.887159232380639e-05,
+      "loss": 0.9746,
+      "step": 9895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.878350503027118e-05,
+      "loss": 1.1074,
+      "step": 9900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.869543497201268e-05,
+      "loss": 1.1333,
+      "step": 9905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.860738222057488e-05,
+      "loss": 1.0711,
+      "step": 9910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.851934684748765e-05,
+      "loss": 1.1405,
+      "step": 9915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.843132892426674e-05,
+      "loss": 1.0734,
+      "step": 9920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.83433285224138e-05,
+      "loss": 1.1478,
+      "step": 9925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.825534571341613e-05,
+      "loss": 1.0633,
+      "step": 9930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.816738056874686e-05,
+      "loss": 1.1151,
+      "step": 9935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.807943315986465e-05,
+      "loss": 1.08,
+      "step": 9940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.799150355821388e-05,
+      "loss": 1.0822,
+      "step": 9945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.790359183522435e-05,
+      "loss": 1.1019,
+      "step": 9950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.781569806231142e-05,
+      "loss": 1.1109,
+      "step": 9955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.772782231087577e-05,
+      "loss": 1.0887,
+      "step": 9960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.763996465230357e-05,
+      "loss": 1.0661,
+      "step": 9965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.755212515796615e-05,
+      "loss": 1.0919,
+      "step": 9970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.746430389922026e-05,
+      "loss": 1.1297,
+      "step": 9975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.737650094740767e-05,
+      "loss": 1.1043,
+      "step": 9980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.728871637385537e-05,
+      "loss": 1.175,
+      "step": 9985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.720095024987539e-05,
+      "loss": 1.0476,
+      "step": 9990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.711320264676476e-05,
+      "loss": 1.148,
+      "step": 9995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.70254736358055e-05,
+      "loss": 1.1077,
+      "step": 10000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.693776328826449e-05,
+      "loss": 1.0244,
+      "step": 10005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.685007167539348e-05,
+      "loss": 1.1144,
+      "step": 10010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.676239886842898e-05,
+      "loss": 1.065,
+      "step": 10015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.667474493859225e-05,
+      "loss": 1.131,
+      "step": 10020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.658710995708917e-05,
+      "loss": 1.0779,
+      "step": 10025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.649949399511027e-05,
+      "loss": 1.074,
+      "step": 10030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.641189712383058e-05,
+      "loss": 1.1598,
+      "step": 10035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.632431941440971e-05,
+      "loss": 1.1384,
+      "step": 10040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.623676093799159e-05,
+      "loss": 1.0922,
+      "step": 10045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.614922176570463e-05,
+      "loss": 1.0746,
+      "step": 10050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.606170196866148e-05,
+      "loss": 1.071,
+      "step": 10055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.597420161795909e-05,
+      "loss": 1.1491,
+      "step": 10060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.588672078467862e-05,
+      "loss": 1.1222,
+      "step": 10065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.579925953988534e-05,
+      "loss": 1.163,
+      "step": 10070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.571181795462866e-05,
+      "loss": 1.1735,
+      "step": 10075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.562439609994192e-05,
+      "loss": 1.1638,
+      "step": 10080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.553699404684255e-05,
+      "loss": 1.1776,
+      "step": 10085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.544961186633184e-05,
+      "loss": 1.1405,
+      "step": 10090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.536224962939497e-05,
+      "loss": 1.0934,
+      "step": 10095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.527490740700084e-05,
+      "loss": 1.0013,
+      "step": 10100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.518758527010217e-05,
+      "loss": 1.1097,
+      "step": 10105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.51002832896353e-05,
+      "loss": 1.0495,
+      "step": 10110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.501300153652027e-05,
+      "loss": 1.0718,
+      "step": 10115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.492574008166058e-05,
+      "loss": 1.1874,
+      "step": 10120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.483849899594334e-05,
+      "loss": 1.089,
+      "step": 10125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.47512783502391e-05,
+      "loss": 1.0803,
+      "step": 10130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.466407821540177e-05,
+      "loss": 1.111,
+      "step": 10135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.457689866226864e-05,
+      "loss": 1.1375,
+      "step": 10140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.448973976166018e-05,
+      "loss": 1.1459,
+      "step": 10145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.440260158438028e-05,
+      "loss": 1.0854,
+      "step": 10150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.431548420121575e-05,
+      "loss": 1.0891,
+      "step": 10155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.42283876829367e-05,
+      "loss": 1.0713,
+      "step": 10160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.414131210029622e-05,
+      "loss": 1.1664,
+      "step": 10165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.405425752403037e-05,
+      "loss": 1.0998,
+      "step": 10170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.396722402485817e-05,
+      "loss": 1.1358,
+      "step": 10175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.388021167348152e-05,
+      "loss": 1.1019,
+      "step": 10180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.379322054058515e-05,
+      "loss": 1.0098,
+      "step": 10185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.370625069683649e-05,
+      "loss": 1.1124,
+      "step": 10190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.361930221288581e-05,
+      "loss": 1.0658,
+      "step": 10195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.353237515936588e-05,
+      "loss": 1.0451,
+      "step": 10200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.344546960689215e-05,
+      "loss": 1.1221,
+      "step": 10205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.335858562606259e-05,
+      "loss": 1.103,
+      "step": 10210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.327172328745762e-05,
+      "loss": 1.0827,
+      "step": 10215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.318488266164012e-05,
+      "loss": 1.0917,
+      "step": 10220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.309806381915527e-05,
+      "loss": 1.1364,
+      "step": 10225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.301126683053058e-05,
+      "loss": 1.1422,
+      "step": 10230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.292449176627586e-05,
+      "loss": 1.1408,
+      "step": 10235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.283773869688303e-05,
+      "loss": 1.1349,
+      "step": 10240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.275100769282623e-05,
+      "loss": 1.0405,
+      "step": 10245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.266429882456157e-05,
+      "loss": 1.1005,
+      "step": 10250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.257761216252723e-05,
+      "loss": 1.1948,
+      "step": 10255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.24909477771434e-05,
+      "loss": 1.0878,
+      "step": 10260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.240430573881205e-05,
+      "loss": 1.1491,
+      "step": 10265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.23176861179172e-05,
+      "loss": 1.1067,
+      "step": 10270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.223108898482441e-05,
+      "loss": 1.112,
+      "step": 10275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.214451440988117e-05,
+      "loss": 1.1295,
+      "step": 10280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.205796246341652e-05,
+      "loss": 1.102,
+      "step": 10285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.197143321574119e-05,
+      "loss": 1.0947,
+      "step": 10290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.188492673714744e-05,
+      "loss": 1.1683,
+      "step": 10295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.179844309790904e-05,
+      "loss": 1.1363,
+      "step": 10300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.171198236828122e-05,
+      "loss": 1.0505,
+      "step": 10305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.16255446185006e-05,
+      "loss": 1.0529,
+      "step": 10310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.153912991878506e-05,
+      "loss": 1.0862,
+      "step": 10315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.145273833933388e-05,
+      "loss": 1.1673,
+      "step": 10320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.136636995032742e-05,
+      "loss": 1.1195,
+      "step": 10325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.12800248219273e-05,
+      "loss": 1.0825,
+      "step": 10330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.11937030242763e-05,
+      "loss": 1.1358,
+      "step": 10335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.110740462749805e-05,
+      "loss": 1.1831,
+      "step": 10340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.102112970169738e-05,
+      "loss": 1.0786,
+      "step": 10345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.093487831695985e-05,
+      "loss": 1.1177,
+      "step": 10350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.08486505433521e-05,
+      "loss": 1.2008,
+      "step": 10355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.076244645092141e-05,
+      "loss": 1.1226,
+      "step": 10360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.067626610969602e-05,
+      "loss": 1.1201,
+      "step": 10365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.059010958968462e-05,
+      "loss": 1.1061,
+      "step": 10370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.05039769608768e-05,
+      "loss": 1.1028,
+      "step": 10375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.041786829324258e-05,
+      "loss": 1.1542,
+      "step": 10380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.033178365673258e-05,
+      "loss": 1.1056,
+      "step": 10385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.02457231212779e-05,
+      "loss": 1.1801,
+      "step": 10390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.015968675679003e-05,
+      "loss": 1.2161,
+      "step": 10395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.007367463316088e-05,
+      "loss": 1.1018,
+      "step": 10400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.998768682026256e-05,
+      "loss": 1.1212,
+      "step": 10405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.990172338794759e-05,
+      "loss": 0.9987,
+      "step": 10410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.981578440604852e-05,
+      "loss": 1.0532,
+      "step": 10415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.972986994437815e-05,
+      "loss": 1.1031,
+      "step": 10420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.964398007272927e-05,
+      "loss": 1.1004,
+      "step": 10425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.955811486087484e-05,
+      "loss": 1.0524,
+      "step": 10430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.947227437856764e-05,
+      "loss": 1.0695,
+      "step": 10435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.938645869554042e-05,
+      "loss": 1.0736,
+      "step": 10440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.930066788150577e-05,
+      "loss": 1.1102,
+      "step": 10445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.92149020061561e-05,
+      "loss": 1.1338,
+      "step": 10450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.912916113916357e-05,
+      "loss": 1.0925,
+      "step": 10455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.904344535017995e-05,
+      "loss": 1.1167,
+      "step": 10460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.895775470883676e-05,
+      "loss": 1.0626,
+      "step": 10465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.887208928474502e-05,
+      "loss": 1.1136,
+      "step": 10470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.878644914749521e-05,
+      "loss": 1.0736,
+      "step": 10475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.870083436665737e-05,
+      "loss": 1.0951,
+      "step": 10480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.861524501178084e-05,
+      "loss": 1.0115,
+      "step": 10485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.852968115239444e-05,
+      "loss": 1.1083,
+      "step": 10490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.844414285800614e-05,
+      "loss": 1.0884,
+      "step": 10495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.835863019810325e-05,
+      "loss": 1.19,
+      "step": 10500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.827314324215216e-05,
+      "loss": 1.1527,
+      "step": 10505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.818768205959847e-05,
+      "loss": 1.0989,
+      "step": 10510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.810224671986675e-05,
+      "loss": 1.1104,
+      "step": 10515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.801683729236071e-05,
+      "loss": 1.1296,
+      "step": 10520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.793145384646284e-05,
+      "loss": 1.0106,
+      "step": 10525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.784609645153472e-05,
+      "loss": 1.0996,
+      "step": 10530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.776076517691654e-05,
+      "loss": 1.0468,
+      "step": 10535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.76754600919275e-05,
+      "loss": 1.0788,
+      "step": 10540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.759018126586531e-05,
+      "loss": 1.0402,
+      "step": 10545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.750492876800655e-05,
+      "loss": 1.127,
+      "step": 10550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.74197026676062e-05,
+      "loss": 1.1174,
+      "step": 10555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.733450303389805e-05,
+      "loss": 1.0851,
+      "step": 10560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.724932993609413e-05,
+      "loss": 1.1548,
+      "step": 10565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.716418344338511e-05,
+      "loss": 1.0381,
+      "step": 10570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.707906362493998e-05,
+      "loss": 1.1459,
+      "step": 10575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.699397054990599e-05,
+      "loss": 1.1529,
+      "step": 10580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.690890428740877e-05,
+      "loss": 1.1147,
+      "step": 10585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.682386490655209e-05,
+      "loss": 1.1362,
+      "step": 10590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.673885247641799e-05,
+      "loss": 1.1775,
+      "step": 10595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.665386706606646e-05,
+      "loss": 1.1248,
+      "step": 10600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.656890874453575e-05,
+      "loss": 1.199,
+      "step": 10605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.648397758084183e-05,
+      "loss": 1.0962,
+      "step": 10610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.63990736439789e-05,
+      "loss": 1.1644,
+      "step": 10615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.631419700291875e-05,
+      "loss": 1.179,
+      "step": 10620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.62293477266113e-05,
+      "loss": 1.0606,
+      "step": 10625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.614452588398394e-05,
+      "loss": 1.119,
+      "step": 10630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.605973154394207e-05,
+      "loss": 1.1478,
+      "step": 10635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.597496477536848e-05,
+      "loss": 1.1199,
+      "step": 10640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.58902256471238e-05,
+      "loss": 1.1428,
+      "step": 10645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.580551422804593e-05,
+      "loss": 1.1196,
+      "step": 10650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.572083058695056e-05,
+      "loss": 1.0792,
+      "step": 10655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.563617479263061e-05,
+      "loss": 1.0675,
+      "step": 10660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.555154691385648e-05,
+      "loss": 1.0904,
+      "step": 10665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.54669470193758e-05,
+      "loss": 1.071,
+      "step": 10670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.538237517791359e-05,
+      "loss": 1.0972,
+      "step": 10675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.529783145817193e-05,
+      "loss": 1.1276,
+      "step": 10680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.521331592883018e-05,
+      "loss": 1.1124,
+      "step": 10685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.51288286585448e-05,
+      "loss": 1.0746,
+      "step": 10690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.504436971594916e-05,
+      "loss": 1.1259,
+      "step": 10695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.495993916965383e-05,
+      "loss": 1.0786,
+      "step": 10700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.487553708824606e-05,
+      "loss": 1.1774,
+      "step": 10705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.479116354029023e-05,
+      "loss": 1.126,
+      "step": 10710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.47068185943273e-05,
+      "loss": 1.1933,
+      "step": 10715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.462250231887521e-05,
+      "loss": 1.1841,
+      "step": 10720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.453821478242847e-05,
+      "loss": 1.1088,
+      "step": 10725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.44539560534583e-05,
+      "loss": 1.1384,
+      "step": 10730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.436972620041249e-05,
+      "loss": 1.0844,
+      "step": 10735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.42855252917154e-05,
+      "loss": 1.0835,
+      "step": 10740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.420135339576784e-05,
+      "loss": 1.1662,
+      "step": 10745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.411721058094713e-05,
+      "loss": 1.0605,
+      "step": 10750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.403309691560681e-05,
+      "loss": 1.1286,
+      "step": 10755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.394901246807695e-05,
+      "loss": 1.1304,
+      "step": 10760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.386495730666375e-05,
+      "loss": 1.1059,
+      "step": 10765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.378093149964961e-05,
+      "loss": 1.0618,
+      "step": 10770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.369693511529314e-05,
+      "loss": 1.1487,
+      "step": 10775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.361296822182906e-05,
+      "loss": 1.1809,
+      "step": 10780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.352903088746802e-05,
+      "loss": 1.0398,
+      "step": 10785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.34451231803968e-05,
+      "loss": 1.128,
+      "step": 10790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.33612451687781e-05,
+      "loss": 1.1885,
+      "step": 10795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.327739692075033e-05,
+      "loss": 1.1196,
+      "step": 10800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.319357850442793e-05,
+      "loss": 1.0162,
+      "step": 10805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.310978998790096e-05,
+      "loss": 1.1401,
+      "step": 10810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.30260314392353e-05,
+      "loss": 1.0988,
+      "step": 10815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.29423029264724e-05,
+      "loss": 1.1285,
+      "step": 10820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.285860451762935e-05,
+      "loss": 1.1223,
+      "step": 10825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.277493628069878e-05,
+      "loss": 1.1253,
+      "step": 10830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.269129828364881e-05,
+      "loss": 1.1792,
+      "step": 10835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.260769059442296e-05,
+      "loss": 1.0359,
+      "step": 10840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.252411328094021e-05,
+      "loss": 1.0991,
+      "step": 10845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.244056641109478e-05,
+      "loss": 1.0626,
+      "step": 10850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.235705005275622e-05,
+      "loss": 1.1192,
+      "step": 10855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.227356427376921e-05,
+      "loss": 1.0149,
+      "step": 10860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.219010914195376e-05,
+      "loss": 1.1014,
+      "step": 10865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.210668472510472e-05,
+      "loss": 1.1001,
+      "step": 10870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.20232910909923e-05,
+      "loss": 1.1319,
+      "step": 10875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.193992830736139e-05,
+      "loss": 1.0524,
+      "step": 10880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.185659644193211e-05,
+      "loss": 1.0679,
+      "step": 10885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.177329556239921e-05,
+      "loss": 1.1015,
+      "step": 10890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.169002573643245e-05,
+      "loss": 1.1231,
+      "step": 10895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.160678703167629e-05,
+      "loss": 1.0801,
+      "step": 10900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.15235795157499e-05,
+      "loss": 1.1222,
+      "step": 10905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.144040325624716e-05,
+      "loss": 1.0823,
+      "step": 10910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.135725832073645e-05,
+      "loss": 1.0739,
+      "step": 10915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.12741447767609e-05,
+      "loss": 1.1289,
+      "step": 10920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.119106269183793e-05,
+      "loss": 1.0874,
+      "step": 10925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.110801213345952e-05,
+      "loss": 1.1188,
+      "step": 10930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.102499316909199e-05,
+      "loss": 1.1673,
+      "step": 10935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.094200586617609e-05,
+      "loss": 1.1936,
+      "step": 10940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.085905029212668e-05,
+      "loss": 1.1333,
+      "step": 10945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.077612651433306e-05,
+      "loss": 1.1491,
+      "step": 10950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0693234600158456e-05,
+      "loss": 1.1421,
+      "step": 10955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.061037461694047e-05,
+      "loss": 1.069,
+      "step": 10960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.05275466319905e-05,
+      "loss": 1.1936,
+      "step": 10965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0444750712594206e-05,
+      "loss": 1.1178,
+      "step": 10970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.036198692601096e-05,
+      "loss": 1.0836,
+      "step": 10975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.027925533947426e-05,
+      "loss": 1.1179,
+      "step": 10980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.019655602019122e-05,
+      "loss": 1.1624,
+      "step": 10985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0113889035342935e-05,
+      "loss": 1.1446,
+      "step": 10990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0031254452084106e-05,
+      "loss": 1.1677,
+      "step": 10995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.994865233754321e-05,
+      "loss": 1.1706,
+      "step": 11000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.986608275882224e-05,
+      "loss": 1.1413,
+      "step": 11005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.978354578299683e-05,
+      "loss": 1.1753,
+      "step": 11010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.970104147711616e-05,
+      "loss": 1.1574,
+      "step": 11015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.961856990820276e-05,
+      "loss": 1.0639,
+      "step": 11020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9536131143252725e-05,
+      "loss": 1.0847,
+      "step": 11025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.945372524923526e-05,
+      "loss": 1.1702,
+      "step": 11030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9371352293093194e-05,
+      "loss": 1.205,
+      "step": 11035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.928901234174222e-05,
+      "loss": 1.1253,
+      "step": 11040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.920670546207162e-05,
+      "loss": 1.0925,
+      "step": 11045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.912443172094344e-05,
+      "loss": 1.1342,
+      "step": 11050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.90421911851931e-05,
+      "loss": 1.1636,
+      "step": 11055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.8959983921628806e-05,
+      "loss": 1.1584,
+      "step": 11060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.887780999703197e-05,
+      "loss": 1.0906,
+      "step": 11065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.879566947815669e-05,
+      "loss": 1.1653,
+      "step": 11070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.871356243173015e-05,
+      "loss": 1.1211,
+      "step": 11075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.8631488924452114e-05,
+      "loss": 1.1169,
+      "step": 11080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.85494490229953e-05,
+      "loss": 1.0977,
+      "step": 11085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.846744279400499e-05,
+      "loss": 1.1312,
+      "step": 11090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.8385470304099204e-05,
+      "loss": 1.078,
+      "step": 11095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.8303531619868496e-05,
+      "loss": 1.0715,
+      "step": 11100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.8221626807876e-05,
+      "loss": 1.1397,
+      "step": 11105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.813975593465725e-05,
+      "loss": 1.1986,
+      "step": 11110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.8057919066720335e-05,
+      "loss": 1.115,
+      "step": 11115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.797611627054563e-05,
+      "loss": 1.0957,
+      "step": 11120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.789434761258581e-05,
+      "loss": 1.0986,
+      "step": 11125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.781261315926596e-05,
+      "loss": 1.1246,
+      "step": 11130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.773091297698318e-05,
+      "loss": 1.0196,
+      "step": 11135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.7649247132107e-05,
+      "loss": 1.1791,
+      "step": 11140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.756761569097866e-05,
+      "loss": 1.135,
+      "step": 11145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.748601871991183e-05,
+      "loss": 1.2191,
+      "step": 11150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.7404456285191963e-05,
+      "loss": 1.1379,
+      "step": 11155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.732292845307664e-05,
+      "loss": 1.1218,
+      "step": 11160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.724143528979505e-05,
+      "loss": 1.1387,
+      "step": 11165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.715997686154849e-05,
+      "loss": 1.0241,
+      "step": 11170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.7078553234509864e-05,
+      "loss": 1.1862,
+      "step": 11175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.699716447482399e-05,
+      "loss": 1.1326,
+      "step": 11180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.691581064860709e-05,
+      "loss": 1.1234,
+      "step": 11185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6834491821947295e-05,
+      "loss": 1.1737,
+      "step": 11190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.675320806090408e-05,
+      "loss": 1.0744,
+      "step": 11195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.66719594315086e-05,
+      "loss": 1.0945,
+      "step": 11200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6590745999763365e-05,
+      "loss": 1.0955,
+      "step": 11205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.650956783164233e-05,
+      "loss": 1.0479,
+      "step": 11210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6428424993090744e-05,
+      "loss": 1.1357,
+      "step": 11215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6347317550025305e-05,
+      "loss": 1.0897,
+      "step": 11220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6266245568333815e-05,
+      "loss": 1.0691,
+      "step": 11225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6185209113875275e-05,
+      "loss": 1.1319,
+      "step": 11230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.610420825247994e-05,
+      "loss": 1.1202,
+      "step": 11235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.602324304994901e-05,
+      "loss": 1.1487,
+      "step": 11240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.594231357205492e-05,
+      "loss": 1.1044,
+      "step": 11245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.586141988454076e-05,
+      "loss": 1.1353,
+      "step": 11250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.578056205312085e-05,
+      "loss": 1.1617,
+      "step": 11255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.5699740143480226e-05,
+      "loss": 1.1061,
+      "step": 11260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.561895422127481e-05,
+      "loss": 1.1968,
+      "step": 11265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.5538204352131275e-05,
+      "loss": 1.1268,
+      "step": 11270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.545749060164696e-05,
+      "loss": 1.1069,
+      "step": 11275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.5376813035389874e-05,
+      "loss": 1.2296,
+      "step": 11280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.5296171718898724e-05,
+      "loss": 1.0729,
+      "step": 11285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.521556671768267e-05,
+      "loss": 1.129,
+      "step": 11290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.513499809722139e-05,
+      "loss": 1.1066,
+      "step": 11295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.505446592296501e-05,
+      "loss": 1.0274,
+      "step": 11300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.4973970260334096e-05,
+      "loss": 1.1432,
+      "step": 11305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.48935111747195e-05,
+      "loss": 1.1238,
+      "step": 11310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.481308873148236e-05,
+      "loss": 1.1235,
+      "step": 11315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.473270299595405e-05,
+      "loss": 1.1306,
+      "step": 11320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.465235403343619e-05,
+      "loss": 1.1266,
+      "step": 11325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.4572041909200406e-05,
+      "loss": 1.1028,
+      "step": 11330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.449176668848857e-05,
+      "loss": 1.1316,
+      "step": 11335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.4411528436512414e-05,
+      "loss": 1.115,
+      "step": 11340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.4331327218453646e-05,
+      "loss": 1.144,
+      "step": 11345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.425116309946403e-05,
+      "loss": 1.0982,
+      "step": 11350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.417103614466507e-05,
+      "loss": 1.1867,
+      "step": 11355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.409094641914812e-05,
+      "loss": 1.1451,
+      "step": 11360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.4010893987974224e-05,
+      "loss": 1.1376,
+      "step": 11365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.3930878916174284e-05,
+      "loss": 1.1039,
+      "step": 11370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.385090126874873e-05,
+      "loss": 1.1259,
+      "step": 11375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.377096111066758e-05,
+      "loss": 1.0517,
+      "step": 11380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.369105850687044e-05,
+      "loss": 1.1109,
+      "step": 11385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.361119352226645e-05,
+      "loss": 1.0659,
+      "step": 11390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.353136622173414e-05,
+      "loss": 1.0613,
+      "step": 11395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.345157667012141e-05,
+      "loss": 1.0763,
+      "step": 11400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.337182493224547e-05,
+      "loss": 1.1335,
+      "step": 11405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.3292111072892946e-05,
+      "loss": 1.16,
+      "step": 11410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.321243515681953e-05,
+      "loss": 1.194,
+      "step": 11415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.313279724875029e-05,
+      "loss": 1.046,
+      "step": 11420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.3053197413379116e-05,
+      "loss": 1.0739,
+      "step": 11425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.2973635715369285e-05,
+      "loss": 1.1689,
+      "step": 11430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.289411221935287e-05,
+      "loss": 1.1096,
+      "step": 11435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.281462698993106e-05,
+      "loss": 1.1251,
+      "step": 11440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.2735180091673896e-05,
+      "loss": 1.1601,
+      "step": 11445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.265577158912023e-05,
+      "loss": 1.0862,
+      "step": 11450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.257640154677783e-05,
+      "loss": 1.1508,
+      "step": 11455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.249707002912312e-05,
+      "loss": 1.1167,
+      "step": 11460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.24177771006013e-05,
+      "loss": 1.108,
+      "step": 11465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.2338522825626135e-05,
+      "loss": 1.1568,
+      "step": 11470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.225930726858013e-05,
+      "loss": 1.131,
+      "step": 11475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.2180130493814185e-05,
+      "loss": 1.0945,
+      "step": 11480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.210099256564787e-05,
+      "loss": 1.0948,
+      "step": 11485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.202189354836895e-05,
+      "loss": 1.1538,
+      "step": 11490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.194283350623384e-05,
+      "loss": 1.1023,
+      "step": 11495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.186381250346709e-05,
+      "loss": 1.2009,
+      "step": 11500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.178483060426174e-05,
+      "loss": 1.1419,
+      "step": 11505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1705887872778805e-05,
+      "loss": 1.1852,
+      "step": 11510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.162698437314773e-05,
+      "loss": 1.0519,
+      "step": 11515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.154812016946591e-05,
+      "loss": 1.1591,
+      "step": 11520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1469295325799024e-05,
+      "loss": 1.1399,
+      "step": 11525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.139050990618047e-05,
+      "loss": 1.1471,
+      "step": 11530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.131176397461192e-05,
+      "loss": 1.1197,
+      "step": 11535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.123305759506277e-05,
+      "loss": 1.083,
+      "step": 11540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1154390831470443e-05,
+      "loss": 1.1562,
+      "step": 11545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.107576374774004e-05,
+      "loss": 1.1215,
+      "step": 11550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.099717640774446e-05,
+      "loss": 1.1243,
+      "step": 11555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.091862887532439e-05,
+      "loss": 1.1023,
+      "step": 11560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.084012121428808e-05,
+      "loss": 1.1031,
+      "step": 11565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.076165348841155e-05,
+      "loss": 1.1212,
+      "step": 11570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.0683225761438104e-05,
+      "loss": 1.0882,
+      "step": 11575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.0604838097078844e-05,
+      "loss": 1.1344,
+      "step": 11580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.0526490559012096e-05,
+      "loss": 1.113,
+      "step": 11585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.044818321088385e-05,
+      "loss": 1.1068,
+      "step": 11590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.036991611630713e-05,
+      "loss": 1.1007,
+      "step": 11595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.029168933886252e-05,
+      "loss": 1.0702,
+      "step": 11600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.021350294209767e-05,
+      "loss": 1.1364,
+      "step": 11605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.013535698952759e-05,
+      "loss": 1.1071,
+      "step": 11610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.005725154463433e-05,
+      "loss": 1.2017,
+      "step": 11615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9979186670867034e-05,
+      "loss": 1.1132,
+      "step": 11620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.99011624316419e-05,
+      "loss": 1.0277,
+      "step": 11625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.982317889034219e-05,
+      "loss": 1.1167,
+      "step": 11630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9745236110318016e-05,
+      "loss": 1.0663,
+      "step": 11635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9667334154886396e-05,
+      "loss": 1.1204,
+      "step": 11640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9589473087331173e-05,
+      "loss": 1.1003,
+      "step": 11645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.951165297090304e-05,
+      "loss": 1.1447,
+      "step": 11650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.943387386881936e-05,
+      "loss": 1.1249,
+      "step": 11655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.935613584426416e-05,
+      "loss": 1.115,
+      "step": 11660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9278438960388205e-05,
+      "loss": 1.098,
+      "step": 11665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.920078328030869e-05,
+      "loss": 1.0942,
+      "step": 11670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.912316886710948e-05,
+      "loss": 1.0794,
+      "step": 11675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9045595783840843e-05,
+      "loss": 1.1648,
+      "step": 11680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.896806409351945e-05,
+      "loss": 1.1045,
+      "step": 11685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.889057385912834e-05,
+      "loss": 1.1336,
+      "step": 11690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.881312514361702e-05,
+      "loss": 1.1451,
+      "step": 11695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873571800990111e-05,
+      "loss": 1.1223,
+      "step": 11700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.865835252086249e-05,
+      "loss": 1.1818,
+      "step": 11705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8581028739349186e-05,
+      "loss": 1.2081,
+      "step": 11710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.850374672817548e-05,
+      "loss": 1.1436,
+      "step": 11715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.842650655012156e-05,
+      "loss": 1.103,
+      "step": 11720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.834930826793371e-05,
+      "loss": 1.103,
+      "step": 11725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8272151944324115e-05,
+      "loss": 1.1512,
+      "step": 11730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8195037641971e-05,
+      "loss": 1.0671,
+      "step": 11735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8117965423518296e-05,
+      "loss": 1.0607,
+      "step": 11740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.804093535157593e-05,
+      "loss": 1.1404,
+      "step": 11745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.796394748871937e-05,
+      "loss": 1.0394,
+      "step": 11750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7887001897489995e-05,
+      "loss": 1.1789,
+      "step": 11755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7810098640394706e-05,
+      "loss": 1.0647,
+      "step": 11760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.773323777990616e-05,
+      "loss": 1.1012,
+      "step": 11765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7656419378462356e-05,
+      "loss": 1.0892,
+      "step": 11770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7579643498467e-05,
+      "loss": 1.163,
+      "step": 11775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.750291020228921e-05,
+      "loss": 1.0868,
+      "step": 11780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.742621955226347e-05,
+      "loss": 1.1675,
+      "step": 11785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.7349571610689625e-05,
+      "loss": 1.1235,
+      "step": 11790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.727296643983279e-05,
+      "loss": 1.1036,
+      "step": 11795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.719640410192346e-05,
+      "loss": 1.0805,
+      "step": 11800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.711988465915723e-05,
+      "loss": 1.0863,
+      "step": 11805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.7043408173694846e-05,
+      "loss": 1.1648,
+      "step": 11810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6966974707662195e-05,
+      "loss": 1.1605,
+      "step": 11815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.689058432315024e-05,
+      "loss": 1.1116,
+      "step": 11820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.681423708221487e-05,
+      "loss": 1.1126,
+      "step": 11825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.673793304687707e-05,
+      "loss": 1.1478,
+      "step": 11830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.666167227912247e-05,
+      "loss": 1.0948,
+      "step": 11835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.658545484090184e-05,
+      "loss": 1.1132,
+      "step": 11840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.650928079413055e-05,
+      "loss": 1.1758,
+      "step": 11845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6433150200688896e-05,
+      "loss": 1.1221,
+      "step": 11850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6357063122421613e-05,
+      "loss": 1.0648,
+      "step": 11855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.628101962113834e-05,
+      "loss": 1.109,
+      "step": 11860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6205019758613185e-05,
+      "loss": 1.2088,
+      "step": 11865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.612906359658489e-05,
+      "loss": 1.0691,
+      "step": 11870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.605315119675662e-05,
+      "loss": 1.1582,
+      "step": 11875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.597728262079599e-05,
+      "loss": 1.0383,
+      "step": 11880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.590145793033509e-05,
+      "loss": 1.0629,
+      "step": 11885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.582567718697023e-05,
+      "loss": 1.1229,
+      "step": 11890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.574994045226226e-05,
+      "loss": 1.1112,
+      "step": 11895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.567424778773592e-05,
+      "loss": 1.1251,
+      "step": 11900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.559859925488048e-05,
+      "loss": 1.1147,
+      "step": 11905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.552299491514917e-05,
+      "loss": 1.1154,
+      "step": 11910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5447434829959475e-05,
+      "loss": 1.122,
+      "step": 11915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.537191906069269e-05,
+      "loss": 1.2,
+      "step": 11920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5296447668694366e-05,
+      "loss": 1.0759,
+      "step": 11925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5221020715273816e-05,
+      "loss": 1.1403,
+      "step": 11930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.51456382617044e-05,
+      "loss": 1.1571,
+      "step": 11935
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5070300369223215e-05,
+      "loss": 1.1304,
+      "step": 11940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.499500709903121e-05,
+      "loss": 1.0566,
+      "step": 11945
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4919758512293036e-05,
+      "loss": 1.0922,
+      "step": 11950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4844554670137145e-05,
+      "loss": 1.1554,
+      "step": 11955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4769395633655554e-05,
+      "loss": 1.1302,
+      "step": 11960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.469428146390392e-05,
+      "loss": 1.0698,
+      "step": 11965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4619212221901376e-05,
+      "loss": 1.0959,
+      "step": 11970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.454418796863072e-05,
+      "loss": 1.0745,
+      "step": 11975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.446920876503807e-05,
+      "loss": 1.1268,
+      "step": 11980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.439427467203294e-05,
+      "loss": 1.0728,
+      "step": 11985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4319385750488315e-05,
+      "loss": 1.1349,
+      "step": 11990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.424454206124032e-05,
+      "loss": 1.1114,
+      "step": 11995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4169743665088565e-05,
+      "loss": 1.0812,
+      "step": 12000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.409499062279558e-05,
+      "loss": 1.1699,
+      "step": 12005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.402028299508729e-05,
+      "loss": 1.0348,
+      "step": 12010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.394562084265259e-05,
+      "loss": 1.121,
+      "step": 12015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.387100422614353e-05,
+      "loss": 1.1061,
+      "step": 12020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.37964332061751e-05,
+      "loss": 1.1052,
+      "step": 12025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.372190784332526e-05,
+      "loss": 1.0799,
+      "step": 12030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3647428198134845e-05,
+      "loss": 1.1057,
+      "step": 12035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.357299433110767e-05,
+      "loss": 1.1581,
+      "step": 12040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.349860630271027e-05,
+      "loss": 1.0838,
+      "step": 12045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.342426417337194e-05,
+      "loss": 1.142,
+      "step": 12050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.334996800348468e-05,
+      "loss": 1.0359,
+      "step": 12055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3275717853403255e-05,
+      "loss": 1.0875,
+      "step": 12060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.320151378344493e-05,
+      "loss": 1.128,
+      "step": 12065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3127355853889584e-05,
+      "loss": 1.12,
+      "step": 12070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.305324412497959e-05,
+      "loss": 1.084,
+      "step": 12075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.297917865691987e-05,
+      "loss": 1.1244,
+      "step": 12080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.290515950987763e-05,
+      "loss": 1.0532,
+      "step": 12085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.283118674398262e-05,
+      "loss": 1.0696,
+      "step": 12090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.27572604193267e-05,
+      "loss": 1.1579,
+      "step": 12095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.268338059596417e-05,
+      "loss": 1.0662,
+      "step": 12100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.260954733391156e-05,
+      "loss": 1.0738,
+      "step": 12105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.253576069314747e-05,
+      "loss": 1.1181,
+      "step": 12110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.2462020733612656e-05,
+      "loss": 1.0766,
+      "step": 12115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.2388327515209975e-05,
+      "loss": 1.1526,
+      "step": 12120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.231468109780436e-05,
+      "loss": 1.0989,
+      "step": 12125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.224108154122264e-05,
+      "loss": 1.1137,
+      "step": 12130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.216752890525362e-05,
+      "loss": 1.0953,
+      "step": 12135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.209402324964793e-05,
+      "loss": 1.0427,
+      "step": 12140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.202056463411818e-05,
+      "loss": 1.1042,
+      "step": 12145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.1947153118338596e-05,
+      "loss": 1.0568,
+      "step": 12150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.187378876194533e-05,
+      "loss": 1.1007,
+      "step": 12155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.180047162453595e-05,
+      "loss": 1.1365,
+      "step": 12160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.1727201765669956e-05,
+      "loss": 1.0907,
+      "step": 12165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.165397924486825e-05,
+      "loss": 1.0548,
+      "step": 12170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.158080412161344e-05,
+      "loss": 1.107,
+      "step": 12175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.150767645534939e-05,
+      "loss": 1.0936,
+      "step": 12180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.143459630548167e-05,
+      "loss": 1.1255,
+      "step": 12185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.1361563731377055e-05,
+      "loss": 1.1627,
+      "step": 12190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.128857879236385e-05,
+      "loss": 1.1539,
+      "step": 12195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.121564154773146e-05,
+      "loss": 1.0531,
+      "step": 12200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.114275205673069e-05,
+      "loss": 1.063,
+      "step": 12205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.1069910378573575e-05,
+      "loss": 1.1276,
+      "step": 12210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.099711657243317e-05,
+      "loss": 1.1043,
+      "step": 12215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.092437069744382e-05,
+      "loss": 1.1477,
+      "step": 12220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.085167281270068e-05,
+      "loss": 1.1241,
+      "step": 12225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.077902297726019e-05,
+      "loss": 1.1134,
+      "step": 12230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.070642125013956e-05,
+      "loss": 1.0757,
+      "step": 12235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.063386769031712e-05,
+      "loss": 1.1748,
+      "step": 12240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.056136235673179e-05,
+      "loss": 1.0801,
+      "step": 12245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.0488905308283596e-05,
+      "loss": 1.127,
+      "step": 12250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.041649660383313e-05,
+      "loss": 1.0326,
+      "step": 12255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.034413630220192e-05,
+      "loss": 1.1034,
+      "step": 12260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.027182446217188e-05,
+      "loss": 1.0684,
+      "step": 12265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.0199561142485864e-05,
+      "loss": 1.0526,
+      "step": 12270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.012734640184709e-05,
+      "loss": 1.1638,
+      "step": 12275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.005518029891947e-05,
+      "loss": 1.0238,
+      "step": 12280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.998306289232731e-05,
+      "loss": 1.221,
+      "step": 12285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.991099424065536e-05,
+      "loss": 1.1718,
+      "step": 12290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.983897440244875e-05,
+      "loss": 1.1192,
+      "step": 12295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.976700343621307e-05,
+      "loss": 1.0926,
+      "step": 12300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.969508140041409e-05,
+      "loss": 1.118,
+      "step": 12305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.962320835347782e-05,
+      "loss": 0.9906,
+      "step": 12310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.955138435379061e-05,
+      "loss": 1.1031,
+      "step": 12315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.94796094596988e-05,
+      "loss": 1.1357,
+      "step": 12320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.940788372950902e-05,
+      "loss": 1.1203,
+      "step": 12325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9336207221487735e-05,
+      "loss": 1.066,
+      "step": 12330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.926457999386164e-05,
+      "loss": 1.1502,
+      "step": 12335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.919300210481726e-05,
+      "loss": 1.062,
+      "step": 12340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.912147361250111e-05,
+      "loss": 1.1143,
+      "step": 12345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.904999457501957e-05,
+      "loss": 1.1342,
+      "step": 12350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.89785650504388e-05,
+      "loss": 1.0738,
+      "step": 12355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.890718509678472e-05,
+      "loss": 1.0528,
+      "step": 12360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.883585477204313e-05,
+      "loss": 1.0769,
+      "step": 12365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876457413415934e-05,
+      "loss": 1.0464,
+      "step": 12370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869334324103839e-05,
+      "loss": 1.1172,
+      "step": 12375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862216215054485e-05,
+      "loss": 1.1176,
+      "step": 12380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.855103092050295e-05,
+      "loss": 1.129,
+      "step": 12385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8479949608696285e-05,
+      "loss": 1.1021,
+      "step": 12390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.840891827286795e-05,
+      "loss": 1.0312,
+      "step": 12395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.83379369707204e-05,
+      "loss": 1.1089,
+      "step": 12400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.826700575991557e-05,
+      "loss": 1.1446,
+      "step": 12405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8196124698074564e-05,
+      "loss": 1.1468,
+      "step": 12410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8125293842777874e-05,
+      "loss": 1.0899,
+      "step": 12415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.805451325156502e-05,
+      "loss": 1.1116,
+      "step": 12420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.798378298193487e-05,
+      "loss": 1.0851,
+      "step": 12425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.791310309134537e-05,
+      "loss": 1.1426,
+      "step": 12430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.78424736372135e-05,
+      "loss": 1.1121,
+      "step": 12435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.777189467691529e-05,
+      "loss": 1.0453,
+      "step": 12440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.770136626778569e-05,
+      "loss": 1.1383,
+      "step": 12445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7630888467118734e-05,
+      "loss": 1.0837,
+      "step": 12450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7560461332167223e-05,
+      "loss": 1.0824,
+      "step": 12455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.749008492014281e-05,
+      "loss": 1.084,
+      "step": 12460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.741975928821595e-05,
+      "loss": 1.116,
+      "step": 12465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.734948449351591e-05,
+      "loss": 1.0822,
+      "step": 12470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7279260593130584e-05,
+      "loss": 1.0547,
+      "step": 12475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.720908764410653e-05,
+      "loss": 1.1336,
+      "step": 12480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.713896570344891e-05,
+      "loss": 1.1628,
+      "step": 12485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7068894828121545e-05,
+      "loss": 1.136,
+      "step": 12490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.699887507504664e-05,
+      "loss": 1.0782,
+      "step": 12495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6928906501105005e-05,
+      "loss": 1.132,
+      "step": 12500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6858989163135656e-05,
+      "loss": 1.0995,
+      "step": 12505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6789123117936266e-05,
+      "loss": 1.0418,
+      "step": 12510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.671930842226262e-05,
+      "loss": 1.1233,
+      "step": 12515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.664954513282898e-05,
+      "loss": 1.1218,
+      "step": 12520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.657983330630761e-05,
+      "loss": 1.1092,
+      "step": 12525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6510172999329144e-05,
+      "loss": 0.9964,
+      "step": 12530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6440564268482386e-05,
+      "loss": 1.1084,
+      "step": 12535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.637100717031411e-05,
+      "loss": 1.0999,
+      "step": 12540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6301501761329226e-05,
+      "loss": 1.1879,
+      "step": 12545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.623204809799061e-05,
+      "loss": 1.1341,
+      "step": 12550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6162646236719204e-05,
+      "loss": 1.1048,
+      "step": 12555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.609329623389372e-05,
+      "loss": 1.0768,
+      "step": 12560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.602399814585093e-05,
+      "loss": 1.0301,
+      "step": 12565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.595475202888518e-05,
+      "loss": 1.1138,
+      "step": 12570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.588555793924885e-05,
+      "loss": 1.129,
+      "step": 12575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5816415933151864e-05,
+      "loss": 1.1351,
+      "step": 12580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5747326066762005e-05,
+      "loss": 1.1261,
+      "step": 12585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.567828839620451e-05,
+      "loss": 1.1093,
+      "step": 12590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.560930297756239e-05,
+      "loss": 1.1134,
+      "step": 12595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.554036986687609e-05,
+      "loss": 1.0971,
+      "step": 12600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.547148912014368e-05,
+      "loss": 1.1343,
+      "step": 12605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.540266079332052e-05,
+      "loss": 1.075,
+      "step": 12610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.533388494231956e-05,
+      "loss": 1.0977,
+      "step": 12615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5265161623011004e-05,
+      "loss": 1.1513,
+      "step": 12620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5196490891222467e-05,
+      "loss": 1.1216,
+      "step": 12625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.512787280273878e-05,
+      "loss": 1.1924,
+      "step": 12630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5059307413302026e-05,
+      "loss": 1.0665,
+      "step": 12635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.499079477861148e-05,
+      "loss": 1.068,
+      "step": 12640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4922334954323564e-05,
+      "loss": 1.143,
+      "step": 12645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.485392799605186e-05,
+      "loss": 1.1131,
+      "step": 12650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.478557395936686e-05,
+      "loss": 1.1357,
+      "step": 12655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.471727289979619e-05,
+      "loss": 1.1969,
+      "step": 12660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4649024872824374e-05,
+      "loss": 1.181,
+      "step": 12665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.458082993389296e-05,
+      "loss": 1.2171,
+      "step": 12670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.451268813840017e-05,
+      "loss": 1.1365,
+      "step": 12675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4444599541701256e-05,
+      "loss": 1.1166,
+      "step": 12680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.437656419910813e-05,
+      "loss": 1.1471,
+      "step": 12685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.4308582165889556e-05,
+      "loss": 1.0413,
+      "step": 12690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.4240653497270855e-05,
+      "loss": 1.0813,
+      "step": 12695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.417277824843409e-05,
+      "loss": 1.0251,
+      "step": 12700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.410495647451786e-05,
+      "loss": 1.0604,
+      "step": 12705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.4037188230617426e-05,
+      "loss": 1.1318,
+      "step": 12710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.396947357178449e-05,
+      "loss": 1.1136,
+      "step": 12715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.39018125530272e-05,
+      "loss": 1.0477,
+      "step": 12720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.383420522931017e-05,
+      "loss": 1.1374,
+      "step": 12725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.376665165555445e-05,
+      "loss": 1.1639,
+      "step": 12730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3699151886637325e-05,
+      "loss": 1.093,
+      "step": 12735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3631705977392414e-05,
+      "loss": 1.1244,
+      "step": 12740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3564313982609544e-05,
+      "loss": 1.1487,
+      "step": 12745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3496975957034825e-05,
+      "loss": 1.1063,
+      "step": 12750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3429691955370526e-05,
+      "loss": 1.1197,
+      "step": 12755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3362462032274924e-05,
+      "loss": 1.1169,
+      "step": 12760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.329528624236247e-05,
+      "loss": 1.1236,
+      "step": 12765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3228164640203524e-05,
+      "loss": 1.141,
+      "step": 12770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3161097280324594e-05,
+      "loss": 1.1073,
+      "step": 12775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3094084217207994e-05,
+      "loss": 1.1008,
+      "step": 12780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3027125505291964e-05,
+      "loss": 1.133,
+      "step": 12785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.296022119897054e-05,
+      "loss": 1.1476,
+      "step": 12790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.28933713525937e-05,
+      "loss": 1.112,
+      "step": 12795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.282657602046706e-05,
+      "loss": 1.1707,
+      "step": 12800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2759835256851994e-05,
+      "loss": 1.0002,
+      "step": 12805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.269314911596549e-05,
+      "loss": 1.1404,
+      "step": 12810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.262651765198029e-05,
+      "loss": 1.0785,
+      "step": 12815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2559940919024584e-05,
+      "loss": 1.1305,
+      "step": 12820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2493418971182253e-05,
+      "loss": 1.2168,
+      "step": 12825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.242695186249245e-05,
+      "loss": 1.0668,
+      "step": 12830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.236053964695e-05,
+      "loss": 1.1709,
+      "step": 12835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2294182378505e-05,
+      "loss": 1.1051,
+      "step": 12840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.222788011106303e-05,
+      "loss": 1.1427,
+      "step": 12845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2161632898484786e-05,
+      "loss": 1.1116,
+      "step": 12850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2095440794586457e-05,
+      "loss": 1.1097,
+      "step": 12855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.202930385313938e-05,
+      "loss": 1.0516,
+      "step": 12860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.196322212787007e-05,
+      "loss": 1.1658,
+      "step": 12865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.189719567246018e-05,
+      "loss": 1.1148,
+      "step": 12870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.183122454054646e-05,
+      "loss": 1.1051,
+      "step": 12875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.176530878572077e-05,
+      "loss": 1.0867,
+      "step": 12880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.169944846152996e-05,
+      "loss": 1.124,
+      "step": 12885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.16336436214758e-05,
+      "loss": 1.198,
+      "step": 12890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.156789431901503e-05,
+      "loss": 1.0649,
+      "step": 12895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1502200607559316e-05,
+      "loss": 1.1083,
+      "step": 12900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.143656254047509e-05,
+      "loss": 1.1248,
+      "step": 12905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1370980171083654e-05,
+      "loss": 1.0997,
+      "step": 12910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.130545355266094e-05,
+      "loss": 1.1382,
+      "step": 12915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.123998273843774e-05,
+      "loss": 1.1247,
+      "step": 12920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.11745677815994e-05,
+      "loss": 1.219,
+      "step": 12925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.110920873528603e-05,
+      "loss": 1.0724,
+      "step": 12930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1043905652592085e-05,
+      "loss": 1.1091,
+      "step": 12935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0978658586566806e-05,
+      "loss": 1.1715,
+      "step": 12940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.091346759021377e-05,
+      "loss": 1.1078,
+      "step": 12945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0848332716491084e-05,
+      "loss": 1.1069,
+      "step": 12950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.078325401831125e-05,
+      "loss": 1.0679,
+      "step": 12955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.071823154854111e-05,
+      "loss": 1.1401,
+      "step": 12960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.065326536000181e-05,
+      "loss": 1.1073,
+      "step": 12965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.058835550546883e-05,
+      "loss": 1.119,
+      "step": 12970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.052350203767193e-05,
+      "loss": 1.1249,
+      "step": 12975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0458705009294886e-05,
+      "loss": 1.0811,
+      "step": 12980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0393964472975814e-05,
+      "loss": 1.1094,
+      "step": 12985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.032928048130681e-05,
+      "loss": 1.112,
+      "step": 12990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0264653086834182e-05,
+      "loss": 1.1141,
+      "step": 12995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0200082342058043e-05,
+      "loss": 1.0816,
+      "step": 13000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.01355682994327e-05,
+      "loss": 1.1427,
+      "step": 13005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0071111011366258e-05,
+      "loss": 1.1684,
+      "step": 13010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0006710530220815e-05,
+      "loss": 1.1816,
+      "step": 13015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9942366908312257e-05,
+      "loss": 1.1468,
+      "step": 13020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.987808019791029e-05,
+      "loss": 1.1237,
+      "step": 13025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9813850451238364e-05,
+      "loss": 1.1578,
+      "step": 13030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9749677720473768e-05,
+      "loss": 1.0485,
+      "step": 13035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9685562057747327e-05,
+      "loss": 1.1203,
+      "step": 13040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.962150351514361e-05,
+      "loss": 1.2114,
+      "step": 13045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9557502144700686e-05,
+      "loss": 1.1761,
+      "step": 13050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.949355799841031e-05,
+      "loss": 1.0751,
+      "step": 13055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.942967112821765e-05,
+      "loss": 1.1175,
+      "step": 13060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9365841586021382e-05,
+      "loss": 1.1024,
+      "step": 13065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.930206942367357e-05,
+      "loss": 1.0135,
+      "step": 13070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.923835469297975e-05,
+      "loss": 1.1041,
+      "step": 13075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.917469744569879e-05,
+      "loss": 1.1339,
+      "step": 13080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.911109773354272e-05,
+      "loss": 1.1029,
+      "step": 13085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.904755560817702e-05,
+      "loss": 1.1495,
+      "step": 13090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.898407112122024e-05,
+      "loss": 1.2055,
+      "step": 13095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.892064432424425e-05,
+      "loss": 1.0567,
+      "step": 13100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8857275268773956e-05,
+      "loss": 1.0984,
+      "step": 13105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8793964006287357e-05,
+      "loss": 1.1613,
+      "step": 13110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.87307105882155e-05,
+      "loss": 1.0997,
+      "step": 13115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8667515065942552e-05,
+      "loss": 1.0995,
+      "step": 13120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8604377490805517e-05,
+      "loss": 1.1035,
+      "step": 13125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8541297914094368e-05,
+      "loss": 1.1854,
+      "step": 13130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8478276387051948e-05,
+      "loss": 1.1477,
+      "step": 13135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8415312960874014e-05,
+      "loss": 1.1755,
+      "step": 13140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.835240768670906e-05,
+      "loss": 1.1125,
+      "step": 13145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8289560615658364e-05,
+      "loss": 1.1011,
+      "step": 13150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8226771798775853e-05,
+      "loss": 1.1177,
+      "step": 13155
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8164041287068264e-05,
+      "loss": 1.1192,
+      "step": 13160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.810136913149484e-05,
+      "loss": 1.1319,
+      "step": 13165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8038755382967577e-05,
+      "loss": 1.0785,
+      "step": 13170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7976200092350784e-05,
+      "loss": 1.0592,
+      "step": 13175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7913703310461507e-05,
+      "loss": 1.1038,
+      "step": 13180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.785126508806913e-05,
+      "loss": 1.1802,
+      "step": 13185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7788885475895558e-05,
+      "loss": 1.0879,
+      "step": 13190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.772656452461501e-05,
+      "loss": 1.1564,
+      "step": 13195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.766430228485405e-05,
+      "loss": 1.1016,
+      "step": 13200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7602098807191637e-05,
+      "loss": 1.1713,
+      "step": 13205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.75399541421589e-05,
+      "loss": 1.0543,
+      "step": 13210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.747786834023924e-05,
+      "loss": 1.1137,
+      "step": 13215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7415841451868165e-05,
+      "loss": 1.0905,
+      "step": 13220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7353873527433476e-05,
+      "loss": 1.0709,
+      "step": 13225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7291964617274895e-05,
+      "loss": 1.0793,
+      "step": 13230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7230114771684407e-05,
+      "loss": 1.1697,
+      "step": 13235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.716832404090575e-05,
+      "loss": 1.0943,
+      "step": 13240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7106592475134916e-05,
+      "loss": 1.0944,
+      "step": 13245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7044920124519638e-05,
+      "loss": 1.1024,
+      "step": 13250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6983307039159674e-05,
+      "loss": 1.1247,
+      "step": 13255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6921753269106496e-05,
+      "loss": 1.1211,
+      "step": 13260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.686025886436356e-05,
+      "loss": 1.181,
+      "step": 13265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.679882387488593e-05,
+      "loss": 1.0776,
+      "step": 13270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6737448350580598e-05,
+      "loss": 1.081,
+      "step": 13275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6676132341306005e-05,
+      "loss": 1.0723,
+      "step": 13280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.661487589687246e-05,
+      "loss": 1.0979,
+      "step": 13285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6553679067041762e-05,
+      "loss": 1.1002,
+      "step": 13290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6492541901527327e-05,
+      "loss": 1.1407,
+      "step": 13295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.643146444999417e-05,
+      "loss": 1.0677,
+      "step": 13300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6370446762058602e-05,
+      "loss": 1.1042,
+      "step": 13305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.630948888728859e-05,
+      "loss": 1.0832,
+      "step": 13310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6248590875203353e-05,
+      "loss": 1.1264,
+      "step": 13315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.618775277527368e-05,
+      "loss": 1.1957,
+      "step": 13320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6126974636921422e-05,
+      "loss": 1.0779,
+      "step": 13325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6066256509519965e-05,
+      "loss": 1.0938,
+      "step": 13330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.600559844239377e-05,
+      "loss": 1.1281,
+      "step": 13335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5945000484818674e-05,
+      "loss": 1.136,
+      "step": 13340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.588446268602148e-05,
+      "loss": 1.111,
+      "step": 13345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.582398509518031e-05,
+      "loss": 1.1041,
+      "step": 13350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.576356776142424e-05,
+      "loss": 1.0408,
+      "step": 13355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5703210733833508e-05,
+      "loss": 1.1333,
+      "step": 13360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.56429140614393e-05,
+      "loss": 1.1083,
+      "step": 13365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.558267779322377e-05,
+      "loss": 1.1371,
+      "step": 13370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5522501978119967e-05,
+      "loss": 1.1701,
+      "step": 13375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5462386665011944e-05,
+      "loss": 1.1124,
+      "step": 13380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.540233190273452e-05,
+      "loss": 1.145,
+      "step": 13385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5342337740073327e-05,
+      "loss": 1.0931,
+      "step": 13390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5282404225764765e-05,
+      "loss": 1.1021,
+      "step": 13395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5222531408496007e-05,
+      "loss": 1.0982,
+      "step": 13400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.516271933690497e-05,
+      "loss": 1.0976,
+      "step": 13405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.510296805958001e-05,
+      "loss": 1.0351,
+      "step": 13410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5043277625060342e-05,
+      "loss": 1.0522,
+      "step": 13415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.498364808183559e-05,
+      "loss": 1.149,
+      "step": 13420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.492407947834603e-05,
+      "loss": 1.0945,
+      "step": 13425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4864571862982354e-05,
+      "loss": 1.1975,
+      "step": 13430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.480512528408573e-05,
+      "loss": 1.0145,
+      "step": 13435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4745739789947707e-05,
+      "loss": 1.0774,
+      "step": 13440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4686415428810324e-05,
+      "loss": 1.0966,
+      "step": 13445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.462715224886584e-05,
+      "loss": 1.1288,
+      "step": 13450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.456795029825688e-05,
+      "loss": 1.1367,
+      "step": 13455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.450880962507626e-05,
+      "loss": 1.1045,
+      "step": 13460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4449730277367138e-05,
+      "loss": 1.098,
+      "step": 13465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4390712303122747e-05,
+      "loss": 1.1466,
+      "step": 13470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4331755750286497e-05,
+      "loss": 1.1213,
+      "step": 13475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4272860666751874e-05,
+      "loss": 1.107,
+      "step": 13480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4214027100362504e-05,
+      "loss": 1.0435,
+      "step": 13485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4155255098911936e-05,
+      "loss": 1.1201,
+      "step": 13490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4096544710143864e-05,
+      "loss": 1.182,
+      "step": 13495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.403789598175171e-05,
+      "loss": 1.1396,
+      "step": 13500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3979308961379008e-05,
+      "loss": 1.1197,
+      "step": 13505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.392078369661903e-05,
+      "loss": 1.0763,
+      "step": 13510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.386232023501497e-05,
+      "loss": 1.0941,
+      "step": 13515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3803918624059778e-05,
+      "loss": 1.083,
+      "step": 13520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3745578911196125e-05,
+      "loss": 1.1338,
+      "step": 13525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3687301143816476e-05,
+      "loss": 1.1166,
+      "step": 13530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3629085369262915e-05,
+      "loss": 1.1572,
+      "step": 13535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3570931634827172e-05,
+      "loss": 1.0818,
+      "step": 13540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.351283998775057e-05,
+      "loss": 1.0831,
+      "step": 13545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3454810475224076e-05,
+      "loss": 1.046,
+      "step": 13550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3396843144388093e-05,
+      "loss": 1.0966,
+      "step": 13555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3338938042332523e-05,
+      "loss": 1.1668,
+      "step": 13560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3281095216096715e-05,
+      "loss": 1.1433,
+      "step": 13565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.32233147126695e-05,
+      "loss": 1.2075,
+      "step": 13570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.316559657898896e-05,
+      "loss": 1.0886,
+      "step": 13575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3107940861942667e-05,
+      "loss": 1.1546,
+      "step": 13580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.30503476083673e-05,
+      "loss": 1.0657,
+      "step": 13585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.2992816865048957e-05,
+      "loss": 1.0969,
+      "step": 13590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.293534867872283e-05,
+      "loss": 1.077,
+      "step": 13595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.287794309607346e-05,
+      "loss": 1.1192,
+      "step": 13600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2820600163734308e-05,
+      "loss": 1.0944,
+      "step": 13605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2763319928288128e-05,
+      "loss": 1.1207,
+      "step": 13610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.270610243626664e-05,
+      "loss": 1.1778,
+      "step": 13615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2648947734150673e-05,
+      "loss": 1.0632,
+      "step": 13620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.259185586836997e-05,
+      "loss": 1.1142,
+      "step": 13625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.253482688530324e-05,
+      "loss": 1.1039,
+      "step": 13630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.247786083127818e-05,
+      "loss": 1.1981,
+      "step": 13635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2420957752571277e-05,
+      "loss": 1.0875,
+      "step": 13640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2364117695407966e-05,
+      "loss": 1.1427,
+      "step": 13645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.230734070596232e-05,
+      "loss": 1.1005,
+      "step": 13650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.225062683035735e-05,
+      "loss": 1.1464,
+      "step": 13655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.219397611466468e-05,
+      "loss": 1.1583,
+      "step": 13660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2137388604904757e-05,
+      "loss": 1.0324,
+      "step": 13665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.208086434704649e-05,
+      "loss": 1.1044,
+      "step": 13670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.202440338700757e-05,
+      "loss": 1.1663,
+      "step": 13675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1968005770654178e-05,
+      "loss": 1.0342,
+      "step": 13680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1911671543801118e-05,
+      "loss": 1.1859,
+      "step": 13685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.185540075221162e-05,
+      "loss": 1.1405,
+      "step": 13690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1799193441597412e-05,
+      "loss": 1.1567,
+      "step": 13695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.174304965761864e-05,
+      "loss": 1.0746,
+      "step": 13700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1686969445883893e-05,
+      "loss": 1.103,
+      "step": 13705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.163095285195006e-05,
+      "loss": 1.0605,
+      "step": 13710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1574999921322382e-05,
+      "loss": 1.0985,
+      "step": 13715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.151911069945434e-05,
+      "loss": 1.0536,
+      "step": 13720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.146328523174774e-05,
+      "loss": 1.0509,
+      "step": 13725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1407523563552542e-05,
+      "loss": 1.1405,
+      "step": 13730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.135182574016684e-05,
+      "loss": 1.139,
+      "step": 13735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.129619180683697e-05,
+      "loss": 1.158,
+      "step": 13740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1240621808757265e-05,
+      "loss": 1.1466,
+      "step": 13745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.118511579107024e-05,
+      "loss": 1.1264,
+      "step": 13750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1129673798866227e-05,
+      "loss": 1.1122,
+      "step": 13755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1074295877183792e-05,
+      "loss": 1.1042,
+      "step": 13760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1018982071009263e-05,
+      "loss": 1.0984,
+      "step": 13765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.0963732425277016e-05,
+      "loss": 1.1177,
+      "step": 13770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.0908546984869227e-05,
+      "loss": 1.0712,
+      "step": 13775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.085342579461593e-05,
+      "loss": 1.0738,
+      "step": 13780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0798368899294965e-05,
+      "loss": 1.1011,
+      "step": 13785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0743376343631973e-05,
+      "loss": 1.0885,
+      "step": 13790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.068844817230029e-05,
+      "loss": 1.1315,
+      "step": 13795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.063358442992097e-05,
+      "loss": 1.1061,
+      "step": 13800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0578785161062694e-05,
+      "loss": 1.1283,
+      "step": 13805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0524050410241843e-05,
+      "loss": 1.0953,
+      "step": 13810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0469380221922317e-05,
+      "loss": 1.1648,
+      "step": 13815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0414774640515587e-05,
+      "loss": 1.0729,
+      "step": 13820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.036023371038061e-05,
+      "loss": 1.1419,
+      "step": 13825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.030575747582393e-05,
+      "loss": 1.1052,
+      "step": 13830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0251345981099388e-05,
+      "loss": 1.0747,
+      "step": 13835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0196999270408374e-05,
+      "loss": 1.1237,
+      "step": 13840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0142717387899564e-05,
+      "loss": 1.1346,
+      "step": 13845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0088500377668944e-05,
+      "loss": 1.0706,
+      "step": 13850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.003434828375992e-05,
+      "loss": 1.1789,
+      "step": 13855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.998026115016306e-05,
+      "loss": 1.0862,
+      "step": 13860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9926239020816193e-05,
+      "loss": 1.1025,
+      "step": 13865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9872281939604308e-05,
+      "loss": 1.0944,
+      "step": 13870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.981838995035964e-05,
+      "loss": 1.1507,
+      "step": 13875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9764563096861463e-05,
+      "loss": 1.1189,
+      "step": 13880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9710801422836175e-05,
+      "loss": 1.0884,
+      "step": 13885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.965710497195719e-05,
+      "loss": 1.1377,
+      "step": 13890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9603473787844994e-05,
+      "loss": 1.1438,
+      "step": 13895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9549907914067013e-05,
+      "loss": 1.1214,
+      "step": 13900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9496407394137676e-05,
+      "loss": 1.1477,
+      "step": 13905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.944297227151817e-05,
+      "loss": 1.0481,
+      "step": 13910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9389602589616738e-05,
+      "loss": 1.184,
+      "step": 13915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.933629839178835e-05,
+      "loss": 1.1532,
+      "step": 13920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9283059721334877e-05,
+      "loss": 1.0982,
+      "step": 13925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9229886621504803e-05,
+      "loss": 1.1322,
+      "step": 13930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.917677913549351e-05,
+      "loss": 1.156,
+      "step": 13935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9123737306442957e-05,
+      "loss": 1.1052,
+      "step": 13940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9070761177441863e-05,
+      "loss": 1.0726,
+      "step": 13945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.901785079152551e-05,
+      "loss": 1.1467,
+      "step": 13950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8965006191675762e-05,
+      "loss": 1.1082,
+      "step": 13955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.891222742082113e-05,
+      "loss": 1.1113,
+      "step": 13960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8859514521836554e-05,
+      "loss": 1.121,
+      "step": 13965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8806867537543508e-05,
+      "loss": 1.0504,
+      "step": 13970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.875428651070989e-05,
+      "loss": 1.0968,
+      "step": 13975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8701771484050075e-05,
+      "loss": 1.0976,
+      "step": 13980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8649322500224738e-05,
+      "loss": 1.1125,
+      "step": 13985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8596939601841044e-05,
+      "loss": 1.1515,
+      "step": 13990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8544622831452263e-05,
+      "loss": 1.1904,
+      "step": 13995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.849237223155814e-05,
+      "loss": 1.1123,
+      "step": 14000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8440187844604552e-05,
+      "loss": 1.1667,
+      "step": 14005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.838806971298369e-05,
+      "loss": 1.1128,
+      "step": 14010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8336017879033766e-05,
+      "loss": 1.1282,
+      "step": 14015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8284032385039298e-05,
+      "loss": 1.1173,
+      "step": 14020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.823211327323079e-05,
+      "loss": 1.0989,
+      "step": 14025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.818026058578493e-05,
+      "loss": 1.0901,
+      "step": 14030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8128474364824348e-05,
+      "loss": 1.1538,
+      "step": 14035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.807675465241774e-05,
+      "loss": 1.1019,
+      "step": 14040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8025101490579708e-05,
+      "loss": 1.0927,
+      "step": 14045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7973514921270904e-05,
+      "loss": 1.1338,
+      "step": 14050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7921994986397788e-05,
+      "loss": 1.0442,
+      "step": 14055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.787054172781269e-05,
+      "loss": 1.1012,
+      "step": 14060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7819155187313846e-05,
+      "loss": 1.1339,
+      "step": 14065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.776783540664521e-05,
+      "loss": 1.1106,
+      "step": 14070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7716582427496617e-05,
+      "loss": 1.184,
+      "step": 14075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7665396291503452e-05,
+      "loss": 1.156,
+      "step": 14080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7614277040247003e-05,
+      "loss": 1.1387,
+      "step": 14085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.756322471525406e-05,
+      "loss": 1.1431,
+      "step": 14090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.751223935799719e-05,
+      "loss": 1.1367,
+      "step": 14095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7461321009894452e-05,
+      "loss": 1.187,
+      "step": 14100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7410469712309488e-05,
+      "loss": 1.0791,
+      "step": 14105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7359685506551474e-05,
+      "loss": 1.0798,
+      "step": 14110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.730896843387515e-05,
+      "loss": 1.1235,
+      "step": 14115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7258318535480632e-05,
+      "loss": 1.076,
+      "step": 14120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7207735852513505e-05,
+      "loss": 1.0844,
+      "step": 14125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7157220426064732e-05,
+      "loss": 1.1181,
+      "step": 14130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7106772297170688e-05,
+      "loss": 1.1591,
+      "step": 14135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7056391506813026e-05,
+      "loss": 1.1628,
+      "step": 14140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.700607809591873e-05,
+      "loss": 1.0763,
+      "step": 14145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.695583210535999e-05,
+      "loss": 1.1178,
+      "step": 14150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6905653575954345e-05,
+      "loss": 1.1503,
+      "step": 14155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6855542548464414e-05,
+      "loss": 1.096,
+      "step": 14160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6805499063598006e-05,
+      "loss": 1.0913,
+      "step": 14165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6755523162008128e-05,
+      "loss": 1.1116,
+      "step": 14170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.670561488429279e-05,
+      "loss": 1.1686,
+      "step": 14175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.665577427099517e-05,
+      "loss": 1.0464,
+      "step": 14180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.660600136260338e-05,
+      "loss": 1.1262,
+      "step": 14185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6556296199550593e-05,
+      "loss": 1.1405,
+      "step": 14190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6506658822214906e-05,
+      "loss": 1.0876,
+      "step": 14195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6457089270919424e-05,
+      "loss": 1.0455,
+      "step": 14200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6407587585932083e-05,
+      "loss": 1.0927,
+      "step": 14205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6358153807465704e-05,
+      "loss": 1.1686,
+      "step": 14210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6308787975677943e-05,
+      "loss": 1.1128,
+      "step": 14215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6259490130671283e-05,
+      "loss": 1.1565,
+      "step": 14220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6210260312492962e-05,
+      "loss": 1.1049,
+      "step": 14225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.616109856113494e-05,
+      "loss": 1.1708,
+      "step": 14230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.611200491653389e-05,
+      "loss": 1.1054,
+      "step": 14235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.606297941857121e-05,
+      "loss": 1.1213,
+      "step": 14240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.601402210707283e-05,
+      "loss": 1.1641,
+      "step": 14245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.596513302180945e-05,
+      "loss": 1.0863,
+      "step": 14250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.591631220249613e-05,
+      "loss": 1.0899,
+      "step": 14255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5867559688792686e-05,
+      "loss": 1.156,
+      "step": 14260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5818875520303297e-05,
+      "loss": 1.0327,
+      "step": 14265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5770259736576746e-05,
+      "loss": 1.0896,
+      "step": 14270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5721712377106107e-05,
+      "loss": 1.0675,
+      "step": 14275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5673233481329e-05,
+      "loss": 1.0433,
+      "step": 14280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5624823088627418e-05,
+      "loss": 1.1567,
+      "step": 14285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5576481238327648e-05,
+      "loss": 1.1344,
+      "step": 14290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5528207969700317e-05,
+      "loss": 1.1423,
+      "step": 14295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5480003321960316e-05,
+      "loss": 1.0844,
+      "step": 14300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5431867334266886e-05,
+      "loss": 1.1175,
+      "step": 14305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5383800045723364e-05,
+      "loss": 1.094,
+      "step": 14310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5335801495377417e-05,
+      "loss": 1.1264,
+      "step": 14315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5287871722220698e-05,
+      "loss": 1.1052,
+      "step": 14320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5240010765189155e-05,
+      "loss": 1.1341,
+      "step": 14325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5192218663162729e-05,
+      "loss": 1.0309,
+      "step": 14330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5144495454965528e-05,
+      "loss": 1.1309,
+      "step": 14335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.509684117936553e-05,
+      "loss": 1.1607,
+      "step": 14340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5049255875074885e-05,
+      "loss": 1.1157,
+      "step": 14345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5001739580749606e-05,
+      "loss": 1.2165,
+      "step": 14350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4954292334989738e-05,
+      "loss": 1.1538,
+      "step": 14355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4906914176339105e-05,
+      "loss": 1.06,
+      "step": 14360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4859605143285526e-05,
+      "loss": 1.1644,
+      "step": 14365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4812365274260596e-05,
+      "loss": 1.0295,
+      "step": 14370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4765194607639782e-05,
+      "loss": 1.0812,
+      "step": 14375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4718093181742291e-05,
+      "loss": 1.0472,
+      "step": 14380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4671061034831063e-05,
+      "loss": 1.0802,
+      "step": 14385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.462409820511284e-05,
+      "loss": 1.1038,
+      "step": 14390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4577204730737952e-05,
+      "loss": 1.1691,
+      "step": 14395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4530380649800507e-05,
+      "loss": 1.129,
+      "step": 14400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4483626000338084e-05,
+      "loss": 1.1195,
+      "step": 14405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4436940820332006e-05,
+      "loss": 1.0641,
+      "step": 14410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4390325147707078e-05,
+      "loss": 1.1007,
+      "step": 14415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4343779020331729e-05,
+      "loss": 1.1644,
+      "step": 14420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4297302476017729e-05,
+      "loss": 1.1147,
+      "step": 14425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4250895552520505e-05,
+      "loss": 1.0384,
+      "step": 14430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4204558287538793e-05,
+      "loss": 1.13,
+      "step": 14435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4158290718714839e-05,
+      "loss": 1.0701,
+      "step": 14440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4112092883634211e-05,
+      "loss": 1.0848,
+      "step": 14445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4065964819825853e-05,
+      "loss": 1.0651,
+      "step": 14450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4019906564761966e-05,
+      "loss": 1.0654,
+      "step": 14455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3973918155858178e-05,
+      "loss": 1.1258,
+      "step": 14460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3927999630473253e-05,
+      "loss": 1.0602,
+      "step": 14465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3882151025909237e-05,
+      "loss": 1.1346,
+      "step": 14470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.383637237941135e-05,
+      "loss": 1.1683,
+      "step": 14475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3790663728168021e-05,
+      "loss": 1.0821,
+      "step": 14480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.37450251093108e-05,
+      "loss": 1.069,
+      "step": 14485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3699456559914309e-05,
+      "loss": 1.0604,
+      "step": 14490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.365395811699633e-05,
+      "loss": 1.1048,
+      "step": 14495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3608529817517584e-05,
+      "loss": 1.121,
+      "step": 14500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3563171698381941e-05,
+      "loss": 1.0915,
+      "step": 14505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3517883796436152e-05,
+      "loss": 1.1266,
+      "step": 14510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3472666148469976e-05,
+      "loss": 1.1148,
+      "step": 14515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3427518791216048e-05,
+      "loss": 1.1668,
+      "step": 14520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3382441761350023e-05,
+      "loss": 1.0674,
+      "step": 14525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3337435095490296e-05,
+      "loss": 1.1686,
+      "step": 14530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3292498830198169e-05,
+      "loss": 1.1115,
+      "step": 14535
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3247633001977705e-05,
+      "loss": 1.1087,
+      "step": 14540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3202837647275842e-05,
+      "loss": 1.1455,
+      "step": 14545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3158112802482167e-05,
+      "loss": 1.1759,
+      "step": 14550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3113458503929029e-05,
+      "loss": 1.1278,
+      "step": 14555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3068874787891457e-05,
+      "loss": 1.0648,
+      "step": 14560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.302436169058719e-05,
+      "loss": 1.1783,
+      "step": 14565
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2979919248176514e-05,
+      "loss": 1.0736,
+      "step": 14570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2935547496762438e-05,
+      "loss": 1.1644,
+      "step": 14575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2891246472390373e-05,
+      "loss": 1.0614,
+      "step": 14580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.284701621104847e-05,
+      "loss": 1.1576,
+      "step": 14585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2802856748667237e-05,
+      "loss": 1.1394,
+      "step": 14590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2758768121119792e-05,
+      "loss": 1.0738,
+      "step": 14595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2714750364221584e-05,
+      "loss": 1.0264,
+      "step": 14600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2670803513730577e-05,
+      "loss": 1.0791,
+      "step": 14605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2626927605347147e-05,
+      "loss": 1.0826,
+      "step": 14610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2583122674713974e-05,
+      "loss": 1.1423,
+      "step": 14615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2539388757416136e-05,
+      "loss": 1.0908,
+      "step": 14620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2495725888980948e-05,
+      "loss": 1.1082,
+      "step": 14625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2452134104878124e-05,
+      "loss": 1.1416,
+      "step": 14630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2408613440519524e-05,
+      "loss": 1.1877,
+      "step": 14635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2365163931259293e-05,
+      "loss": 1.1972,
+      "step": 14640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2321785612393721e-05,
+      "loss": 1.2007,
+      "step": 14645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2278478519161363e-05,
+      "loss": 1.1861,
+      "step": 14650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2235242686742796e-05,
+      "loss": 1.1312,
+      "step": 14655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2192078150260844e-05,
+      "loss": 1.1126,
+      "step": 14660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2148984944780229e-05,
+      "loss": 1.152,
+      "step": 14665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2105963105307906e-05,
+      "loss": 1.0607,
+      "step": 14670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.206301266679274e-05,
+      "loss": 1.1657,
+      "step": 14675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2020133664125698e-05,
+      "loss": 1.1544,
+      "step": 14680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1977326132139577e-05,
+      "loss": 1.0904,
+      "step": 14685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1934590105609232e-05,
+      "loss": 1.0232,
+      "step": 14690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1891925619251365e-05,
+      "loss": 1.1031,
+      "step": 14695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1849332707724626e-05,
+      "loss": 1.1519,
+      "step": 14700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.180681140562946e-05,
+      "loss": 1.0821,
+      "step": 14705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1764361747508125e-05,
+      "loss": 1.1709,
+      "step": 14710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.172198376784478e-05,
+      "loss": 1.0843,
+      "step": 14715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1679677501065233e-05,
+      "loss": 1.1104,
+      "step": 14720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1637442981537173e-05,
+      "loss": 1.1851,
+      "step": 14725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1595280243569817e-05,
+      "loss": 1.149,
+      "step": 14730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.155318932141426e-05,
+      "loss": 1.1588,
+      "step": 14735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1511170249263138e-05,
+      "loss": 1.0147,
+      "step": 14740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1469223061250799e-05,
+      "loss": 1.1911,
+      "step": 14745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.142734779145308e-05,
+      "loss": 1.1319,
+      "step": 14750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1385544473887543e-05,
+      "loss": 1.076,
+      "step": 14755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.134381314251317e-05,
+      "loss": 1.0629,
+      "step": 14760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1302153831230565e-05,
+      "loss": 1.0782,
+      "step": 14765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.126056657388177e-05,
+      "loss": 1.0539,
+      "step": 14770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.121905140425029e-05,
+      "loss": 1.1337,
+      "step": 14775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1177608356061075e-05,
+      "loss": 1.129,
+      "step": 14780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1136237462980521e-05,
+      "loss": 1.0913,
+      "step": 14785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1094938758616386e-05,
+      "loss": 1.0456,
+      "step": 14790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1053712276517785e-05,
+      "loss": 1.1827,
+      "step": 14795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.101255805017512e-05,
+      "loss": 1.1398,
+      "step": 14800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0971476113020192e-05,
+      "loss": 1.1424,
+      "step": 14805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0930466498426e-05,
+      "loss": 1.1412,
+      "step": 14810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0889529239706811e-05,
+      "loss": 1.177,
+      "step": 14815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0848664370118102e-05,
+      "loss": 1.1508,
+      "step": 14820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0807871922856573e-05,
+      "loss": 1.151,
+      "step": 14825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0767151931060129e-05,
+      "loss": 1.1165,
+      "step": 14830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0726504427807686e-05,
+      "loss": 1.047,
+      "step": 14835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0685929446119395e-05,
+      "loss": 1.1675,
+      "step": 14840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0645427018956423e-05,
+      "loss": 1.1338,
+      "step": 14845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0604997179221066e-05,
+      "loss": 1.1181,
+      "step": 14850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0564639959756595e-05,
+      "loss": 1.1002,
+      "step": 14855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0524355393347297e-05,
+      "loss": 1.132,
+      "step": 14860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0484143512718438e-05,
+      "loss": 1.1145,
+      "step": 14865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0444004350536286e-05,
+      "loss": 1.0696,
+      "step": 14870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0403937939407958e-05,
+      "loss": 1.0861,
+      "step": 14875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0363944311881546e-05,
+      "loss": 1.0521,
+      "step": 14880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0324023500445933e-05,
+      "loss": 1.188,
+      "step": 14885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0284175537530948e-05,
+      "loss": 1.0528,
+      "step": 14890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0244400455507161e-05,
+      "loss": 1.1431,
+      "step": 14895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0204698286685984e-05,
+      "loss": 1.0853,
+      "step": 14900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.016506906331954e-05,
+      "loss": 1.1475,
+      "step": 14905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0125512817600798e-05,
+      "loss": 1.177,
+      "step": 14910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0086029581663333e-05,
+      "loss": 1.1124,
+      "step": 14915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0046619387581497e-05,
+      "loss": 1.1108,
+      "step": 14920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0007282267370218e-05,
+      "loss": 1.0483,
+      "step": 14925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.96801825298513e-06,
+      "loss": 1.1628,
+      "step": 14930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.9288273763225e-06,
+      "loss": 1.1352,
+      "step": 14935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.88970966921909e-06,
+      "loss": 1.0802,
+      "step": 14940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.850665163452308e-06,
+      "loss": 1.119,
+      "step": 14945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.811693890740034e-06,
+      "loss": 1.156,
+      "step": 14950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.772795882740712e-06,
+      "loss": 1.163,
+      "step": 14955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.733971171053233e-06,
+      "loss": 1.1857,
+      "step": 14960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.695219787216966e-06,
+      "loss": 1.0886,
+      "step": 14965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.656541762711666e-06,
+      "loss": 1.1179,
+      "step": 14970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.617937128957576e-06,
+      "loss": 1.0918,
+      "step": 14975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.57940591731523e-06,
+      "loss": 1.1592,
+      "step": 14980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.540948159085627e-06,
+      "loss": 1.1356,
+      "step": 14985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.502563885509961e-06,
+      "loss": 1.1696,
+      "step": 14990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.464253127769838e-06,
+      "loss": 1.1137,
+      "step": 14995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.426015916987071e-06,
+      "loss": 1.1584,
+      "step": 15000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.387852284223826e-06,
+      "loss": 1.1233,
+      "step": 15005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.34976226048236e-06,
+      "loss": 1.0998,
+      "step": 15010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.311745876705242e-06,
+      "loss": 1.1126,
+      "step": 15015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.27380316377518e-06,
+      "loss": 1.0879,
+      "step": 15020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.235934152515064e-06,
+      "loss": 1.1092,
+      "step": 15025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.198138873687829e-06,
+      "loss": 1.1679,
+      "step": 15030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.160417357996594e-06,
+      "loss": 1.1006,
+      "step": 15035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.122769636084569e-06,
+      "loss": 1.0688,
+      "step": 15040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.085195738534952e-06,
+      "loss": 1.1219,
+      "step": 15045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.047695695871012e-06,
+      "loss": 1.1509,
+      "step": 15050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.010269538555977e-06,
+      "loss": 1.1096,
+      "step": 15055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.972917296993138e-06,
+      "loss": 1.0901,
+      "step": 15060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.935639001525636e-06,
+      "loss": 1.0876,
+      "step": 15065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.898434682436659e-06,
+      "loss": 1.1571,
+      "step": 15070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.861304369949158e-06,
+      "loss": 1.1578,
+      "step": 15075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.8242480942261e-06,
+      "loss": 1.0897,
+      "step": 15080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.787265885370216e-06,
+      "loss": 1.1255,
+      "step": 15085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.750357773424152e-06,
+      "loss": 1.1242,
+      "step": 15090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.713523788370244e-06,
+      "loss": 1.0924,
+      "step": 15095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.676763960130736e-06,
+      "loss": 1.0899,
+      "step": 15100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.64007831856752e-06,
+      "loss": 1.1433,
+      "step": 15105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.603466893482332e-06,
+      "loss": 1.1505,
+      "step": 15110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.566929714616527e-06,
+      "loss": 1.0991,
+      "step": 15115
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.530466811651198e-06,
+      "loss": 1.1318,
+      "step": 15120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.494078214207062e-06,
+      "loss": 1.1226,
+      "step": 15125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.457763951844514e-06,
+      "loss": 1.1172,
+      "step": 15130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.421524054063524e-06,
+      "loss": 1.0945,
+      "step": 15135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.385358550303693e-06,
+      "loss": 1.0909,
+      "step": 15140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.349267469944135e-06,
+      "loss": 1.1227,
+      "step": 15145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.313250842303544e-06,
+      "loss": 1.1964,
+      "step": 15150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.277308696640186e-06,
+      "loss": 1.1688,
+      "step": 15155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.241441062151657e-06,
+      "loss": 1.0771,
+      "step": 15160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.205647967975205e-06,
+      "loss": 1.1096,
+      "step": 15165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.169929443187396e-06,
+      "loss": 1.1335,
+      "step": 15170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.13428551680431e-06,
+      "loss": 1.1183,
+      "step": 15175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.09871621778137e-06,
+      "loss": 1.1953,
+      "step": 15180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.063221575013382e-06,
+      "loss": 1.0554,
+      "step": 15185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.027801617334511e-06,
+      "loss": 1.0197,
+      "step": 15190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.992456373518278e-06,
+      "loss": 1.044,
+      "step": 15195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.957185872277487e-06,
+      "loss": 1.069,
+      "step": 15200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.921990142264202e-06,
+      "loss": 1.0493,
+      "step": 15205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.886869212069771e-06,
+      "loss": 1.1269,
+      "step": 15210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.851823110224788e-06,
+      "loss": 1.0462,
+      "step": 15215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.81685186519906e-06,
+      "loss": 1.0593,
+      "step": 15220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.78195550540155e-06,
+      "loss": 1.1268,
+      "step": 15225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.747134059180393e-06,
+      "loss": 1.0949,
+      "step": 15230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.71238755482293e-06,
+      "loss": 1.0858,
+      "step": 15235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.677716020555556e-06,
+      "loss": 1.1019,
+      "step": 15240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.643119484543788e-06,
+      "loss": 1.0684,
+      "step": 15245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.608597974892178e-06,
+      "loss": 1.0584,
+      "step": 15250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.574151519644415e-06,
+      "loss": 1.1031,
+      "step": 15255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.539780146783173e-06,
+      "loss": 1.0976,
+      "step": 15260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.505483884230105e-06,
+      "loss": 1.0825,
+      "step": 15265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.471262759845878e-06,
+      "loss": 1.0622,
+      "step": 15270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.4371168014301e-06,
+      "loss": 1.1313,
+      "step": 15275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.403046036721362e-06,
+      "loss": 1.1014,
+      "step": 15280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.369050493397123e-06,
+      "loss": 1.1541,
+      "step": 15285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.335130199073759e-06,
+      "loss": 1.0662,
+      "step": 15290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.301285181306505e-06,
+      "loss": 1.1495,
+      "step": 15295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.267515467589458e-06,
+      "loss": 1.2118,
+      "step": 15300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.233821085355541e-06,
+      "loss": 1.0682,
+      "step": 15305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.200202061976469e-06,
+      "loss": 1.0419,
+      "step": 15310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.16665842476274e-06,
+      "loss": 1.1222,
+      "step": 15315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.133190200963624e-06,
+      "loss": 1.087,
+      "step": 15320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.0997974177671175e-06,
+      "loss": 1.1532,
+      "step": 15325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.066480102299978e-06,
+      "loss": 1.2219,
+      "step": 15330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.0332382816275524e-06,
+      "loss": 1.0025,
+      "step": 15335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.000071982753964e-06,
+      "loss": 1.1892,
+      "step": 15340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.966981232621916e-06,
+      "loss": 1.0898,
+      "step": 15345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9339660581128175e-06,
+      "loss": 1.0582,
+      "step": 15350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.901026486046568e-06,
+      "loss": 1.0936,
+      "step": 15355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.868162543181776e-06,
+      "loss": 1.0732,
+      "step": 15360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8353742562155085e-06,
+      "loss": 1.1495,
+      "step": 15365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.802661651783457e-06,
+      "loss": 1.1067,
+      "step": 15370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.770024756459777e-06,
+      "loss": 1.1011,
+      "step": 15375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.737463596757115e-06,
+      "loss": 1.1788,
+      "step": 15380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.70497819912666e-06,
+      "loss": 1.0994,
+      "step": 15385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.672568589957978e-06,
+      "loss": 1.0815,
+      "step": 15390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6402347955791525e-06,
+      "loss": 1.1746,
+      "step": 15395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.607976842256551e-06,
+      "loss": 1.0857,
+      "step": 15400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.575794756195067e-06,
+      "loss": 1.1396,
+      "step": 15405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.543688563537878e-06,
+      "loss": 1.099,
+      "step": 15410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5116582903665645e-06,
+      "loss": 1.0743,
+      "step": 15415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.479703962700958e-06,
+      "loss": 1.081,
+      "step": 15420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.447825606499293e-06,
+      "loss": 1.1219,
+      "step": 15425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.416023247657989e-06,
+      "loss": 1.1589,
+      "step": 15430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.384296912011835e-06,
+      "loss": 1.0749,
+      "step": 15435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.352646625333748e-06,
+      "loss": 1.0635,
+      "step": 15440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.321072413334962e-06,
+      "loss": 1.0475,
+      "step": 15445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.289574301664848e-06,
+      "loss": 1.0372,
+      "step": 15450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.258152315911004e-06,
+      "loss": 1.1169,
+      "step": 15455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.226806481599146e-06,
+      "loss": 1.1027,
+      "step": 15460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.195536824193171e-06,
+      "loss": 1.0746,
+      "step": 15465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1643433690950274e-06,
+      "loss": 1.1372,
+      "step": 15470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1332261416448325e-06,
+      "loss": 1.1275,
+      "step": 15475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1021851671207775e-06,
+      "loss": 1.1642,
+      "step": 15480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0712204707390146e-06,
+      "loss": 1.0729,
+      "step": 15485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.040332077653843e-06,
+      "loss": 1.1197,
+      "step": 15490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.009520012957515e-06,
+      "loss": 1.0777,
+      "step": 15495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.978784301680329e-06,
+      "loss": 1.1061,
+      "step": 15500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9481249687904586e-06,
+      "loss": 1.0993,
+      "step": 15505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.917542039194157e-06,
+      "loss": 1.0466,
+      "step": 15510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.887035537735497e-06,
+      "loss": 1.1489,
+      "step": 15515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.856605489196563e-06,
+      "loss": 1.0831,
+      "step": 15520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8262519182972746e-06,
+      "loss": 1.0411,
+      "step": 15525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.795974849695416e-06,
+      "loss": 1.0776,
+      "step": 15530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.765774307986659e-06,
+      "loss": 1.1137,
+      "step": 15535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7356503177044986e-06,
+      "loss": 1.1409,
+      "step": 15540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.705602903320238e-06,
+      "loss": 1.057,
+      "step": 15545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.675632089242975e-06,
+      "loss": 1.1074,
+      "step": 15550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.645737899819556e-06,
+      "loss": 1.094,
+      "step": 15555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.615920359334656e-06,
+      "loss": 1.1088,
+      "step": 15560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5861794920105995e-06,
+      "loss": 1.1397,
+      "step": 15565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5565153220074585e-06,
+      "loss": 1.1354,
+      "step": 15570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5269278734230025e-06,
+      "loss": 1.1178,
+      "step": 15575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.497417170292683e-06,
+      "loss": 1.1433,
+      "step": 15580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.467983236589624e-06,
+      "loss": 1.0541,
+      "step": 15585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4386260962245185e-06,
+      "loss": 1.1599,
+      "step": 15590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4093457730457355e-06,
+      "loss": 1.1396,
+      "step": 15595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.380142290839215e-06,
+      "loss": 1.0739,
+      "step": 15600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3510156733285045e-06,
+      "loss": 1.1661,
+      "step": 15605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.321965944174686e-06,
+      "loss": 1.151,
+      "step": 15610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.292993126976373e-06,
+      "loss": 1.0758,
+      "step": 15615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.264097245269703e-06,
+      "loss": 1.0851,
+      "step": 15620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.235278322528345e-06,
+      "loss": 1.1199,
+      "step": 15625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.206536382163425e-06,
+      "loss": 1.0436,
+      "step": 15630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.177871447523525e-06,
+      "loss": 1.1559,
+      "step": 15635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.149283541894672e-06,
+      "loss": 1.1605,
+      "step": 15640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.120772688500369e-06,
+      "loss": 1.1924,
+      "step": 15645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.092338910501429e-06,
+      "loss": 1.1025,
+      "step": 15650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.06398223099617e-06,
+      "loss": 1.1805,
+      "step": 15655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0357026730201625e-06,
+      "loss": 1.1369,
+      "step": 15660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0075002595464075e-06,
+      "loss": 1.0946,
+      "step": 15665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.979375013485201e-06,
+      "loss": 1.1092,
+      "step": 15670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.951326957684189e-06,
+      "loss": 1.1243,
+      "step": 15675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.923356114928245e-06,
+      "loss": 1.1355,
+      "step": 15680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8954625079395844e-06,
+      "loss": 1.1517,
+      "step": 15685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.867646159377636e-06,
+      "loss": 1.121,
+      "step": 15690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.839907091839112e-06,
+      "loss": 1.1251,
+      "step": 15695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.812245327857899e-06,
+      "loss": 1.16,
+      "step": 15700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.784660889905099e-06,
+      "loss": 1.1407,
+      "step": 15705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.757153800389022e-06,
+      "loss": 1.1172,
+      "step": 15710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.729724081655126e-06,
+      "loss": 1.1653,
+      "step": 15715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.702371755986008e-06,
+      "loss": 1.1363,
+      "step": 15720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.675096845601386e-06,
+      "loss": 1.1129,
+      "step": 15725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.647899372658144e-06,
+      "loss": 1.1746,
+      "step": 15730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.620779359250189e-06,
+      "loss": 1.0906,
+      "step": 15735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.593736827408574e-06,
+      "loss": 1.0549,
+      "step": 15740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.566771799101328e-06,
+      "loss": 1.1265,
+      "step": 15745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.539884296233606e-06,
+      "loss": 1.146,
+      "step": 15750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.513074340647501e-06,
+      "loss": 1.1519,
+      "step": 15755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.486341954122204e-06,
+      "loss": 1.0767,
+      "step": 15760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.459687158373789e-06,
+      "loss": 1.0886,
+      "step": 15765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.433109975055379e-06,
+      "loss": 1.1667,
+      "step": 15770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.406610425757003e-06,
+      "loss": 1.0441,
+      "step": 15775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3801885320056755e-06,
+      "loss": 1.0611,
+      "step": 15780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.35384431526526e-06,
+      "loss": 1.0406,
+      "step": 15785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.327577796936572e-06,
+      "loss": 1.0848,
+      "step": 15790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.301388998357265e-06,
+      "loss": 1.0957,
+      "step": 15795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2752779408018865e-06,
+      "loss": 1.1554,
+      "step": 15800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.24924464548186e-06,
+      "loss": 1.0796,
+      "step": 15805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2232891335453555e-06,
+      "loss": 1.1156,
+      "step": 15810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.197411426077429e-06,
+      "loss": 1.0907,
+      "step": 15815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.171611544099889e-06,
+      "loss": 1.0705,
+      "step": 15820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1458895085713695e-06,
+      "loss": 1.0766,
+      "step": 15825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.120245340387189e-06,
+      "loss": 1.0711,
+      "step": 15830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.094679060379506e-06,
+      "loss": 1.1613,
+      "step": 15835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.069190689317107e-06,
+      "loss": 1.0675,
+      "step": 15840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.043780247905584e-06,
+      "loss": 1.0787,
+      "step": 15845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.01844775678715e-06,
+      "loss": 1.1046,
+      "step": 15850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.99319323654074e-06,
+      "loss": 1.1347,
+      "step": 15855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.968016707681899e-06,
+      "loss": 1.0645,
+      "step": 15860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.942918190662893e-06,
+      "loss": 1.0998,
+      "step": 15865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.917897705872543e-06,
+      "loss": 1.0556,
+      "step": 15870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.892955273636313e-06,
+      "loss": 1.1649,
+      "step": 15875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.868090914216239e-06,
+      "loss": 1.0941,
+      "step": 15880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.843304647810975e-06,
+      "loss": 1.0874,
+      "step": 15885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.818596494555704e-06,
+      "loss": 1.0992,
+      "step": 15890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.793966474522159e-06,
+      "loss": 1.1281,
+      "step": 15895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7694146077185796e-06,
+      "loss": 1.033,
+      "step": 15900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7449409140897778e-06,
+      "loss": 1.1218,
+      "step": 15905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7205454135170163e-06,
+      "loss": 1.14,
+      "step": 15910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.696228125818022e-06,
+      "loss": 1.1457,
+      "step": 15915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6719890707470366e-06,
+      "loss": 1.0548,
+      "step": 15920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6478282679947095e-06,
+      "loss": 1.1437,
+      "step": 15925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6237457371881293e-06,
+      "loss": 1.0699,
+      "step": 15930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.599741497890825e-06,
+      "loss": 1.1399,
+      "step": 15935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.575815569602676e-06,
+      "loss": 1.0962,
+      "step": 15940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5519679717599906e-06,
+      "loss": 1.0773,
+      "step": 15945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5281987237354267e-06,
+      "loss": 1.1114,
+      "step": 15950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5045078448380067e-06,
+      "loss": 1.1133,
+      "step": 15955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.4808953543130573e-06,
+      "loss": 1.0785,
+      "step": 15960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.457361271342257e-06,
+      "loss": 1.0664,
+      "step": 15965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.4339056150436027e-06,
+      "loss": 1.0698,
+      "step": 15970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4105284044713407e-06,
+      "loss": 1.2109,
+      "step": 15975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.3872296586160133e-06,
+      "loss": 1.0391,
+      "step": 15980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.3640093964044033e-06,
+      "loss": 1.1354,
+      "step": 15985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.340867636699577e-06,
+      "loss": 1.0609,
+      "step": 15990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.3178043983007974e-06,
+      "loss": 1.1171,
+      "step": 15995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.2948196999435657e-06,
+      "loss": 1.0477,
+      "step": 16000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.2719135602995126e-06,
+      "loss": 1.0792,
+      "step": 16005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.249085997976555e-06,
+      "loss": 1.0743,
+      "step": 16010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.226337031518702e-06,
+      "loss": 1.0793,
+      "step": 16015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.2036666794061378e-06,
+      "loss": 1.1765,
+      "step": 16020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.1810749600551983e-06,
+      "loss": 1.1106,
+      "step": 16025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.1585618918183025e-06,
+      "loss": 1.1193,
+      "step": 16030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.136127492984021e-06,
+      "loss": 1.139,
+      "step": 16035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.113771781776987e-06,
+      "loss": 1.1757,
+      "step": 16040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.09149477635794e-06,
+      "loss": 1.1187,
+      "step": 16045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0692964948236377e-06,
+      "loss": 1.1375,
+      "step": 16050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0471769552069324e-06,
+      "loss": 1.0921,
+      "step": 16055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.025136175476684e-06,
+      "loss": 1.0905,
+      "step": 16060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.003174173537804e-06,
+      "loss": 1.1505,
+      "step": 16065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9812909672311435e-06,
+      "loss": 1.1504,
+      "step": 16070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9594865743336145e-06,
+      "loss": 1.1149,
+      "step": 16075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9377610125580487e-06,
+      "loss": 1.1724,
+      "step": 16080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9161142995533067e-06,
+      "loss": 1.0924,
+      "step": 16085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8945464529041e-06,
+      "loss": 1.1057,
+      "step": 16090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8730574901311692e-06,
+      "loss": 1.0862,
+      "step": 16095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8516474286910942e-06,
+      "loss": 1.0754,
+      "step": 16100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.830316285976452e-06,
+      "loss": 1.1125,
+      "step": 16105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.809064079315591e-06,
+      "loss": 1.0761,
+      "step": 16110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.787890825972836e-06,
+      "loss": 1.1002,
+      "step": 16115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.766796543148306e-06,
+      "loss": 1.1259,
+      "step": 16120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7457812479780493e-06,
+      "loss": 1.1162,
+      "step": 16125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7248449575338536e-06,
+      "loss": 1.112,
+      "step": 16130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7039876888233695e-06,
+      "loss": 1.146,
+      "step": 16135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6832094587900884e-06,
+      "loss": 1.1394,
+      "step": 16140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.662510284313231e-06,
+      "loss": 1.0629,
+      "step": 16145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6418901822078575e-06,
+      "loss": 1.0954,
+      "step": 16150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6213491692247248e-06,
+      "loss": 1.1435,
+      "step": 16155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6008872620504172e-06,
+      "loss": 1.1263,
+      "step": 16160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.580504477307183e-06,
+      "loss": 1.1138,
+      "step": 16165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5602008315530776e-06,
+      "loss": 1.1343,
+      "step": 16170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5399763412817743e-06,
+      "loss": 1.1663,
+      "step": 16175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.51983102292273e-06,
+      "loss": 1.0963,
+      "step": 16180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.499764892841011e-06,
+      "loss": 1.0759,
+      "step": 16185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.479777967337432e-06,
+      "loss": 1.0689,
+      "step": 16190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.459870262648405e-06,
+      "loss": 1.1193,
+      "step": 16195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.440041794946017e-06,
+      "loss": 1.0101,
+      "step": 16200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4202925803379493e-06,
+      "loss": 1.229,
+      "step": 16205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4006226348675694e-06,
+      "loss": 1.0983,
+      "step": 16210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.381031974513784e-06,
+      "loss": 1.2061,
+      "step": 16215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.361520615191129e-06,
+      "loss": 1.0888,
+      "step": 16220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3420885727496924e-06,
+      "loss": 1.0964,
+      "step": 16225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3227358629751696e-06,
+      "loss": 1.0691,
+      "step": 16230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3034625015887848e-06,
+      "loss": 1.0945,
+      "step": 16235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2842685042472802e-06,
+      "loss": 1.117,
+      "step": 16240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.265153886542983e-06,
+      "loss": 1.121,
+      "step": 16245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.246118664003673e-06,
+      "loss": 1.096,
+      "step": 16250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2271628520926916e-06,
+      "loss": 1.0896,
+      "step": 16255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2082864662088325e-06,
+      "loss": 1.1378,
+      "step": 16260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.189489521686384e-06,
+      "loss": 1.1348,
+      "step": 16265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1707720337950773e-06,
+      "loss": 1.1282,
+      "step": 16270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.152134017740139e-06,
+      "loss": 1.1035,
+      "step": 16275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1335754886621886e-06,
+      "loss": 1.0911,
+      "step": 16280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.115096461637313e-06,
+      "loss": 1.1321,
+      "step": 16285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0966969516769816e-06,
+      "loss": 1.1175,
+      "step": 16290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0783769737281155e-06,
+      "loss": 1.0464,
+      "step": 16295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0601365426729763e-06,
+      "loss": 1.0491,
+      "step": 16300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0419756733292233e-06,
+      "loss": 1.1104,
+      "step": 16305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.0238943804498886e-06,
+      "loss": 1.1716,
+      "step": 16310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.005892678723377e-06,
+      "loss": 1.1465,
+      "step": 16315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.987970582773414e-06,
+      "loss": 1.1245,
+      "step": 16320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9701281071590427e-06,
+      "loss": 1.1121,
+      "step": 16325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9523652663746682e-06,
+      "loss": 1.0917,
+      "step": 16330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9346820748499707e-06,
+      "loss": 1.1602,
+      "step": 16335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.917078546949936e-06,
+      "loss": 1.104,
+      "step": 16340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.899554696974849e-06,
+      "loss": 1.1388,
+      "step": 16345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8821105391602541e-06,
+      "loss": 1.1104,
+      "step": 16350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8647460876769391e-06,
+      "loss": 1.1324,
+      "step": 16355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8474613566309977e-06,
+      "loss": 1.121,
+      "step": 16360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8302563600637091e-06,
+      "loss": 1.0966,
+      "step": 16365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8131311119515934e-06,
+      "loss": 1.1196,
+      "step": 16370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.796085626206389e-06,
+      "loss": 1.1594,
+      "step": 16375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.779119916675053e-06,
+      "loss": 1.1135,
+      "step": 16380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7622339971397384e-06,
+      "loss": 1.085,
+      "step": 16385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7454278813177395e-06,
+      "loss": 1.0639,
+      "step": 16390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7287015828615571e-06,
+      "loss": 1.1301,
+      "step": 16395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7120551153588557e-06,
+      "loss": 1.1418,
+      "step": 16400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6954884923324298e-06,
+      "loss": 1.1234,
+      "step": 16405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6790017272402356e-06,
+      "loss": 1.1029,
+      "step": 16410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6625948334753047e-06,
+      "loss": 1.0818,
+      "step": 16415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6462678243658534e-06,
+      "loss": 1.1578,
+      "step": 16420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6300207131751606e-06,
+      "loss": 1.1124,
+      "step": 16425
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6138535131016246e-06,
+      "loss": 1.1645,
+      "step": 16430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5977662372786842e-06,
+      "loss": 1.1968,
+      "step": 16435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5817588987749081e-06,
+      "loss": 1.1123,
+      "step": 16440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5658315105938826e-06,
+      "loss": 1.1023,
+      "step": 16445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5499840856742805e-06,
+      "loss": 1.1495,
+      "step": 16450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.534216636889793e-06,
+      "loss": 1.0852,
+      "step": 16455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5185291770491406e-06,
+      "loss": 1.1549,
+      "step": 16460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5029217188960855e-06,
+      "loss": 1.1739,
+      "step": 16465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.487394275109385e-06,
+      "loss": 1.0818,
+      "step": 16470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4719468583028047e-06,
+      "loss": 1.0607,
+      "step": 16475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4565794810250846e-06,
+      "loss": 1.1059,
+      "step": 16480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4412921557599614e-06,
+      "loss": 1.0518,
+      "step": 16485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4260848949261344e-06,
+      "loss": 1.0182,
+      "step": 16490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4109577108772653e-06,
+      "loss": 1.1413,
+      "step": 16495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3959106159019473e-06,
+      "loss": 1.1225,
+      "step": 16500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3809436222237576e-06,
+      "loss": 1.1613,
+      "step": 16505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3660567420011375e-06,
+      "loss": 1.1519,
+      "step": 16510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3512499873275244e-06,
+      "loss": 1.1568,
+      "step": 16515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3365233702311965e-06,
+      "loss": 1.1024,
+      "step": 16520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3218769026753619e-06,
+      "loss": 1.1473,
+      "step": 16525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3073105965581255e-06,
+      "loss": 1.1397,
+      "step": 16530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2928244637124661e-06,
+      "loss": 1.14,
+      "step": 16535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2784185159062376e-06,
+      "loss": 1.1294,
+      "step": 16540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.264092764842145e-06,
+      "loss": 1.1437,
+      "step": 16545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.249847222157735e-06,
+      "loss": 1.0628,
+      "step": 16550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2356818994254386e-06,
+      "loss": 1.1149,
+      "step": 16555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2215968081524741e-06,
+      "loss": 1.1076,
+      "step": 16560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2075919597809113e-06,
+      "loss": 1.0568,
+      "step": 16565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.193667365687638e-06,
+      "loss": 1.085,
+      "step": 16570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1798230371843058e-06,
+      "loss": 1.1119,
+      "step": 16575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1660589855174297e-06,
+      "loss": 1.1855,
+      "step": 16580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1523752218682435e-06,
+      "loss": 1.0604,
+      "step": 16585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1387717573528108e-06,
+      "loss": 1.1328,
+      "step": 16590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1252486030219139e-06,
+      "loss": 1.0743,
+      "step": 16595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1118057698611544e-06,
+      "loss": 1.1038,
+      "step": 16600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0984432687908412e-06,
+      "loss": 1.1044,
+      "step": 16605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0851611106660467e-06,
+      "loss": 1.0952,
+      "step": 16610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0719593062765512e-06,
+      "loss": 1.1749,
+      "step": 16615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.058837866346887e-06,
+      "loss": 1.1268,
+      "step": 16620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.045796801536283e-06,
+      "loss": 1.0945,
+      "step": 16625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.032836122438698e-06,
+      "loss": 1.0407,
+      "step": 16630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.019955839582744e-06,
+      "loss": 1.1344,
+      "step": 16635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0071559634317957e-06,
+      "loss": 1.1925,
+      "step": 16640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.944365043838461e-07,
+      "loss": 1.1355,
+      "step": 16645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.817974727715744e-07,
+      "loss": 1.1649,
+      "step": 16650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.692388788623341e-07,
+      "loss": 1.1269,
+      "step": 16655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.567607328581428e-07,
+      "loss": 1.0955,
+      "step": 16660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.443630448956471e-07,
+      "loss": 1.1368,
+      "step": 16665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.320458250461577e-07,
+      "loss": 1.1313,
+      "step": 16670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.198090833155817e-07,
+      "loss": 1.0815,
+      "step": 16675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.076528296444788e-07,
+      "loss": 1.1637,
+      "step": 16680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.955770739080159e-07,
+      "loss": 1.1099,
+      "step": 16685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.835818259159576e-07,
+      "loss": 1.0788,
+      "step": 16690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.716670954126869e-07,
+      "loss": 1.1443,
+      "step": 16695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.5983289207715e-07,
+      "loss": 1.0945,
+      "step": 16700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.480792255229126e-07,
+      "loss": 1.0568,
+      "step": 16705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.364061052980813e-07,
+      "loss": 1.1136,
+      "step": 16710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.248135408853598e-07,
+      "loss": 1.1255,
+      "step": 16715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.133015417019929e-07,
+      "loss": 1.1145,
+      "step": 16720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.018701170997888e-07,
+      "loss": 1.0732,
+      "step": 16725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.905192763650856e-07,
+      "loss": 1.0835,
+      "step": 16730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.792490287187856e-07,
+      "loss": 1.0513,
+      "step": 16735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.680593833162869e-07,
+      "loss": 1.1288,
+      "step": 16740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.56950349247541e-07,
+      "loss": 1.1648,
+      "step": 16745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.459219355369955e-07,
+      "loss": 1.1551,
+      "step": 16750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.349741511436173e-07,
+      "loss": 1.1016,
+      "step": 16755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.241070049608591e-07,
+      "loss": 1.0629,
+      "step": 16760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.133205058166926e-07,
+      "loss": 1.1038,
+      "step": 16765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.026146624735308e-07,
+      "loss": 1.1666,
+      "step": 16770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.919894836283277e-07,
+      "loss": 1.0821,
+      "step": 16775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.814449779124462e-07,
+      "loss": 1.2119,
+      "step": 16780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.70981153891767e-07,
+      "loss": 1.1009,
+      "step": 16785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.605980200665911e-07,
+      "loss": 1.1142,
+      "step": 16790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.502955848716933e-07,
+      "loss": 1.0964,
+      "step": 16795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.4007385667629e-07,
+      "loss": 1.0647,
+      "step": 16800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.299328437840058e-07,
+      "loss": 1.1979,
+      "step": 16805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.198725544329508e-07,
+      "loss": 1.0791,
+      "step": 16810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.09892996795599e-07,
+      "loss": 1.0483,
+      "step": 16815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.999941789788985e-07,
+      "loss": 1.103,
+      "step": 16820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.901761090241506e-07,
+      "loss": 1.1139,
+      "step": 16825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.804387949071194e-07,
+      "loss": 1.1437,
+      "step": 16830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.707822445379223e-07,
+      "loss": 1.0156,
+      "step": 16835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.612064657610954e-07,
+      "loss": 1.0878,
+      "step": 16840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.517114663555271e-07,
+      "loss": 1.0785,
+      "step": 16845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.422972540345254e-07,
+      "loss": 1.0821,
+      "step": 16850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.329638364457391e-07,
+      "loss": 1.1105,
+      "step": 16855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.237112211712036e-07,
+      "loss": 1.1316,
+      "step": 16860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.145394157273064e-07,
+      "loss": 1.0996,
+      "step": 16865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.054484275647874e-07,
+      "loss": 1.0521,
+      "step": 16870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.964382640687169e-07,
+      "loss": 1.1272,
+      "step": 16875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.875089325585622e-07,
+      "loss": 1.1102,
+      "step": 16880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.786604402880541e-07,
+      "loss": 1.1545,
+      "step": 16885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.698927944452985e-07,
+      "loss": 1.1228,
+      "step": 16890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.612060021527314e-07,
+      "loss": 1.1967,
+      "step": 16895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.526000704670855e-07,
+      "loss": 1.1296,
+      "step": 16900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.440750063794022e-07,
+      "loss": 1.1456,
+      "step": 16905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.356308168150415e-07,
+      "loss": 1.0568,
+      "step": 16910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.272675086336831e-07,
+      "loss": 1.1082,
+      "step": 16915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.1898508862925897e-07,
+      "loss": 1.1298,
+      "step": 16920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.107835635300205e-07,
+      "loss": 1.0705,
+      "step": 16925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.026629399985158e-07,
+      "loss": 1.1249,
+      "step": 16930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.946232246315348e-07,
+      "loss": 1.0854,
+      "step": 16935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8666442396016426e-07,
+      "loss": 1.1833,
+      "step": 16940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.787865444497658e-07,
+      "loss": 1.1123,
+      "step": 16945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.709895924999535e-07,
+      "loss": 1.0506,
+      "step": 16950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6327357444460526e-07,
+      "loss": 1.0753,
+      "step": 16955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5563849655184035e-07,
+      "loss": 1.0932,
+      "step": 16960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4808436502404195e-07,
+      "loss": 1.1486,
+      "step": 16965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4061118599783455e-07,
+      "loss": 1.0935,
+      "step": 16970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.33218965544051e-07,
+      "loss": 1.0516,
+      "step": 16975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.259077096678097e-07,
+      "loss": 1.1001,
+      "step": 16980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1867742430841546e-07,
+      "loss": 1.0619,
+      "step": 16985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.115281153394143e-07,
+      "loss": 1.0998,
+      "step": 16990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0445978856856037e-07,
+      "loss": 1.0915,
+      "step": 16995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9747244973782734e-07,
+      "loss": 1.1161,
+      "step": 17000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9056610452340806e-07,
+      "loss": 1.068,
+      "step": 17005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8374075853568145e-07,
+      "loss": 1.0757,
+      "step": 17010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7699641731924553e-07,
+      "loss": 1.0625,
+      "step": 17015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.703330863528844e-07,
+      "loss": 1.0942,
+      "step": 17020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6375077104955705e-07,
+      "loss": 1.1192,
+      "step": 17025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5724947675643055e-07,
+      "loss": 1.1503,
+      "step": 17030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.508292087548693e-07,
+      "loss": 1.0777,
+      "step": 17035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4448997226036786e-07,
+      "loss": 1.1206,
+      "step": 17040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.382317724226291e-07,
+      "loss": 1.1276,
+      "step": 17045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3205461432553085e-07,
+      "loss": 0.985,
+      "step": 17050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.259585029870924e-07,
+      "loss": 1.1301,
+      "step": 17055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1994344335950796e-07,
+      "loss": 1.1161,
+      "step": 17060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.140094403291356e-07,
+      "loss": 1.0812,
+      "step": 17065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.081564987164637e-07,
+      "loss": 1.0704,
+      "step": 17070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0238462327615564e-07,
+      "loss": 1.0955,
+      "step": 17075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.966938186970163e-07,
+      "loss": 1.0662,
+      "step": 17080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.910840896019699e-07,
+      "loss": 1.1334,
+      "step": 17085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8555544054810452e-07,
+      "loss": 1.1211,
+      "step": 17090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8010787602662758e-07,
+      "loss": 1.0898,
+      "step": 17095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7474140046289915e-07,
+      "loss": 1.1766,
+      "step": 17100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6945601821636532e-07,
+      "loss": 1.0579,
+      "step": 17105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6425173358063605e-07,
+      "loss": 1.1526,
+      "step": 17110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.591285507834184e-07,
+      "loss": 1.1208,
+      "step": 17115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5408647398654996e-07,
+      "loss": 1.1807,
+      "step": 17120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4912550728597645e-07,
+      "loss": 1.1635,
+      "step": 17125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4424565471175196e-07,
+      "loss": 1.0432,
+      "step": 17130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3944692022802775e-07,
+      "loss": 1.1401,
+      "step": 17135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.347293077330855e-07,
+      "loss": 1.1203,
+      "step": 17140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3009282105927082e-07,
+      "loss": 1.0678,
+      "step": 17145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2553746397305962e-07,
+      "loss": 1.1245,
+      "step": 17150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2106324017500294e-07,
+      "loss": 1.146,
+      "step": 17155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1667015329976e-07,
+      "loss": 1.1014,
+      "step": 17160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1235820691605392e-07,
+      "loss": 1.1257,
+      "step": 17165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0812740452670492e-07,
+      "loss": 1.0855,
+      "step": 17170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0397774956861934e-07,
+      "loss": 1.1447,
+      "step": 17175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.99092454127896e-08,
+      "loss": 1.0901,
+      "step": 17180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.592189536426074e-08,
+      "loss": 1.0761,
+      "step": 17185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.201570266218619e-08,
+      "loss": 1.0526,
+      "step": 17190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.819067047974989e-08,
+      "loss": 1.1825,
+      "step": 17195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.444680192424406e-08,
+      "loss": 1.1301,
+      "step": 17200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.078410003700251e-08,
+      "loss": 1.1353,
+      "step": 17205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.720256779343409e-08,
+      "loss": 1.0628,
+      "step": 17210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.370220810301142e-08,
+      "loss": 1.1315,
+      "step": 17215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.028302380924884e-08,
+      "loss": 1.106,
+      "step": 17220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.694501768973561e-08,
+      "loss": 1.1035,
+      "step": 17225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.368819245611376e-08,
+      "loss": 1.1582,
+      "step": 17230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.051255075406694e-08,
+      "loss": 1.166,
+      "step": 17235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.7418095163342706e-08,
+      "loss": 1.0483,
+      "step": 17240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4404828197719124e-08,
+      "loss": 1.1408,
+      "step": 17245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.1472752305049216e-08,
+      "loss": 1.1449,
+      "step": 17250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8621869867205465e-08,
+      "loss": 1.1723,
+      "step": 17255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5852183200102e-08,
+      "loss": 1.0561,
+      "step": 17260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.316369455371681e-08,
+      "loss": 1.0003,
+      "step": 17265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.05564061120362e-08,
+      "loss": 1.1429,
+      "step": 17270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.803031999309926e-08,
+      "loss": 1.1214,
+      "step": 17275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5585438248986716e-08,
+      "loss": 1.1354,
+      "step": 17280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.322176286579875e-08,
+      "loss": 1.0623,
+      "step": 17285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0939295763677164e-08,
+      "loss": 1.144,
+      "step": 17290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.873803879678325e-08,
+      "loss": 1.1006,
+      "step": 17295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.661799375331992e-08,
+      "loss": 1.2163,
+      "step": 17300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4579162355520632e-08,
+      "loss": 1.2416,
+      "step": 17305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.26215462596161e-08,
+      "loss": 1.1587,
+      "step": 17310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0745147055900888e-08,
+      "loss": 1.078,
+      "step": 17315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8949966268655696e-08,
+      "loss": 1.091,
+      "step": 17320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.723600535620288e-08,
+      "loss": 1.1071,
+      "step": 17325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5603265710895336e-08,
+      "loss": 1.0908,
+      "step": 17330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4051748659083202e-08,
+      "loss": 1.0811,
+      "step": 17335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2581455461147166e-08,
+      "loss": 1.137,
+      "step": 17340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1192387311487366e-08,
+      "loss": 1.0535,
+      "step": 17345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.88454533851228e-09,
+      "loss": 1.1146,
+      "step": 17350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.657930604660933e-09,
+      "loss": 1.1813,
+      "step": 17355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.51254410635849e-09,
+      "loss": 1.1633,
+      "step": 17360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.448386774082868e-09,
+      "loss": 1.1483,
+      "step": 17365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.465459472298123e-09,
+      "loss": 1.1346,
+      "step": 17370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.563762999476653e-09,
+      "loss": 1.0953,
+      "step": 17375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.743298088121406e-09,
+      "loss": 1.1749,
+      "step": 17380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0040654047547757e-09,
+      "loss": 1.2187,
+      "step": 17385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3460655498630877e-09,
+      "loss": 1.1367,
+      "step": 17390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7692990580076276e-09,
+      "loss": 1.1282,
+      "step": 17395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.273766397702511e-09,
+      "loss": 1.1553,
+      "step": 17400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.594679715035046e-10,
+      "loss": 1.0399,
+      "step": 17405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.264041159636159e-10,
+      "loss": 1.1995,
+      "step": 17410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.745751016552985e-10,
+      "loss": 1.0904,
+      "step": 17415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0398113314824695e-10,
+      "loss": 1.0648,
+      "step": 17420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4622349031601304e-11,
+      "loss": 1.1043,
+      "step": 17425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.498821880640549e-12,
+      "loss": 1.0461,
+      "step": 17430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.961055830119079e-11,
+      "loss": 1.1392,
+      "step": 17435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3395749890742223e-10,
+      "loss": 1.0029,
+      "step": 17440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.695395183107465e-10,
+      "loss": 1.1768,
+      "step": 17445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.863564251309187e-10,
+      "loss": 1.0712,
+      "step": 17450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1844079620182416e-09,
+      "loss": 1.0349,
+      "step": 17455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6636938055869524e-09,
+      "loss": 1.1187,
+      "step": 17460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.224213566515143e-09,
+      "loss": 1.0561,
+      "step": 17465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8659667894337388e-09,
+      "loss": 1.1697,
+      "step": 17470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5889529530486194e-09,
+      "loss": 1.069,
+      "step": 17475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3931714700073956e-09,
+      "loss": 1.1701,
+      "step": 17480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.278621687021535e-09,
+      "loss": 1.0673,
+      "step": 17485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.245302884788639e-09,
+      "loss": 1.159,
+      "step": 17490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.293214278014658e-09,
+      "loss": 1.1931,
+      "step": 17495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.422355015424986e-09,
+      "loss": 1.0897,
+      "step": 17500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.632724179775562e-09,
+      "loss": 1.1737,
+      "step": 17505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0924320787808472e-08,
+      "loss": 1.1856,
+      "step": 17510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2297143790296336e-08,
+      "loss": 1.1193,
+      "step": 17515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3751192072009026e-08,
+      "loss": 1.0823,
+      "step": 17520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5286464451769163e-08,
+      "loss": 1.0698,
+      "step": 17525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6902959682374414e-08,
+      "loss": 1.0934,
+      "step": 17530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.860067645067518e-08,
+      "loss": 1.0701,
+      "step": 17535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0379613377508023e-08,
+      "loss": 1.091,
+      "step": 17540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2239769017773337e-08,
+      "loss": 1.1573,
+      "step": 17545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4181141860346563e-08,
+      "loss": 1.0598,
+      "step": 17550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6203730328155897e-08,
+      "loss": 1.1075,
+      "step": 17555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8307532778160077e-08,
+      "loss": 1.1165,
+      "step": 17560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0492547501303994e-08,
+      "loss": 1.0897,
+      "step": 17565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2758772722607476e-08,
+      "loss": 1.0659,
+      "step": 17570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5106206601076506e-08,
+      "loss": 1.0895,
+      "step": 17575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.753484722978091e-08,
+      "loss": 1.0961,
+      "step": 17580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.004469263580996e-08,
+      "loss": 1.1173,
+      "step": 17585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.26357407802724e-08,
+      "loss": 1.0851,
+      "step": 17590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5307989558329686e-08,
+      "loss": 1.1711,
+      "step": 17595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.806143679917382e-08,
+      "loss": 1.0771,
+      "step": 17600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.089608026603849e-08,
+      "loss": 1.046,
+      "step": 17605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.381191765618794e-08,
+      "loss": 1.1066,
+      "step": 17610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.680894660092802e-08,
+      "loss": 1.0681,
+      "step": 17615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.988716466561736e-08,
+      "loss": 1.0917,
+      "step": 17620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.304656934966734e-08,
+      "loss": 1.1276,
+      "step": 17625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.628715808653093e-08,
+      "loss": 1.1136,
+      "step": 17630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.960892824368069e-08,
+      "loss": 1.1298,
+      "step": 17635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.301187712268617e-08,
+      "loss": 1.1391,
+      "step": 17640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.64960019591587e-08,
+      "loss": 1.1331,
+      "step": 17645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.006129992274011e-08,
+      "loss": 1.0241,
+      "step": 17650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.37077681171694e-08,
+      "loss": 1.1279,
+      "step": 17655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.743540358021604e-08,
+      "loss": 1.0849,
+      "step": 17660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.124420328373574e-08,
+      "loss": 1.096,
+      "step": 17665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.513416413363673e-08,
+      "loss": 1.1706,
+      "step": 17670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.910528296989132e-08,
+      "loss": 1.0976,
+      "step": 17675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0315755656655769e-07,
+      "loss": 1.089,
+      "step": 17680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.07290981631758e-07,
+      "loss": 1.1047,
+      "step": 17685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1150555480770042e-07,
+      "loss": 1.0895,
+      "step": 17690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1580127267064585e-07,
+      "loss": 1.1052,
+      "step": 17695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2017813173098576e-07,
+      "loss": 1.1536,
+      "step": 17700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2463612843313095e-07,
+      "loss": 1.1482,
+      "step": 17705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2917525915564495e-07,
+      "loss": 1.1544,
+      "step": 17710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.337955202111551e-07,
+      "loss": 1.0969,
+      "step": 17715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3849690784635265e-07,
+      "loss": 1.1826,
+      "step": 17720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4327941824207048e-07,
+      "loss": 1.0546,
+      "step": 17725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4814304751321617e-07,
+      "loss": 1.1012,
+      "step": 17730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5308779170882804e-07,
+      "loss": 1.0676,
+      "step": 17735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5811364681199704e-07,
+      "loss": 1.1352,
+      "step": 17740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6322060873997795e-07,
+      "loss": 1.0641,
+      "step": 17745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6840867334411152e-07,
+      "loss": 1.046,
+      "step": 17750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.736778364098691e-07,
+      "loss": 1.0918,
+      "step": 17755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.79028093656819e-07,
+      "loss": 1.1782,
+      "step": 17760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8445944073866025e-07,
+      "loss": 1.0212,
+      "step": 17765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8997187324323326e-07,
+      "loss": 1.0893,
+      "step": 17770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9556538669249784e-07,
+      "loss": 1.1477,
+      "step": 17775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0123997654255545e-07,
+      "loss": 1.1655,
+      "step": 17780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0699563818362692e-07,
+      "loss": 1.0668,
+      "step": 17785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1283236694008562e-07,
+      "loss": 1.1685,
+      "step": 17790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.187501580704466e-07,
+      "loss": 1.1222,
+      "step": 17795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.247490067673886e-07,
+      "loss": 1.1557,
+      "step": 17800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.30828908157732e-07,
+      "loss": 1.1365,
+      "step": 17805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3698985730242768e-07,
+      "loss": 1.138,
+      "step": 17810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.432318491966346e-07,
+      "loss": 1.2195,
+      "step": 17815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4955487876965333e-07,
+      "loss": 1.1511,
+      "step": 17820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.559589408849483e-07,
+      "loss": 1.134,
+      "step": 17825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6244403034015876e-07,
+      "loss": 1.1204,
+      "step": 17830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.690101418671098e-07,
+      "loss": 1.0799,
+      "step": 17835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.756572701318017e-07,
+      "loss": 1.0545,
+      "step": 17840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8238540973442034e-07,
+      "loss": 1.0499,
+      "step": 17845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8919455520934887e-07,
+      "loss": 1.0304,
+      "step": 17850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.960847010251455e-07,
+      "loss": 1.15,
+      "step": 17855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0305584158460966e-07,
+      "loss": 1.1098,
+      "step": 17860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.101079712246935e-07,
+      "loss": 1.1001,
+      "step": 17865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1724108421657956e-07,
+      "loss": 1.0966,
+      "step": 17870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.244551747656698e-07,
+      "loss": 1.0998,
+      "step": 17875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3175023701158535e-07,
+      "loss": 1.1206,
+      "step": 17880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.391262650281667e-07,
+      "loss": 1.1344,
+      "step": 17885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4658325282345135e-07,
+      "loss": 1.0534,
+      "step": 17890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.541211943397626e-07,
+      "loss": 1.1075,
+      "step": 17895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6174008345360996e-07,
+      "loss": 1.168,
+      "step": 17900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.694399139757997e-07,
+      "loss": 1.0723,
+      "step": 17905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7722067965132436e-07,
+      "loss": 1.1394,
+      "step": 17910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.850823741594733e-07,
+      "loss": 1.2,
+      "step": 17915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9302499111375516e-07,
+      "loss": 1.105,
+      "step": 17920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.010485240620088e-07,
+      "loss": 1.1311,
+      "step": 17925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.091529664862482e-07,
+      "loss": 1.1579,
+      "step": 17930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.173383118028395e-07,
+      "loss": 1.1507,
+      "step": 17935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2560455336237975e-07,
+      "loss": 1.0406,
+      "step": 17940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.339516844497736e-07,
+      "loss": 1.0647,
+      "step": 17945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.423796982842009e-07,
+      "loss": 1.1317,
+      "step": 17950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5088858801914935e-07,
+      "loss": 1.1455,
+      "step": 17955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.594783467423813e-07,
+      "loss": 1.0541,
+      "step": 17960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.681489674760008e-07,
+      "loss": 1.1335,
+      "step": 17965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.769004431764089e-07,
+      "loss": 1.0877,
+      "step": 17970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.857327667343145e-07,
+      "loss": 1.0963,
+      "step": 17975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.946459309747462e-07,
+      "loss": 1.071,
+      "step": 17980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.036399286570733e-07,
+      "loss": 1.1313,
+      "step": 17985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.127147524750183e-07,
+      "loss": 1.1129,
+      "step": 17990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.218703950566002e-07,
+      "loss": 1.1666,
+      "step": 17995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.311068489642132e-07,
+      "loss": 1.1177,
+      "step": 18000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.404241066946146e-07,
+      "loss": 1.0978,
+      "step": 18005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.498221606788811e-07,
+      "loss": 1.1234,
+      "step": 18010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.593010032824975e-07,
+      "loss": 1.106,
+      "step": 18015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6886062680529e-07,
+      "loss": 1.0697,
+      "step": 18020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.785010234814815e-07,
+      "loss": 1.0271,
+      "step": 18025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.882221854796587e-07,
+      "loss": 1.156,
+      "step": 18030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.980241049028056e-07,
+      "loss": 1.0948,
+      "step": 18035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.079067737883358e-07,
+      "loss": 1.1019,
+      "step": 18040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.17870184108005e-07,
+      "loss": 1.1376,
+      "step": 18045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.279143277680201e-07,
+      "loss": 1.1264,
+      "step": 18050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.380391966090085e-07,
+      "loss": 1.1145,
+      "step": 18055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.482447824059825e-07,
+      "loss": 1.1042,
+      "step": 18060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.585310768684184e-07,
+      "loss": 1.1158,
+      "step": 18065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.688980716402227e-07,
+      "loss": 1.1082,
+      "step": 18070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.79345758299721e-07,
+      "loss": 1.1302,
+      "step": 18075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.898741283597353e-07,
+      "loss": 1.1616,
+      "step": 18080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.004831732674963e-07,
+      "loss": 1.1661,
+      "step": 18085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.1117288440472e-07,
+      "loss": 1.1061,
+      "step": 18090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.219432530875864e-07,
+      "loss": 1.1504,
+      "step": 18095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.327942705667606e-07,
+      "loss": 1.0058,
+      "step": 18100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.43725928027405e-07,
+      "loss": 1.1764,
+      "step": 18105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.547382165891342e-07,
+      "loss": 1.1138,
+      "step": 18110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.65831127306104e-07,
+      "loss": 1.0429,
+      "step": 18115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.770046511669453e-07,
+      "loss": 1.1505,
+      "step": 18120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.882587790948297e-07,
+      "loss": 1.1248,
+      "step": 18125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.995935019474259e-07,
+      "loss": 1.1732,
+      "step": 18130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.110088105169438e-07,
+      "loss": 1.1216,
+      "step": 18135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.225046955301236e-07,
+      "loss": 1.1376,
+      "step": 18140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.340811476482684e-07,
+      "loss": 1.17,
+      "step": 18145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.457381574671901e-07,
+      "loss": 1.1374,
+      "step": 18150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.574757155173083e-07,
+      "loss": 1.1066,
+      "step": 18155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.692938122635603e-07,
+      "loss": 1.1494,
+      "step": 18160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.811924381055153e-07,
+      "loss": 1.1208,
+      "step": 18165
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.931715833772719e-07,
+      "loss": 1.0973,
+      "step": 18170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.052312383475475e-07,
+      "loss": 1.1403,
+      "step": 18175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.173713932196343e-07,
+      "loss": 1.0775,
+      "step": 18180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.295920381314763e-07,
+      "loss": 1.034,
+      "step": 18185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.41893163155571e-07,
+      "loss": 1.1289,
+      "step": 18190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.542747582990776e-07,
+      "loss": 1.1602,
+      "step": 18195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.667368135037646e-07,
+      "loss": 1.0516,
+      "step": 18200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.792793186460524e-07,
+      "loss": 1.0955,
+      "step": 18205
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.919022635370244e-07,
+      "loss": 1.0446,
+      "step": 18210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0046056379223612e-06,
+      "loss": 1.1256,
+      "step": 18215
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0173894314824724e-06,
+      "loss": 1.1325,
+      "step": 18220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0302536338323771e-06,
+      "loss": 1.1127,
+      "step": 18225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0431982345218455e-06,
+      "loss": 1.1377,
+      "step": 18230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0562232230352669e-06,
+      "loss": 1.1232,
+      "step": 18235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0693285887917827e-06,
+      "loss": 1.1032,
+      "step": 18240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.082514321145176e-06,
+      "loss": 1.0565,
+      "step": 18245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.095780409384015e-06,
+      "loss": 1.1019,
+      "step": 18250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1091268427315424e-06,
+      "loss": 1.1256,
+      "step": 18255
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.1316999197006226,
+      "eval_runtime": 1030.3828,
+      "eval_samples_per_second": 15.689,
+      "eval_steps_per_second": 15.689,
+      "step": 18257
+    },
+    {
+      "epoch": 1.0,
+      "step": 18257,
+      "total_flos": 1.8793185671034962e+18,
+      "train_loss": 1.1159586102780843,
+      "train_runtime": 26065.8571,
+      "train_samples_per_second": 5.603,
+      "train_steps_per_second": 0.7
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 18257,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 1.8793185671034962e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}