diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,60019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5981701973662567,
+  "eval_steps": 500,
+  "global_step": 50000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 3.2591,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 3.2777,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 3.2753,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 3.2606,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 3.1962,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 3.184,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 3.2628,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 3.2691,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 3.1692,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 3.216,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 3.1765,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 3.2538,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 3.2171,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-05,
+      "loss": 3.1625,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 3.1747,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 3.1782,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.5e-05,
+      "loss": 3.1535,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-05,
+      "loss": 3.1459,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-05,
+      "loss": 3.233,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 3.2025,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99999991150237e-05,
+      "loss": 3.1818,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999646009479e-05,
+      "loss": 3.2144,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999203521338e-05,
+      "loss": 3.2037,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999998584037962e-05,
+      "loss": 3.2315,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999997787559375e-05,
+      "loss": 3.1742,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999996814085603e-05,
+      "loss": 3.2204,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999995663616681e-05,
+      "loss": 3.0987,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99999433615265e-05,
+      "loss": 3.0905,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999992831693557e-05,
+      "loss": 3.2231,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999991150239456e-05,
+      "loss": 3.1365,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999989291790406e-05,
+      "loss": 3.1346,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999987256346471e-05,
+      "loss": 3.1362,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999985043907725e-05,
+      "loss": 3.1969,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999982654474244e-05,
+      "loss": 3.1699,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999980088046118e-05,
+      "loss": 3.1428,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999977344623432e-05,
+      "loss": 3.1436,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999974424206286e-05,
+      "loss": 3.1048,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999971326794784e-05,
+      "loss": 3.1611,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999968052389032e-05,
+      "loss": 3.1552,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99996460098915e-05,
+      "loss": 3.1622,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999960972595258e-05,
+      "loss": 3.136,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999957167207486e-05,
+      "loss": 3.2335,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999953184825965e-05,
+      "loss": 3.0733,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99994902545084e-05,
+      "loss": 3.1079,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99994468908226e-05,
+      "loss": 3.145,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99994017572037e-05,
+      "loss": 3.1169,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99993548536534e-05,
+      "loss": 3.1683,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999930618017329e-05,
+      "loss": 3.0774,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99992557367651e-05,
+      "loss": 3.0857,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999920352343064e-05,
+      "loss": 3.1848,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999914954017175e-05,
+      "loss": 3.1784,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999909378699036e-05,
+      "loss": 3.1449,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99990362638884e-05,
+      "loss": 3.0663,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999897697086793e-05,
+      "loss": 3.1573,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999891590793104e-05,
+      "loss": 3.165,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999885307507992e-05,
+      "loss": 3.1118,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999878847231678e-05,
+      "loss": 3.1641,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999872209964389e-05,
+      "loss": 3.0653,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99986539570636e-05,
+      "loss": 3.0659,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999858404457835e-05,
+      "loss": 3.147,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999851236219058e-05,
+      "loss": 3.1191,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999843890990286e-05,
+      "loss": 3.1937,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999836368771777e-05,
+      "loss": 3.0073,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999828669563797e-05,
+      "loss": 3.0586,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999820793366622e-05,
+      "loss": 3.0819,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999812740180525e-05,
+      "loss": 3.102,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999804510005797e-05,
+      "loss": 3.1341,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999796102842725e-05,
+      "loss": 3.1658,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999787518691607e-05,
+      "loss": 3.1209,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999778757552751e-05,
+      "loss": 3.1559,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999769819426464e-05,
+      "loss": 3.0269,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999760704313059e-05,
+      "loss": 3.1028,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999751412212866e-05,
+      "loss": 3.1073,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999741943126209e-05,
+      "loss": 3.1339,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999732297053424e-05,
+      "loss": 3.1981,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999722473994852e-05,
+      "loss": 3.0413,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999712473950843e-05,
+      "loss": 3.1155,
+      "step": 385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99970229692175e-05,
+      "loss": 3.1727,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999691942907932e-05,
+      "loss": 3.0563,
+      "step": 395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999681411909756e-05,
+      "loss": 3.185,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999670703927595e-05,
+      "loss": 3.1042,
+      "step": 405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999659818961828e-05,
+      "loss": 3.0887,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999648757012842e-05,
+      "loss": 3.0982,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999637518081027e-05,
+      "loss": 3.1055,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999626102166779e-05,
+      "loss": 3.069,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999614509270504e-05,
+      "loss": 3.0989,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999602739392615e-05,
+      "loss": 3.0905,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999590792533524e-05,
+      "loss": 3.0837,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999578668693656e-05,
+      "loss": 3.1102,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999566367873441e-05,
+      "loss": 3.0921,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999553890073314e-05,
+      "loss": 3.1312,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999541235293716e-05,
+      "loss": 3.0627,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999528403535094e-05,
+      "loss": 3.2333,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999515394797905e-05,
+      "loss": 3.0666,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999502209082607e-05,
+      "loss": 3.1114,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999488846389668e-05,
+      "loss": 3.1012,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999475306719563e-05,
+      "loss": 3.0765,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999461590072767e-05,
+      "loss": 3.0776,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999447696449766e-05,
+      "loss": 3.0968,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999433625851056e-05,
+      "loss": 3.1251,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999419378277132e-05,
+      "loss": 3.111,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999404953728498e-05,
+      "loss": 3.0854,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999390352205667e-05,
+      "loss": 3.1125,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999375573709153e-05,
+      "loss": 3.0592,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999360618239481e-05,
+      "loss": 3.0244,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99934548579718e-05,
+      "loss": 3.1212,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999330176382785e-05,
+      "loss": 3.1276,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999314689996839e-05,
+      "loss": 3.107,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99929902663989e-05,
+      "loss": 3.0883,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999283186312492e-05,
+      "loss": 3.166,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999267169015205e-05,
+      "loss": 3.147,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999250974748599e-05,
+      "loss": 3.0671,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999234603513243e-05,
+      "loss": 3.1182,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99921805530972e-05,
+      "loss": 3.1374,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999201330138614e-05,
+      "loss": 3.1388,
+      "step": 575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999184428000519e-05,
+      "loss": 3.0617,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99916734889603e-05,
+      "loss": 3.1643,
+      "step": 585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999150092825755e-05,
+      "loss": 3.0605,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999132659790302e-05,
+      "loss": 3.1253,
+      "step": 595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99911504979029e-05,
+      "loss": 3.1266,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999097262826342e-05,
+      "loss": 3.0389,
+      "step": 605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999079298899087e-05,
+      "loss": 3.0961,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999061158009162e-05,
+      "loss": 3.0947,
+      "step": 615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999042840157208e-05,
+      "loss": 3.1105,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999024345343875e-05,
+      "loss": 3.0787,
+      "step": 625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999005673569814e-05,
+      "loss": 3.1412,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998986824835693e-05,
+      "loss": 3.0545,
+      "step": 635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998967799142172e-05,
+      "loss": 3.121,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998948596489927e-05,
+      "loss": 3.1307,
+      "step": 645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998929216879638e-05,
+      "loss": 3.0513,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998909660311992e-05,
+      "loss": 3.0244,
+      "step": 655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99888992678768e-05,
+      "loss": 3.1356,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998870016307402e-05,
+      "loss": 3.0378,
+      "step": 665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998849928871861e-05,
+      "loss": 3.0598,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99882966448177e-05,
+      "loss": 3.1288,
+      "step": 675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998809223137844e-05,
+      "loss": 3.0913,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998788604840807e-05,
+      "loss": 3.2005,
+      "step": 685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998767809591391e-05,
+      "loss": 3.0776,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99874683739033e-05,
+      "loss": 3.089,
+      "step": 695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998725688238368e-05,
+      "loss": 3.0196,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998704362136252e-05,
+      "loss": 3.125,
+      "step": 705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998682859084738e-05,
+      "loss": 3.0342,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998661179084588e-05,
+      "loss": 3.1713,
+      "step": 715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998639322136567e-05,
+      "loss": 3.1364,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998617288241451e-05,
+      "loss": 3.0938,
+      "step": 725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998595077400019e-05,
+      "loss": 3.1047,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998572689613057e-05,
+      "loss": 3.0784,
+      "step": 735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998550124881359e-05,
+      "loss": 2.9785,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998527383205721e-05,
+      "loss": 3.1134,
+      "step": 745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998504464586949e-05,
+      "loss": 3.0216,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998481369025857e-05,
+      "loss": 3.0444,
+      "step": 755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99845809652326e-05,
+      "loss": 3.0694,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998434647079981e-05,
+      "loss": 3.0689,
+      "step": 765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998411020696853e-05,
+      "loss": 3.1201,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99838721737471e-05,
+      "loss": 3.1673,
+      "step": 775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998363237114397e-05,
+      "loss": 2.9911,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99833907991676e-05,
+      "loss": 3.1539,
+      "step": 785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998314745782656e-05,
+      "loss": 3.0299,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998290234712944e-05,
+      "loss": 3.104,
+      "step": 795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998265546708495e-05,
+      "loss": 3.0228,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998240681770182e-05,
+      "loss": 3.0504,
+      "step": 805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998215639898882e-05,
+      "loss": 3.0109,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998190421095487e-05,
+      "loss": 3.1135,
+      "step": 815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998165025360886e-05,
+      "loss": 3.0657,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998139452695978e-05,
+      "loss": 3.0811,
+      "step": 825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998113703101669e-05,
+      "loss": 3.0355,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998087776578871e-05,
+      "loss": 3.0186,
+      "step": 835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998061673128502e-05,
+      "loss": 3.0188,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998035392751485e-05,
+      "loss": 3.0557,
+      "step": 845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99800893544875e-05,
+      "loss": 3.0244,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997982301221234e-05,
+      "loss": 3.0035,
+      "step": 855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997955490069881e-05,
+      "loss": 3.0484,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99792850199564e-05,
+      "loss": 3.0623,
+      "step": 865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997901336999462e-05,
+      "loss": 3.0344,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997873995082316e-05,
+      "loss": 3.0852,
+      "step": 875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997846476245163e-05,
+      "loss": 3.1219,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997818780488983e-05,
+      "loss": 3.1688,
+      "step": 885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997790907814751e-05,
+      "loss": 3.0605,
+      "step": 890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997762858223457e-05,
+      "loss": 2.9544,
+      "step": 895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997734631716094e-05,
+      "loss": 3.063,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99770622829366e-05,
+      "loss": 3.0811,
+      "step": 905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99767764795716e-05,
+      "loss": 3.0792,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997648890707607e-05,
+      "loss": 3.0959,
+      "step": 915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997619956546018e-05,
+      "loss": 3.0228,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997590845473419e-05,
+      "loss": 3.1379,
+      "step": 925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997561557490838e-05,
+      "loss": 3.2007,
+      "step": 930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997532092599313e-05,
+      "loss": 2.9925,
+      "step": 935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997502450799887e-05,
+      "loss": 3.0537,
+      "step": 940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99747263209361e-05,
+      "loss": 2.9903,
+      "step": 945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997442636481535e-05,
+      "loss": 3.0693,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997412463964727e-05,
+      "loss": 3.1108,
+      "step": 955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997382114544254e-05,
+      "loss": 3.0721,
+      "step": 960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997351588221187e-05,
+      "loss": 3.0314,
+      "step": 965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997320884996609e-05,
+      "loss": 3.1156,
+      "step": 970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997290004871607e-05,
+      "loss": 3.0082,
+      "step": 975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997258947847275e-05,
+      "loss": 3.1044,
+      "step": 980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997227713924708e-05,
+      "loss": 3.0265,
+      "step": 985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997196303105018e-05,
+      "loss": 3.0057,
+      "step": 990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99716471538931e-05,
+      "loss": 3.0026,
+      "step": 995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99713295077871e-05,
+      "loss": 3.0668,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997101009274337e-05,
+      "loss": 3.0408,
+      "step": 1005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997068890877321e-05,
+      "loss": 3.0537,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997036595588802e-05,
+      "loss": 3.1123,
+      "step": 1015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997004123409923e-05,
+      "loss": 3.0983,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996971474341832e-05,
+      "loss": 3.0946,
+      "step": 1025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996938648385686e-05,
+      "loss": 3.1247,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996905645542645e-05,
+      "loss": 3.001,
+      "step": 1035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996872465813878e-05,
+      "loss": 3.0964,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996839109200563e-05,
+      "loss": 3.0655,
+      "step": 1045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996805575703876e-05,
+      "loss": 3.0552,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996771865325005e-05,
+      "loss": 3.1078,
+      "step": 1055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996737978065147e-05,
+      "loss": 3.1224,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996703913925497e-05,
+      "loss": 2.9502,
+      "step": 1065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996669672907264e-05,
+      "loss": 3.1495,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996635255011657e-05,
+      "loss": 3.0474,
+      "step": 1075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996600660239898e-05,
+      "loss": 3.1221,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996565888593208e-05,
+      "loss": 3.0501,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996530940072822e-05,
+      "loss": 3.0123,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996495814679974e-05,
+      "loss": 3.0347,
+      "step": 1095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996460512415908e-05,
+      "loss": 3.015,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996425033281873e-05,
+      "loss": 3.0922,
+      "step": 1105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996389377279127e-05,
+      "loss": 3.0729,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996353544408931e-05,
+      "loss": 3.0846,
+      "step": 1115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996317534672555e-05,
+      "loss": 3.1496,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99628134807127e-05,
+      "loss": 3.027,
+      "step": 1125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99624498460636e-05,
+      "loss": 3.0253,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99620844427911e-05,
+      "loss": 2.9497,
+      "step": 1135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996171727090817e-05,
+      "loss": 3.033,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996134833042778e-05,
+      "loss": 3.0603,
+      "step": 1145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996097762136298e-05,
+      "loss": 2.9652,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996060514372693e-05,
+      "loss": 3.0828,
+      "step": 1155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99602308975328e-05,
+      "loss": 2.9083,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99598548827938e-05,
+      "loss": 2.9923,
+      "step": 1165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99594770995233e-05,
+      "loss": 3.0421,
+      "step": 1170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995909754773464e-05,
+      "loss": 3.089,
+      "step": 1175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995871622744127e-05,
+      "loss": 3.1198,
+      "step": 1180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995833313865667e-05,
+      "loss": 3.0783,
+      "step": 1185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995794828139443e-05,
+      "loss": 3.0342,
+      "step": 1190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995756165566812e-05,
+      "loss": 3.057,
+      "step": 1195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995717326149149e-05,
+      "loss": 3.0348,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995678309887825e-05,
+      "loss": 3.052,
+      "step": 1205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995639116784222e-05,
+      "loss": 3.0808,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995599746839728e-05,
+      "loss": 3.087,
+      "step": 1215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995560200055736e-05,
+      "loss": 2.9872,
+      "step": 1220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995520476433647e-05,
+      "loss": 3.0481,
+      "step": 1225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995480575974865e-05,
+      "loss": 3.0529,
+      "step": 1230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995440498680803e-05,
+      "loss": 2.9599,
+      "step": 1235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995400244552884e-05,
+      "loss": 2.9881,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995359813592527e-05,
+      "loss": 3.0764,
+      "step": 1245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995319205801166e-05,
+      "loss": 3.0902,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995278421180237e-05,
+      "loss": 3.0722,
+      "step": 1255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995237459731186e-05,
+      "loss": 3.1431,
+      "step": 1260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995196321455462e-05,
+      "loss": 2.9468,
+      "step": 1265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99515500635452e-05,
+      "loss": 3.0437,
+      "step": 1270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995113514429824e-05,
+      "loss": 2.9424,
+      "step": 1275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995071845682843e-05,
+      "loss": 3.0279,
+      "step": 1280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995030000115051e-05,
+      "loss": 3.0837,
+      "step": 1285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99498797772793e-05,
+      "loss": 3.0939,
+      "step": 1290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994945778522966e-05,
+      "loss": 3.0062,
+      "step": 1295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994903402501657e-05,
+      "loss": 2.9881,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994860849665497e-05,
+      "loss": 3.0496,
+      "step": 1305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994818120015996e-05,
+      "loss": 3.0533,
+      "step": 1310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994775213554669e-05,
+      "loss": 3.0604,
+      "step": 1315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994732130283029e-05,
+      "loss": 3.0094,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994688870202604e-05,
+      "loss": 3.1205,
+      "step": 1325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994645433314926e-05,
+      "loss": 3.0397,
+      "step": 1330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994601819621532e-05,
+      "loss": 3.0296,
+      "step": 1335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994558029123965e-05,
+      "loss": 3.1577,
+      "step": 1340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994514061823779e-05,
+      "loss": 3.0443,
+      "step": 1345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994469917722525e-05,
+      "loss": 2.9162,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994425596821768e-05,
+      "loss": 3.0243,
+      "step": 1355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994381099123078e-05,
+      "loss": 3.106,
+      "step": 1360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994336424628027e-05,
+      "loss": 3.0942,
+      "step": 1365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994291573338201e-05,
+      "loss": 3.1291,
+      "step": 1370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994246545255185e-05,
+      "loss": 3.0217,
+      "step": 1375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994201340380573e-05,
+      "loss": 3.0713,
+      "step": 1380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994155958715966e-05,
+      "loss": 2.9684,
+      "step": 1385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99411040026297e-05,
+      "loss": 3.0492,
+      "step": 1390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994064665023196e-05,
+      "loss": 2.9889,
+      "step": 1395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994018752998267e-05,
+      "loss": 2.9818,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993972664189804e-05,
+      "loss": 3.0946,
+      "step": 1405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993926398599441e-05,
+      "loss": 3.0661,
+      "step": 1410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993879956228815e-05,
+      "loss": 3.0142,
+      "step": 1415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99383333707957e-05,
+      "loss": 3.133,
+      "step": 1420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993786541153357e-05,
+      "loss": 3.0596,
+      "step": 1425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993739568451832e-05,
+      "loss": 3.0774,
+      "step": 1430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993692418976657e-05,
+      "loss": 3.0545,
+      "step": 1435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993645092729503e-05,
+      "loss": 2.8985,
+      "step": 1440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993597589712042e-05,
+      "loss": 3.0264,
+      "step": 1445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99354990992596e-05,
+      "loss": 3.0303,
+      "step": 1450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993502053372942e-05,
+      "loss": 3.1349,
+      "step": 1455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993454020054682e-05,
+      "loss": 3.0456,
+      "step": 1460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993405809972881e-05,
+      "loss": 3.0251,
+      "step": 1465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993357423129246e-05,
+      "loss": 3.0916,
+      "step": 1470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993308859525489e-05,
+      "loss": 3.009,
+      "step": 1475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993260119163329e-05,
+      "loss": 2.9889,
+      "step": 1480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993211202044493e-05,
+      "loss": 2.9308,
+      "step": 1485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99316210817071e-05,
+      "loss": 3.052,
+      "step": 1490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99311283754372e-05,
+      "loss": 3.0082,
+      "step": 1495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993063390165267e-05,
+      "loss": 3.0434,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9930137660371e-05,
+      "loss": 3.0938,
+      "step": 1505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992963965160977e-05,
+      "loss": 3.0684,
+      "step": 1510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99291398753866e-05,
+      "loss": 3.0733,
+      "step": 1515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992863833171918e-05,
+      "loss": 3.0009,
+      "step": 1520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992813502062528e-05,
+      "loss": 2.9888,
+      "step": 1525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99276299421227e-05,
+      "loss": 3.1604,
+      "step": 1530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992712309622933e-05,
+      "loss": 3.045,
+      "step": 1535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992661448296311e-05,
+      "loss": 3.0428,
+      "step": 1540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992610410234203e-05,
+      "loss": 2.9848,
+      "step": 1545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992559195438418e-05,
+      "loss": 3.0696,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992507803910768e-05,
+      "loss": 3.0931,
+      "step": 1555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992456235653071e-05,
+      "loss": 3.0427,
+      "step": 1560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992404490667155e-05,
+      "loss": 3.002,
+      "step": 1565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992352568954849e-05,
+      "loss": 3.0721,
+      "step": 1570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992300470517993e-05,
+      "loss": 2.9975,
+      "step": 1575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99224819535843e-05,
+      "loss": 3.0291,
+      "step": 1580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99219574347801e-05,
+      "loss": 3.1075,
+      "step": 1585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992143114878592e-05,
+      "loss": 3.0338,
+      "step": 1590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992090309562039e-05,
+      "loss": 3.0802,
+      "step": 1595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992037327530217e-05,
+      "loss": 2.9946,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991984168785004e-05,
+      "loss": 3.0697,
+      "step": 1605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99193083332828e-05,
+      "loss": 3.064,
+      "step": 1610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991877321161937e-05,
+      "loss": 3.0914,
+      "step": 1615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991823632287864e-05,
+      "loss": 3.0359,
+      "step": 1620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991769766707965e-05,
+      "loss": 3.0663,
+      "step": 1625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991715724424147e-05,
+      "loss": 3.0491,
+      "step": 1630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991661505438321e-05,
+      "loss": 3.0298,
+      "step": 1635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991607109752406e-05,
+      "loss": 3.0551,
+      "step": 1640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99155253736833e-05,
+      "loss": 3.0167,
+      "step": 1645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991497788288024e-05,
+      "loss": 3.0528,
+      "step": 1650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991442862513425e-05,
+      "loss": 3.0228,
+      "step": 1655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991387760046478e-05,
+      "loss": 3.0734,
+      "step": 1660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991332480889133e-05,
+      "loss": 3.0014,
+      "step": 1665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991277025043349e-05,
+      "loss": 3.0098,
+      "step": 1670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991221392511088e-05,
+      "loss": 3.0714,
+      "step": 1675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991165583294317e-05,
+      "loss": 3.101,
+      "step": 1680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991109597395013e-05,
+      "loss": 2.9731,
+      "step": 1685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991053434815158e-05,
+      "loss": 3.0973,
+      "step": 1690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990997095556742e-05,
+      "loss": 2.9184,
+      "step": 1695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990940579621757e-05,
+      "loss": 2.9446,
+      "step": 1700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990883887012205e-05,
+      "loss": 3.001,
+      "step": 1705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990827017730091e-05,
+      "loss": 3.0518,
+      "step": 1710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99076997177743e-05,
+      "loss": 3.0368,
+      "step": 1715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990712749156241e-05,
+      "loss": 3.0616,
+      "step": 1720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99065534986855e-05,
+      "loss": 3.0363,
+      "step": 1725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990597773916386e-05,
+      "loss": 3.0702,
+      "step": 1730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99054002130179e-05,
+      "loss": 3.0159,
+      "step": 1735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990482092026807e-05,
+      "loss": 3.0144,
+      "step": 1740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990423986093485e-05,
+      "loss": 2.9505,
+      "step": 1745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990365703503884e-05,
+      "loss": 3.0405,
+      "step": 1750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990307244260062e-05,
+      "loss": 3.0283,
+      "step": 1755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990248608364094e-05,
+      "loss": 3.0908,
+      "step": 1760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990189795818054e-05,
+      "loss": 2.9963,
+      "step": 1765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990130806624023e-05,
+      "loss": 3.0316,
+      "step": 1770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990071640784089e-05,
+      "loss": 3.0006,
+      "step": 1775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990012298300346e-05,
+      "loss": 3.0472,
+      "step": 1780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989952779174899e-05,
+      "loss": 2.9787,
+      "step": 1785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989893083409846e-05,
+      "loss": 3.003,
+      "step": 1790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989833211007311e-05,
+      "loss": 2.9749,
+      "step": 1795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989773161969406e-05,
+      "loss": 3.0133,
+      "step": 1800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989712936298259e-05,
+      "loss": 3.0971,
+      "step": 1805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989652533996002e-05,
+      "loss": 2.998,
+      "step": 1810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989591955064772e-05,
+      "loss": 3.0452,
+      "step": 1815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989531199506714e-05,
+      "loss": 3.0065,
+      "step": 1820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98947026732398e-05,
+      "loss": 2.9922,
+      "step": 1825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989409158518726e-05,
+      "loss": 3.1558,
+      "step": 1830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989347873093116e-05,
+      "loss": 3.0061,
+      "step": 1835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989286411049317e-05,
+      "loss": 2.9763,
+      "step": 1840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989224772389507e-05,
+      "loss": 3.0319,
+      "step": 1845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989162957115867e-05,
+      "loss": 2.9838,
+      "step": 1850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989100965230586e-05,
+      "loss": 3.0188,
+      "step": 1855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989038796735856e-05,
+      "loss": 3.0322,
+      "step": 1860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988976451633882e-05,
+      "loss": 2.9699,
+      "step": 1865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988913929926869e-05,
+      "loss": 2.998,
+      "step": 1870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988851231617028e-05,
+      "loss": 3.0958,
+      "step": 1875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988788356706581e-05,
+      "loss": 3.0791,
+      "step": 1880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988725305197753e-05,
+      "loss": 2.9758,
+      "step": 1885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988662077092776e-05,
+      "loss": 3.0093,
+      "step": 1890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988598672393887e-05,
+      "loss": 3.0565,
+      "step": 1895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988535091103333e-05,
+      "loss": 3.0078,
+      "step": 1900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988471333223364e-05,
+      "loss": 3.1109,
+      "step": 1905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988407398756234e-05,
+      "loss": 3.0244,
+      "step": 1910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988343287704209e-05,
+      "loss": 3.1623,
+      "step": 1915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98827900006956e-05,
+      "loss": 2.9679,
+      "step": 1920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988214535854558e-05,
+      "loss": 3.0681,
+      "step": 1925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988149895061489e-05,
+      "loss": 2.9678,
+      "step": 1930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988085077692642e-05,
+      "loss": 3.011,
+      "step": 1935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988020083750307e-05,
+      "loss": 3.0665,
+      "step": 1940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987954913236788e-05,
+      "loss": 3.0602,
+      "step": 1945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98788956615439e-05,
+      "loss": 2.897,
+      "step": 1950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987824042505428e-05,
+      "loss": 3.003,
+      "step": 1955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987758342292219e-05,
+      "loss": 3.0566,
+      "step": 1960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987692465517093e-05,
+      "loss": 3.051,
+      "step": 1965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98762641218238e-05,
+      "loss": 3.0486,
+      "step": 1970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987560182290415e-05,
+      "loss": 3.0745,
+      "step": 1975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987493775843547e-05,
+      "loss": 3.0821,
+      "step": 1980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987427192844124e-05,
+      "loss": 3.0065,
+      "step": 1985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987360433294505e-05,
+      "loss": 3.0178,
+      "step": 1990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987293497197051e-05,
+      "loss": 3.1003,
+      "step": 1995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987226384554133e-05,
+      "loss": 3.0375,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987159095368126e-05,
+      "loss": 3.051,
+      "step": 2005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987091629641413e-05,
+      "loss": 2.945,
+      "step": 2010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98702398737638e-05,
+      "loss": 3.0465,
+      "step": 2015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986956168575425e-05,
+      "loss": 3.0049,
+      "step": 2020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986888173240946e-05,
+      "loss": 3.0303,
+      "step": 2025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98682000137535e-05,
+      "loss": 2.9473,
+      "step": 2030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986751652981053e-05,
+      "loss": 2.9811,
+      "step": 2035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98668312806047e-05,
+      "loss": 3.0405,
+      "step": 2040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98661442661603e-05,
+      "loss": 2.952,
+      "step": 2045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986545548650163e-05,
+      "loss": 3.0065,
+      "step": 2050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986476494165309e-05,
+      "loss": 3.0225,
+      "step": 2055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986407263163911e-05,
+      "loss": 2.9511,
+      "step": 2060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986337855648422e-05,
+      "loss": 3.0489,
+      "step": 2065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986268271621298e-05,
+      "loss": 3.0185,
+      "step": 2070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986198511084999e-05,
+      "loss": 3.0419,
+      "step": 2075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986128574041997e-05,
+      "loss": 3.023,
+      "step": 2080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98605846049477e-05,
+      "loss": 3.0055,
+      "step": 2085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985988170445797e-05,
+      "loss": 3.0722,
+      "step": 2090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985917703897566e-05,
+      "loss": 3.0632,
+      "step": 2095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985847060852573e-05,
+      "loss": 2.995,
+      "step": 2100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985776241313319e-05,
+      "loss": 3.0417,
+      "step": 2105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985705245282308e-05,
+      "loss": 3.0233,
+      "step": 2110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985634072762056e-05,
+      "loss": 3.0981,
+      "step": 2115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985562723755082e-05,
+      "loss": 3.0542,
+      "step": 2120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985491198263912e-05,
+      "loss": 3.0818,
+      "step": 2125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985419496291075e-05,
+      "loss": 3.047,
+      "step": 2130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985347617839113e-05,
+      "loss": 3.0325,
+      "step": 2135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985275562910569e-05,
+      "loss": 3.0492,
+      "step": 2140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985203331507993e-05,
+      "loss": 3.0665,
+      "step": 2145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985130923633943e-05,
+      "loss": 3.0714,
+      "step": 2150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985058339290983e-05,
+      "loss": 2.9251,
+      "step": 2155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984985578481678e-05,
+      "loss": 2.9738,
+      "step": 2160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984912641208608e-05,
+      "loss": 3.0148,
+      "step": 2165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984839527474354e-05,
+      "loss": 3.0759,
+      "step": 2170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984766237281504e-05,
+      "loss": 2.9501,
+      "step": 2175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98469277063265e-05,
+      "loss": 2.949,
+      "step": 2180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984619127530398e-05,
+      "loss": 2.9236,
+      "step": 2185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98454530797735e-05,
+      "loss": 3.0203,
+      "step": 2190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984471311976121e-05,
+      "loss": 3.1065,
+      "step": 2195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984397139529332e-05,
+      "loss": 2.9905,
+      "step": 2200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984322790639604e-05,
+      "loss": 2.9946,
+      "step": 2205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984248265309572e-05,
+      "loss": 2.8871,
+      "step": 2210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984173563541876e-05,
+      "loss": 3.0721,
+      "step": 2215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984098685339159e-05,
+      "loss": 3.0289,
+      "step": 2220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984023630704067e-05,
+      "loss": 3.1325,
+      "step": 2225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983948399639264e-05,
+      "loss": 2.9666,
+      "step": 2230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983872992147408e-05,
+      "loss": 2.9644,
+      "step": 2235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983797408231172e-05,
+      "loss": 3.0039,
+      "step": 2240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983721647893229e-05,
+      "loss": 3.0238,
+      "step": 2245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983645711136261e-05,
+      "loss": 3.0778,
+      "step": 2250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983569597962957e-05,
+      "loss": 3.0546,
+      "step": 2255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983493308376011e-05,
+      "loss": 3.0019,
+      "step": 2260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983416842378124e-05,
+      "loss": 2.9693,
+      "step": 2265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983340199972003e-05,
+      "loss": 3.0669,
+      "step": 2270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98326338116036e-05,
+      "loss": 3.0377,
+      "step": 2275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983186385945915e-05,
+      "loss": 3.0225,
+      "step": 2280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983109214331395e-05,
+      "loss": 3.0651,
+      "step": 2285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983031866319528e-05,
+      "loss": 3.0649,
+      "step": 2290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982954341913053e-05,
+      "loss": 3.0538,
+      "step": 2295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982876641114717e-05,
+      "loss": 3.0529,
+      "step": 2300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98279876392727e-05,
+      "loss": 2.9447,
+      "step": 2305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982720710353468e-05,
+      "loss": 3.0008,
+      "step": 2310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982642480396074e-05,
+      "loss": 2.9785,
+      "step": 2315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982564074057857e-05,
+      "loss": 3.0183,
+      "step": 2320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982485491341591e-05,
+      "loss": 3.0484,
+      "step": 2325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98240673225006e-05,
+      "loss": 2.9624,
+      "step": 2330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982327796786053e-05,
+      "loss": 3.0749,
+      "step": 2335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982248684952362e-05,
+      "loss": 3.0323,
+      "step": 2340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982169396751788e-05,
+      "loss": 3.031,
+      "step": 2345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982089932187137e-05,
+      "loss": 2.9852,
+      "step": 2350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982010291261223e-05,
+      "loss": 2.9158,
+      "step": 2355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981930473976866e-05,
+      "loss": 2.9657,
+      "step": 2360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98185048033689e-05,
+      "loss": 3.0377,
+      "step": 2365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981770310344127e-05,
+      "loss": 3.0126,
+      "step": 2370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981689964001417e-05,
+      "loss": 3.0059,
+      "step": 2375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981609441311602e-05,
+      "loss": 3.0344,
+      "step": 2380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981528742277531e-05,
+      "loss": 3.0183,
+      "step": 2385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981447866902063e-05,
+      "loss": 2.9703,
+      "step": 2390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98136681518806e-05,
+      "loss": 2.9756,
+      "step": 2395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981285587138391e-05,
+      "loss": 3.0525,
+      "step": 2400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981204182755935e-05,
+      "loss": 3.0522,
+      "step": 2405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981122602043567e-05,
+      "loss": 3.0494,
+      "step": 2410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98104084500418e-05,
+      "loss": 3.0019,
+      "step": 2415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980958911640666e-05,
+      "loss": 3.0223,
+      "step": 2420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980876801955928e-05,
+      "loss": 3.0744,
+      "step": 2425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980794515952867e-05,
+      "loss": 2.9729,
+      "step": 2430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980712053634402e-05,
+      "loss": 3.0289,
+      "step": 2435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980629415003449e-05,
+      "loss": 3.0979,
+      "step": 2440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980546600062932e-05,
+      "loss": 3.0207,
+      "step": 2445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980463608815786e-05,
+      "loss": 3.0574,
+      "step": 2450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980380441264947e-05,
+      "loss": 2.9954,
+      "step": 2455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98029709741336e-05,
+      "loss": 2.9789,
+      "step": 2460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980213577263973e-05,
+      "loss": 2.9664,
+      "step": 2465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980129880819744e-05,
+      "loss": 2.8813,
+      "step": 2470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980046008083637e-05,
+      "loss": 3.0391,
+      "step": 2475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979961959058619e-05,
+      "loss": 3.0051,
+      "step": 2480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979877733747666e-05,
+      "loss": 3.0405,
+      "step": 2485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979793332153762e-05,
+      "loss": 3.0239,
+      "step": 2490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979708754279889e-05,
+      "loss": 3.0185,
+      "step": 2495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979624000129047e-05,
+      "loss": 2.9855,
+      "step": 2500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979539069704231e-05,
+      "loss": 2.986,
+      "step": 2505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979453963008451e-05,
+      "loss": 3.0786,
+      "step": 2510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979368680044719e-05,
+      "loss": 2.9413,
+      "step": 2515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979283220816053e-05,
+      "loss": 2.969,
+      "step": 2520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979197585325479e-05,
+      "loss": 2.9967,
+      "step": 2525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979111773576029e-05,
+      "loss": 3.0836,
+      "step": 2530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979025785570739e-05,
+      "loss": 2.9593,
+      "step": 2535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978939621312653e-05,
+      "loss": 3.0512,
+      "step": 2540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978853280804822e-05,
+      "loss": 2.972,
+      "step": 2545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978766764050305e-05,
+      "loss": 2.9795,
+      "step": 2550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978680071052159e-05,
+      "loss": 2.9579,
+      "step": 2555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978593201813457e-05,
+      "loss": 3.0316,
+      "step": 2560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978506156337273e-05,
+      "loss": 2.9429,
+      "step": 2565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978418934626686e-05,
+      "loss": 2.9628,
+      "step": 2570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978331536684789e-05,
+      "loss": 3.0401,
+      "step": 2575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97824396251467e-05,
+      "loss": 2.9933,
+      "step": 2580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978156212119432e-05,
+      "loss": 3.0071,
+      "step": 2585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97806828550218e-05,
+      "loss": 2.9615,
+      "step": 2590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977980182666029e-05,
+      "loss": 2.9554,
+      "step": 2595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977891903614096e-05,
+      "loss": 2.9756,
+      "step": 2600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977803448349506e-05,
+      "loss": 2.9694,
+      "step": 2605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977714816875388e-05,
+      "loss": 2.9343,
+      "step": 2610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977626009194885e-05,
+      "loss": 3.0038,
+      "step": 2615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977537025311136e-05,
+      "loss": 2.9981,
+      "step": 2620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977447865227291e-05,
+      "loss": 3.0134,
+      "step": 2625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97735852894651e-05,
+      "loss": 2.9817,
+      "step": 2630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977269016471953e-05,
+      "loss": 3.0229,
+      "step": 2635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977179327806787e-05,
+      "loss": 2.9778,
+      "step": 2640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97708946295419e-05,
+      "loss": 2.9776,
+      "step": 2645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97699942191734e-05,
+      "loss": 3.0216,
+      "step": 2650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976909204699428e-05,
+      "loss": 2.9781,
+      "step": 2655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976818811303646e-05,
+      "loss": 3.0725,
+      "step": 2660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976728241733192e-05,
+      "loss": 3.0265,
+      "step": 2665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976637495991275e-05,
+      "loss": 3.0517,
+      "step": 2670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976546574081105e-05,
+      "loss": 3.06,
+      "step": 2675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976455476005902e-05,
+      "loss": 2.9812,
+      "step": 2680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976364201768889e-05,
+      "loss": 3.1274,
+      "step": 2685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976272751373298e-05,
+      "loss": 3.0256,
+      "step": 2690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976181124822368e-05,
+      "loss": 3.052,
+      "step": 2695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976089322119341e-05,
+      "loss": 2.9129,
+      "step": 2700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975997343267467e-05,
+      "loss": 2.9319,
+      "step": 2705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97590518827e-05,
+      "loss": 2.9994,
+      "step": 2710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975812857130207e-05,
+      "loss": 3.0796,
+      "step": 2715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97572034985135e-05,
+      "loss": 2.9394,
+      "step": 2720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975627666436709e-05,
+      "loss": 2.9456,
+      "step": 2725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975534806889563e-05,
+      "loss": 3.0629,
+      "step": 2730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975441771213198e-05,
+      "loss": 3.0888,
+      "step": 2735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975348559410909e-05,
+      "loss": 3.0131,
+      "step": 2740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975255171485996e-05,
+      "loss": 2.9506,
+      "step": 2745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975161607441762e-05,
+      "loss": 3.0424,
+      "step": 2750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975067867281522e-05,
+      "loss": 2.9013,
+      "step": 2755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974973951008592e-05,
+      "loss": 3.0445,
+      "step": 2760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9748798586263e-05,
+      "loss": 3.0438,
+      "step": 2765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974785590137974e-05,
+      "loss": 3.0346,
+      "step": 2770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974691145546952e-05,
+      "loss": 2.9911,
+      "step": 2775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974596524856577e-05,
+      "loss": 3.0648,
+      "step": 2780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974501728070196e-05,
+      "loss": 2.9887,
+      "step": 2785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97440675519117e-05,
+      "loss": 3.0192,
+      "step": 2790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974311606222856e-05,
+      "loss": 3.0332,
+      "step": 2795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974216281168625e-05,
+      "loss": 3.0493,
+      "step": 2800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97412078003185e-05,
+      "loss": 2.9904,
+      "step": 2805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974025102815913e-05,
+      "loss": 3.0827,
+      "step": 2810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9739292495242e-05,
+      "loss": 2.9394,
+      "step": 2815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973833220160105e-05,
+      "loss": 2.928,
+      "step": 2820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973737014727025e-05,
+      "loss": 2.9613,
+      "step": 2825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973640633228369e-05,
+      "loss": 2.9834,
+      "step": 2830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973544075667545e-05,
+      "loss": 3.0304,
+      "step": 2835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973447342047974e-05,
+      "loss": 3.0432,
+      "step": 2840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973350432373081e-05,
+      "loss": 2.9231,
+      "step": 2845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973253346646294e-05,
+      "loss": 2.9617,
+      "step": 2850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973156084871049e-05,
+      "loss": 2.9403,
+      "step": 2855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973058647050792e-05,
+      "loss": 3.0881,
+      "step": 2860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97296103318897e-05,
+      "loss": 3.092,
+      "step": 2865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972863243289042e-05,
+      "loss": 3.0953,
+      "step": 2870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972765277354464e-05,
+      "loss": 3.0723,
+      "step": 2875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972667135388707e-05,
+      "loss": 3.0063,
+      "step": 2880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972568817395245e-05,
+      "loss": 2.9966,
+      "step": 2885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972470323377558e-05,
+      "loss": 2.9165,
+      "step": 2890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972371653339134e-05,
+      "loss": 3.0226,
+      "step": 2895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972272807283465e-05,
+      "loss": 2.9807,
+      "step": 2900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972173785214049e-05,
+      "loss": 3.0246,
+      "step": 2905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972074587134391e-05,
+      "loss": 3.0381,
+      "step": 2910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971975213048005e-05,
+      "loss": 3.0614,
+      "step": 2915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971875662958407e-05,
+      "loss": 2.8764,
+      "step": 2920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971775936869121e-05,
+      "loss": 3.0277,
+      "step": 2925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971676034783678e-05,
+      "loss": 2.967,
+      "step": 2930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971575956705615e-05,
+      "loss": 2.9875,
+      "step": 2935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971475702638471e-05,
+      "loss": 3.0435,
+      "step": 2940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971375272585799e-05,
+      "loss": 2.8592,
+      "step": 2945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971274666551152e-05,
+      "loss": 3.0412,
+      "step": 2950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971173884538093e-05,
+      "loss": 3.0406,
+      "step": 2955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971072926550189e-05,
+      "loss": 3.0705,
+      "step": 2960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970971792591012e-05,
+      "loss": 3.0083,
+      "step": 2965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970870482664143e-05,
+      "loss": 2.9662,
+      "step": 2970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97076899677317e-05,
+      "loss": 2.9632,
+      "step": 2975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970667334921684e-05,
+      "loss": 3.0741,
+      "step": 2980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970565497113284e-05,
+      "loss": 3.0247,
+      "step": 2985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970463483351576e-05,
+      "loss": 3.0036,
+      "step": 2990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97036129364017e-05,
+      "loss": 3.0901,
+      "step": 2995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970258927982681e-05,
+      "loss": 3.0086,
+      "step": 3000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970156386382736e-05,
+      "loss": 2.9753,
+      "step": 3005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970053668843966e-05,
+      "loss": 3.0188,
+      "step": 3010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969950775370004e-05,
+      "loss": 3.0252,
+      "step": 3015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969847705964494e-05,
+      "loss": 2.957,
+      "step": 3020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969744460631082e-05,
+      "loss": 2.9955,
+      "step": 3025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969641039373427e-05,
+      "loss": 2.948,
+      "step": 3030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969537442195187e-05,
+      "loss": 2.9839,
+      "step": 3035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96943366910003e-05,
+      "loss": 3.004,
+      "step": 3040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969329720091631e-05,
+      "loss": 3.003,
+      "step": 3045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969225595173667e-05,
+      "loss": 2.9513,
+      "step": 3050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969121294349826e-05,
+      "loss": 3.0602,
+      "step": 3055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969016817623798e-05,
+      "loss": 2.8861,
+      "step": 3060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968912164999284e-05,
+      "loss": 2.9904,
+      "step": 3065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968807336479986e-05,
+      "loss": 3.0732,
+      "step": 3070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968702332069617e-05,
+      "loss": 3.0611,
+      "step": 3075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968597151771894e-05,
+      "loss": 2.9827,
+      "step": 3080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968491795590538e-05,
+      "loss": 3.011,
+      "step": 3085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968386263529282e-05,
+      "loss": 2.911,
+      "step": 3090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968280555591857e-05,
+      "loss": 3.0809,
+      "step": 3095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968174671782009e-05,
+      "loss": 2.9208,
+      "step": 3100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968068612103485e-05,
+      "loss": 3.042,
+      "step": 3105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96796237656004e-05,
+      "loss": 3.0034,
+      "step": 3110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967855965155432e-05,
+      "loss": 3.0745,
+      "step": 3115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967749377893431e-05,
+      "loss": 3.0036,
+      "step": 3120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96764261477781e-05,
+      "loss": 3.1009,
+      "step": 3125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967535675812346e-05,
+      "loss": 2.9871,
+      "step": 3130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967428561000827e-05,
+      "loss": 2.9476,
+      "step": 3135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96732127034704e-05,
+      "loss": 2.9362,
+      "step": 3140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967213803854791e-05,
+      "loss": 2.9964,
+      "step": 3145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967106161527877e-05,
+      "loss": 2.9279,
+      "step": 3150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96699834337011e-05,
+      "loss": 2.9891,
+      "step": 3155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966890349385312e-05,
+      "loss": 2.9419,
+      "step": 3160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966782179577299e-05,
+      "loss": 2.9961,
+      "step": 3165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966673833949903e-05,
+      "loss": 3.0022,
+      "step": 3170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966565312506959e-05,
+      "loss": 3.0255,
+      "step": 3175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966456615252307e-05,
+      "loss": 2.9403,
+      "step": 3180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966347742189799e-05,
+      "loss": 2.9334,
+      "step": 3185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966238693323286e-05,
+      "loss": 3.0124,
+      "step": 3190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966129468656628e-05,
+      "loss": 2.9999,
+      "step": 3195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966020068193693e-05,
+      "loss": 3.0007,
+      "step": 3200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965910491938351e-05,
+      "loss": 3.0546,
+      "step": 3205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965800739894485e-05,
+      "loss": 3.0056,
+      "step": 3210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965690812065978e-05,
+      "loss": 3.0149,
+      "step": 3215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965580708456718e-05,
+      "loss": 3.0369,
+      "step": 3220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96547042907061e-05,
+      "loss": 3.009,
+      "step": 3225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96535997391155e-05,
+      "loss": 2.9625,
+      "step": 3230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965249342983452e-05,
+      "loss": 3.0495,
+      "step": 3235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965138536290233e-05,
+      "loss": 3.0526,
+      "step": 3240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965027553835814e-05,
+      "loss": 2.9856,
+      "step": 3245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964916395624124e-05,
+      "loss": 3.0284,
+      "step": 3250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964805061659098e-05,
+      "loss": 3.0376,
+      "step": 3255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964693551944676e-05,
+      "loss": 3.0123,
+      "step": 3260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964581866484806e-05,
+      "loss": 3.0034,
+      "step": 3265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964470005283441e-05,
+      "loss": 3.0127,
+      "step": 3270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964357968344545e-05,
+      "loss": 3.0376,
+      "step": 3275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964245755672078e-05,
+      "loss": 3.0893,
+      "step": 3280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964133367270014e-05,
+      "loss": 3.0025,
+      "step": 3285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964020803142335e-05,
+      "loss": 3.1064,
+      "step": 3290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963908063293021e-05,
+      "loss": 3.0391,
+      "step": 3295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963795147726065e-05,
+      "loss": 2.9744,
+      "step": 3300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963682056445463e-05,
+      "loss": 2.9106,
+      "step": 3305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963568789455221e-05,
+      "loss": 2.9824,
+      "step": 3310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963455346759347e-05,
+      "loss": 3.0245,
+      "step": 3315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963341728361855e-05,
+      "loss": 3.0072,
+      "step": 3320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963227934266767e-05,
+      "loss": 3.0221,
+      "step": 3325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963113964478117e-05,
+      "loss": 2.9594,
+      "step": 3330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962999818999932e-05,
+      "loss": 2.9239,
+      "step": 3335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962885497836256e-05,
+      "loss": 3.0166,
+      "step": 3340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962771000991138e-05,
+      "loss": 2.9651,
+      "step": 3345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962656328468627e-05,
+      "loss": 3.0053,
+      "step": 3350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962541480272782e-05,
+      "loss": 3.0101,
+      "step": 3355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962426456407674e-05,
+      "loss": 3.1157,
+      "step": 3360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96231125687737e-05,
+      "loss": 3.0032,
+      "step": 3365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962195881685949e-05,
+      "loss": 2.9722,
+      "step": 3370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962080330837495e-05,
+      "loss": 2.8825,
+      "step": 3375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961964604336099e-05,
+      "loss": 3.095,
+      "step": 3380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961848702185857e-05,
+      "loss": 3.0879,
+      "step": 3385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961732624390874e-05,
+      "loss": 3.0154,
+      "step": 3390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961616370955256e-05,
+      "loss": 3.032,
+      "step": 3395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96149994188312e-05,
+      "loss": 2.9317,
+      "step": 3400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961383337178586e-05,
+      "loss": 3.0456,
+      "step": 3405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961266556845783e-05,
+      "loss": 3.0549,
+      "step": 3410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961149600888845e-05,
+      "loss": 3.0361,
+      "step": 3415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961032469311912e-05,
+      "loss": 3.0992,
+      "step": 3420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960915162119128e-05,
+      "loss": 3.0064,
+      "step": 3425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96079767931465e-05,
+      "loss": 2.9916,
+      "step": 3430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960680020902636e-05,
+      "loss": 2.909,
+      "step": 3435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960562186887246e-05,
+      "loss": 2.925,
+      "step": 3440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960444177272656e-05,
+      "loss": 2.9856,
+      "step": 3445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960325992063042e-05,
+      "loss": 3.0358,
+      "step": 3450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960207631262587e-05,
+      "loss": 3.03,
+      "step": 3455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960089094875482e-05,
+      "loss": 2.932,
+      "step": 3460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959970382905924e-05,
+      "loss": 2.9137,
+      "step": 3465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959851495358112e-05,
+      "loss": 2.9818,
+      "step": 3470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959732432236257e-05,
+      "loss": 3.0376,
+      "step": 3475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959613193544574e-05,
+      "loss": 3.0025,
+      "step": 3480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959493779287283e-05,
+      "loss": 2.9679,
+      "step": 3485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95937418946861e-05,
+      "loss": 3.0676,
+      "step": 3490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95925442409279e-05,
+      "loss": 2.9468,
+      "step": 3495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959134483164063e-05,
+      "loss": 3.0216,
+      "step": 3500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959014366686673e-05,
+      "loss": 3.0904,
+      "step": 3505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958894074664874e-05,
+      "loss": 2.981,
+      "step": 3510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958773607102922e-05,
+      "loss": 3.0052,
+      "step": 3515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958652964005084e-05,
+      "loss": 2.9781,
+      "step": 3520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95853214537563e-05,
+      "loss": 3.0438,
+      "step": 3525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958411151218833e-05,
+      "loss": 3.0753,
+      "step": 3530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958289981538982e-05,
+      "loss": 3.0378,
+      "step": 3535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958168636340363e-05,
+      "loss": 2.9415,
+      "step": 3540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958047115627272e-05,
+      "loss": 2.9369,
+      "step": 3545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95792541940401e-05,
+      "loss": 2.9891,
+      "step": 3550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957803547674886e-05,
+      "loss": 2.9393,
+      "step": 3555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957681500444216e-05,
+      "loss": 2.9487,
+      "step": 3560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957559277716315e-05,
+      "loss": 2.9555,
+      "step": 3565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957436879495514e-05,
+      "loss": 3.0317,
+      "step": 3570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957314305786145e-05,
+      "loss": 2.9519,
+      "step": 3575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957191556592547e-05,
+      "loss": 3.0368,
+      "step": 3580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957068631919064e-05,
+      "loss": 2.8944,
+      "step": 3585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956945531770048e-05,
+      "loss": 2.9985,
+      "step": 3590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956822256149857e-05,
+      "loss": 2.9768,
+      "step": 3595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956698805062855e-05,
+      "loss": 2.908,
+      "step": 3600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956575178513412e-05,
+      "loss": 2.9684,
+      "step": 3605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956451376505902e-05,
+      "loss": 3.0077,
+      "step": 3610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956327399044712e-05,
+      "loss": 2.9965,
+      "step": 3615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956203246134226e-05,
+      "loss": 3.0244,
+      "step": 3620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956078917778841e-05,
+      "loss": 2.9566,
+      "step": 3625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95595441398296e-05,
+      "loss": 2.9665,
+      "step": 3630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955829734750988e-05,
+      "loss": 2.9046,
+      "step": 3635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955704880087337e-05,
+      "loss": 2.9852,
+      "step": 3640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955579849996432e-05,
+      "loss": 2.9264,
+      "step": 3645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955454644482693e-05,
+      "loss": 3.0298,
+      "step": 3650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955329263550556e-05,
+      "loss": 3.0616,
+      "step": 3655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955203707204457e-05,
+      "loss": 2.9497,
+      "step": 3660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955077975448843e-05,
+      "loss": 3.018,
+      "step": 3665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954952068288162e-05,
+      "loss": 3.0602,
+      "step": 3670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954825985726875e-05,
+      "loss": 2.9914,
+      "step": 3675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954699727769441e-05,
+      "loss": 2.9751,
+      "step": 3680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95457329442033e-05,
+      "loss": 2.9519,
+      "step": 3685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954446685684021e-05,
+      "loss": 2.9349,
+      "step": 3690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954319901564993e-05,
+      "loss": 3.0266,
+      "step": 3695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954192942067734e-05,
+      "loss": 2.9789,
+      "step": 3700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95406580719674e-05,
+      "loss": 2.9422,
+      "step": 3705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953938496956508e-05,
+      "loss": 3.0087,
+      "step": 3710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95381101135155e-05,
+      "loss": 3.0031,
+      "step": 3715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953683350386374e-05,
+      "loss": 3.0132,
+      "step": 3720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953555514065502e-05,
+      "loss": 2.979,
+      "step": 3725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953427502393456e-05,
+      "loss": 3.0076,
+      "step": 3730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953299315374771e-05,
+      "loss": 2.8576,
+      "step": 3735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953170953013983e-05,
+      "loss": 3.0865,
+      "step": 3740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953042415315638e-05,
+      "loss": 3.0233,
+      "step": 3745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952913702284284e-05,
+      "loss": 3.0071,
+      "step": 3750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952784813924476e-05,
+      "loss": 2.9994,
+      "step": 3755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95265575024078e-05,
+      "loss": 3.0062,
+      "step": 3760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952526511237763e-05,
+      "loss": 2.9599,
+      "step": 3765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952397096919998e-05,
+      "loss": 2.9478,
+      "step": 3770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95226750729207e-05,
+      "loss": 2.9963,
+      "step": 3775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952137742358566e-05,
+      "loss": 2.9125,
+      "step": 3780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952007802124075e-05,
+      "loss": 2.9913,
+      "step": 3785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951877686593201e-05,
+      "loss": 3.0518,
+      "step": 3790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951747395770548e-05,
+      "loss": 2.9719,
+      "step": 3795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951616929660729e-05,
+      "loss": 3.028,
+      "step": 3800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951486288268361e-05,
+      "loss": 2.9387,
+      "step": 3805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951355471598072e-05,
+      "loss": 2.9776,
+      "step": 3810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951224479654491e-05,
+      "loss": 2.9778,
+      "step": 3815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951093312442253e-05,
+      "loss": 3.0645,
+      "step": 3820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950961969966003e-05,
+      "loss": 3.0319,
+      "step": 3825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950830452230391e-05,
+      "loss": 2.9775,
+      "step": 3830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950698759240071e-05,
+      "loss": 3.0065,
+      "step": 3835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950566890999706e-05,
+      "loss": 2.9211,
+      "step": 3840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950434847513966e-05,
+      "loss": 3.0445,
+      "step": 3845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95030262878752e-05,
+      "loss": 2.9883,
+      "step": 3850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950170234825052e-05,
+      "loss": 3.0185,
+      "step": 3855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950037665631247e-05,
+      "loss": 3.1024,
+      "step": 3860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9499049212108e-05,
+      "loss": 2.9851,
+      "step": 3865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949772001568406e-05,
+      "loss": 3.0239,
+      "step": 3870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949638906708778e-05,
+      "loss": 2.9706,
+      "step": 3875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949505636636617e-05,
+      "loss": 3.1047,
+      "step": 3880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94937219135665e-05,
+      "loss": 3.02,
+      "step": 3885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949238570873596e-05,
+      "loss": 2.9452,
+      "step": 3890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949104775192186e-05,
+      "loss": 3.0124,
+      "step": 3895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948970804317155e-05,
+      "loss": 2.9319,
+      "step": 3900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948836658253247e-05,
+      "loss": 3.0636,
+      "step": 3905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948702337005211e-05,
+      "loss": 3.0625,
+      "step": 3910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9485678405778e-05,
+      "loss": 2.995,
+      "step": 3915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948433168975777e-05,
+      "loss": 3.0327,
+      "step": 3920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948298322203909e-05,
+      "loss": 3.0167,
+      "step": 3925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948163300266965e-05,
+      "loss": 3.0464,
+      "step": 3930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948028103169733e-05,
+      "loss": 3.0685,
+      "step": 3935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947892730916992e-05,
+      "loss": 2.9431,
+      "step": 3940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947757183513538e-05,
+      "loss": 2.9623,
+      "step": 3945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947621460964167e-05,
+      "loss": 2.946,
+      "step": 3950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947485563273684e-05,
+      "loss": 2.9761,
+      "step": 3955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947349490446898e-05,
+      "loss": 3.0113,
+      "step": 3960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94721324248863e-05,
+      "loss": 3.0281,
+      "step": 3965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9470768194037e-05,
+      "loss": 3.0513,
+      "step": 3970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946940221196937e-05,
+      "loss": 3.0266,
+      "step": 3975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946803447873179e-05,
+      "loss": 2.9652,
+      "step": 3980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946666499437265e-05,
+      "loss": 2.9812,
+      "step": 3985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946529375894043e-05,
+      "loss": 2.9361,
+      "step": 3990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946392077248367e-05,
+      "loss": 2.9016,
+      "step": 3995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946254603505101e-05,
+      "loss": 2.9606,
+      "step": 4000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946116954669107e-05,
+      "loss": 3.0147,
+      "step": 4005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94597913074526e-05,
+      "loss": 3.0731,
+      "step": 4010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945841131738436e-05,
+      "loss": 3.0191,
+      "step": 4015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945702957653524e-05,
+      "loss": 2.9803,
+      "step": 4020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945564608495412e-05,
+      "loss": 3.0734,
+      "step": 4025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945426084269e-05,
+      "loss": 3.0425,
+      "step": 4030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94528738497919e-05,
+      "loss": 2.8675,
+      "step": 4035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945148510630892e-05,
+      "loss": 2.9177,
+      "step": 4040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945009461229023e-05,
+      "loss": 2.9476,
+      "step": 4045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944870236778501e-05,
+      "loss": 3.0003,
+      "step": 4050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944730837284261e-05,
+      "loss": 2.9846,
+      "step": 4055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944591262751233e-05,
+      "loss": 3.0009,
+      "step": 4060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944451513184361e-05,
+      "loss": 3.028,
+      "step": 4065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944311588588588e-05,
+      "loss": 2.972,
+      "step": 4070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944171488968872e-05,
+      "loss": 2.8916,
+      "step": 4075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944031214330167e-05,
+      "loss": 3.0353,
+      "step": 4080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943890764677444e-05,
+      "loss": 2.9952,
+      "step": 4085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943750140015671e-05,
+      "loss": 2.951,
+      "step": 4090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943609340349826e-05,
+      "loss": 3.0441,
+      "step": 4095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943468365684896e-05,
+      "loss": 2.9694,
+      "step": 4100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943327216025868e-05,
+      "loss": 2.9807,
+      "step": 4105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943185891377743e-05,
+      "loss": 3.066,
+      "step": 4110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943044391745518e-05,
+      "loss": 2.9567,
+      "step": 4115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942902717134207e-05,
+      "loss": 2.9217,
+      "step": 4120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942760867548822e-05,
+      "loss": 2.9924,
+      "step": 4125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942618842994387e-05,
+      "loss": 2.9544,
+      "step": 4130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942476643475926e-05,
+      "loss": 2.9877,
+      "step": 4135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942334268998476e-05,
+      "loss": 3.0326,
+      "step": 4140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942191719567075e-05,
+      "loss": 3.0064,
+      "step": 4145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94204899518677e-05,
+      "loss": 2.987,
+      "step": 4150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941906095862615e-05,
+      "loss": 2.9484,
+      "step": 4155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941763021599664e-05,
+      "loss": 3.0135,
+      "step": 4160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941619772402986e-05,
+      "loss": 3.0317,
+      "step": 4165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941476348277648e-05,
+      "loss": 2.9238,
+      "step": 4170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941332749228731e-05,
+      "loss": 2.9481,
+      "step": 4175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941188975261317e-05,
+      "loss": 3.0198,
+      "step": 4180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941045026380493e-05,
+      "loss": 2.907,
+      "step": 4185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940900902591358e-05,
+      "loss": 3.0012,
+      "step": 4190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940756603899013e-05,
+      "loss": 2.9967,
+      "step": 4195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940612130308565e-05,
+      "loss": 2.9233,
+      "step": 4200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940467481825126e-05,
+      "loss": 3.0431,
+      "step": 4205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940322658453822e-05,
+      "loss": 3.0415,
+      "step": 4210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940177660199776e-05,
+      "loss": 2.9571,
+      "step": 4215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940032487068122e-05,
+      "loss": 3.0,
+      "step": 4220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939887139064e-05,
+      "loss": 2.9482,
+      "step": 4225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939741616192552e-05,
+      "loss": 2.921,
+      "step": 4230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93959591845893e-05,
+      "loss": 2.9855,
+      "step": 4235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939450045868295e-05,
+      "loss": 3.0015,
+      "step": 4240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939303998425808e-05,
+      "loss": 3.0409,
+      "step": 4245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939157776136639e-05,
+      "loss": 2.9771,
+      "step": 4250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939011379005963e-05,
+      "loss": 2.9898,
+      "step": 4255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938864807038966e-05,
+      "loss": 2.9884,
+      "step": 4260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938718060240833e-05,
+      "loss": 2.9934,
+      "step": 4265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938571138616761e-05,
+      "loss": 2.9244,
+      "step": 4270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938424042171949e-05,
+      "loss": 2.9957,
+      "step": 4275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938276770911605e-05,
+      "loss": 3.0776,
+      "step": 4280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938129324840943e-05,
+      "loss": 3.0554,
+      "step": 4285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937981703965181e-05,
+      "loss": 3.0136,
+      "step": 4290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937833908289547e-05,
+      "loss": 2.987,
+      "step": 4295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937685937819269e-05,
+      "loss": 2.9603,
+      "step": 4300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937537792559588e-05,
+      "loss": 2.9353,
+      "step": 4305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937389472515747e-05,
+      "loss": 2.9074,
+      "step": 4310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937240977692997e-05,
+      "loss": 2.9518,
+      "step": 4315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937092308096595e-05,
+      "loss": 2.9973,
+      "step": 4320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936943463731803e-05,
+      "loss": 2.9662,
+      "step": 4325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93679444460389e-05,
+      "loss": 2.989,
+      "step": 4330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93664525071813e-05,
+      "loss": 2.9576,
+      "step": 4335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936495882079806e-05,
+      "loss": 3.0048,
+      "step": 4340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936346338694206e-05,
+      "loss": 2.9974,
+      "step": 4345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936196620566623e-05,
+      "loss": 2.9549,
+      "step": 4350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936046727702355e-05,
+      "loss": 2.9728,
+      "step": 4355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93589666010671e-05,
+      "loss": 2.9477,
+      "step": 4360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935746417785e-05,
+      "loss": 2.9688,
+      "step": 4365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935596000742544e-05,
+      "loss": 3.0714,
+      "step": 4370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935445408984665e-05,
+      "loss": 3.0204,
+      "step": 4375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935294642516696e-05,
+      "loss": 2.9928,
+      "step": 4380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935143701343972e-05,
+      "loss": 3.0121,
+      "step": 4385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934992585471837e-05,
+      "loss": 2.9934,
+      "step": 4390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934841294905642e-05,
+      "loss": 3.0334,
+      "step": 4395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934689829650739e-05,
+      "loss": 2.9623,
+      "step": 4400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93453818971249e-05,
+      "loss": 2.9682,
+      "step": 4405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934386375096267e-05,
+      "loss": 3.0411,
+      "step": 4410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934234385807442e-05,
+      "loss": 2.9381,
+      "step": 4415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934082221851393e-05,
+      "loss": 2.9476,
+      "step": 4420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933929883233508e-05,
+      "loss": 3.0033,
+      "step": 4425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933777369959182e-05,
+      "loss": 2.9673,
+      "step": 4430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93362468203381e-05,
+      "loss": 3.0322,
+      "step": 4435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9334718194628e-05,
+      "loss": 3.048,
+      "step": 4440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933318782251562e-05,
+      "loss": 3.0039,
+      "step": 4445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933165570405513e-05,
+      "loss": 3.0445,
+      "step": 4450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933012183930077e-05,
+      "loss": 3.0747,
+      "step": 4455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932858622830683e-05,
+      "loss": 2.936,
+      "step": 4460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932704887112768e-05,
+      "loss": 3.0247,
+      "step": 4465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932550976781774e-05,
+      "loss": 3.0899,
+      "step": 4470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932396891843147e-05,
+      "loss": 3.0345,
+      "step": 4475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932242632302347e-05,
+      "loss": 3.0082,
+      "step": 4480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932088198164828e-05,
+      "loss": 2.9942,
+      "step": 4485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931933589436061e-05,
+      "loss": 2.9459,
+      "step": 4490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931778806121517e-05,
+      "loss": 2.9913,
+      "step": 4495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931623848226678e-05,
+      "loss": 3.0135,
+      "step": 4500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931468715757026e-05,
+      "loss": 2.9421,
+      "step": 4505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931313408718054e-05,
+      "loss": 2.9794,
+      "step": 4510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93115792711526e-05,
+      "loss": 3.0632,
+      "step": 4515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931002270954148e-05,
+      "loss": 3.0003,
+      "step": 4520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930846440240227e-05,
+      "loss": 2.9596,
+      "step": 4525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930690434979014e-05,
+      "loss": 3.0602,
+      "step": 4530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93053425517603e-05,
+      "loss": 2.9632,
+      "step": 4535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930377900836807e-05,
+      "loss": 2.9344,
+      "step": 4540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930221371966877e-05,
+      "loss": 2.9447,
+      "step": 4545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930064668571781e-05,
+      "loss": 2.9725,
+      "step": 4550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929907790657069e-05,
+      "loss": 2.948,
+      "step": 4555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929750738228291e-05,
+      "loss": 2.9843,
+      "step": 4560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929593511291007e-05,
+      "loss": 2.9521,
+      "step": 4565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929436109850785e-05,
+      "loss": 3.0104,
+      "step": 4570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929278533913194e-05,
+      "loss": 2.9796,
+      "step": 4575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929120783483813e-05,
+      "loss": 2.9258,
+      "step": 4580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.928962858568228e-05,
+      "loss": 2.9531,
+      "step": 4585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.928804759172027e-05,
+      "loss": 2.8915,
+      "step": 4590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.928646485300808e-05,
+      "loss": 2.9925,
+      "step": 4595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928488036960172e-05,
+      "loss": 3.0131,
+      "step": 4600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928329414155731e-05,
+      "loss": 2.9812,
+      "step": 4605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928170616893095e-05,
+      "loss": 2.97,
+      "step": 4610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928011645177892e-05,
+      "loss": 2.9834,
+      "step": 4615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927852499015745e-05,
+      "loss": 2.8973,
+      "step": 4620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927693178412287e-05,
+      "loss": 3.0356,
+      "step": 4625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927533683373163e-05,
+      "loss": 2.9351,
+      "step": 4630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927374013904011e-05,
+      "loss": 2.9307,
+      "step": 4635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92721417001049e-05,
+      "loss": 3.0307,
+      "step": 4640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927054151698256e-05,
+      "loss": 3.0859,
+      "step": 4645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926893958972972e-05,
+      "loss": 2.9174,
+      "step": 4650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926733591840312e-05,
+      "loss": 2.9855,
+      "step": 4655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92657305030595e-05,
+      "loss": 2.985,
+      "step": 4660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926412334375569e-05,
+      "loss": 2.879,
+      "step": 4665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926251444054859e-05,
+      "loss": 2.9829,
+      "step": 4670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926090379349516e-05,
+      "loss": 2.9555,
+      "step": 4675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925929140265242e-05,
+      "loss": 2.9301,
+      "step": 4680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925767726807742e-05,
+      "loss": 3.0118,
+      "step": 4685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925606138982733e-05,
+      "loss": 2.9655,
+      "step": 4690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925444376795932e-05,
+      "loss": 3.0041,
+      "step": 4695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925282440253068e-05,
+      "loss": 3.0387,
+      "step": 4700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925120329359871e-05,
+      "loss": 3.0178,
+      "step": 4705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924958044122083e-05,
+      "loss": 2.9266,
+      "step": 4710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924795584545444e-05,
+      "loss": 2.9661,
+      "step": 4715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92463295063571e-05,
+      "loss": 3.0314,
+      "step": 4720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924470142398634e-05,
+      "loss": 2.9874,
+      "step": 4725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92430715983998e-05,
+      "loss": 3.0332,
+      "step": 4730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92414400296552e-05,
+      "loss": 2.9598,
+      "step": 4735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923980671781029e-05,
+      "loss": 3.0391,
+      "step": 4740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923817166292284e-05,
+      "loss": 2.9106,
+      "step": 4745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923653486505078e-05,
+      "loss": 2.8826,
+      "step": 4750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923489632425205e-05,
+      "loss": 2.9736,
+      "step": 4755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923325604058463e-05,
+      "loss": 3.0148,
+      "step": 4760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92316140141066e-05,
+      "loss": 3.0648,
+      "step": 4765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922997024487605e-05,
+      "loss": 2.9879,
+      "step": 4770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922832473295123e-05,
+      "loss": 2.9626,
+      "step": 4775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922667747839036e-05,
+      "loss": 3.0272,
+      "step": 4780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922502848125173e-05,
+      "loss": 3.0741,
+      "step": 4785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922337774159374e-05,
+      "loss": 2.9682,
+      "step": 4790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922172525947482e-05,
+      "loss": 3.0335,
+      "step": 4795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922007103495346e-05,
+      "loss": 3.0608,
+      "step": 4800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921841506808823e-05,
+      "loss": 2.9817,
+      "step": 4805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921675735893772e-05,
+      "loss": 3.0345,
+      "step": 4810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921509790756064e-05,
+      "loss": 3.0117,
+      "step": 4815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921343671401573e-05,
+      "loss": 3.0429,
+      "step": 4820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921177377836177e-05,
+      "loss": 2.9856,
+      "step": 4825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921010910065766e-05,
+      "loss": 2.9468,
+      "step": 4830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920844268096232e-05,
+      "loss": 3.0407,
+      "step": 4835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920677451933473e-05,
+      "loss": 2.9912,
+      "step": 4840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920510461583395e-05,
+      "loss": 3.0118,
+      "step": 4845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920343297051908e-05,
+      "loss": 3.0022,
+      "step": 4850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92017595834493e-05,
+      "loss": 2.9615,
+      "step": 4855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920008445468385e-05,
+      "loss": 2.9936,
+      "step": 4860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919840758428202e-05,
+      "loss": 2.9749,
+      "step": 4865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919672897230319e-05,
+      "loss": 3.0364,
+      "step": 4870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919504861880675e-05,
+      "loss": 3.0127,
+      "step": 4875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919336652385223e-05,
+      "loss": 2.9593,
+      "step": 4880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919168268749912e-05,
+      "loss": 2.9752,
+      "step": 4885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918999710980705e-05,
+      "loss": 3.0523,
+      "step": 4890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91883097908357e-05,
+      "loss": 3.0365,
+      "step": 4895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918662073064479e-05,
+      "loss": 2.9108,
+      "step": 4900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91849299292941e-05,
+      "loss": 2.9407,
+      "step": 4905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91832373868435e-05,
+      "loss": 3.1141,
+      "step": 4910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91815431033529e-05,
+      "loss": 2.9403,
+      "step": 4915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917984707888226e-05,
+      "loss": 3.1246,
+      "step": 4920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917814931349163e-05,
+      "loss": 2.9886,
+      "step": 4925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917644980724112e-05,
+      "loss": 3.0826,
+      "step": 4930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91747485601909e-05,
+      "loss": 2.9224,
+      "step": 4935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917304557240114e-05,
+      "loss": 3.0875,
+      "step": 4940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917134084393216e-05,
+      "loss": 2.9607,
+      "step": 4945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916963437484432e-05,
+      "loss": 3.0444,
+      "step": 4950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9167926165198e-05,
+      "loss": 3.04,
+      "step": 4955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916621621505369e-05,
+      "loss": 2.9488,
+      "step": 4960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91645045244719e-05,
+      "loss": 3.0168,
+      "step": 4965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916279109351323e-05,
+      "loss": 2.8925,
+      "step": 4970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916107592223834e-05,
+      "loss": 2.9446,
+      "step": 4975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915935901070794e-05,
+      "loss": 2.9646,
+      "step": 4980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915764035898281e-05,
+      "loss": 2.9684,
+      "step": 4985
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915591996712378e-05,
+      "loss": 3.0147,
+      "step": 4990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915419783519177e-05,
+      "loss": 3.0094,
+      "step": 4995
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915247396324773e-05,
+      "loss": 3.0365,
+      "step": 5000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915074835135269e-05,
+      "loss": 2.9625,
+      "step": 5005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914902099956771e-05,
+      "loss": 3.0555,
+      "step": 5010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914729190795396e-05,
+      "loss": 2.9605,
+      "step": 5015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914556107657264e-05,
+      "loss": 2.9801,
+      "step": 5020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914382850548504e-05,
+      "loss": 2.9232,
+      "step": 5025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914209419475247e-05,
+      "loss": 2.9808,
+      "step": 5030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914035814443631e-05,
+      "loss": 2.9777,
+      "step": 5035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913862035459804e-05,
+      "loss": 3.0873,
+      "step": 5040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913688082529917e-05,
+      "loss": 2.9267,
+      "step": 5045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913513955660127e-05,
+      "loss": 2.9052,
+      "step": 5050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913339654856597e-05,
+      "loss": 2.9362,
+      "step": 5055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913165180125502e-05,
+      "loss": 3.0014,
+      "step": 5060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912990531473012e-05,
+      "loss": 2.9072,
+      "step": 5065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912815708905313e-05,
+      "loss": 3.0049,
+      "step": 5070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912640712428593e-05,
+      "loss": 3.0028,
+      "step": 5075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912465542049046e-05,
+      "loss": 3.0148,
+      "step": 5080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912290197772873e-05,
+      "loss": 3.0118,
+      "step": 5085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912114679606283e-05,
+      "loss": 2.9201,
+      "step": 5090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911938987555485e-05,
+      "loss": 3.017,
+      "step": 5095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911763121626702e-05,
+      "loss": 2.9293,
+      "step": 5100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911587081826159e-05,
+      "loss": 3.0529,
+      "step": 5105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911410868160087e-05,
+      "loss": 2.9329,
+      "step": 5110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911234480634723e-05,
+      "loss": 2.9468,
+      "step": 5115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911057919256312e-05,
+      "loss": 2.9985,
+      "step": 5120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910881184031104e-05,
+      "loss": 3.067,
+      "step": 5125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910704274965355e-05,
+      "loss": 2.8689,
+      "step": 5130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910527192065327e-05,
+      "loss": 2.8975,
+      "step": 5135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910349935337289e-05,
+      "loss": 2.9399,
+      "step": 5140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910172504787516e-05,
+      "loss": 2.9549,
+      "step": 5145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909994900422288e-05,
+      "loss": 2.9154,
+      "step": 5150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909817122247894e-05,
+      "loss": 2.9689,
+      "step": 5155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909639170270626e-05,
+      "loss": 2.974,
+      "step": 5160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909461044496782e-05,
+      "loss": 2.9197,
+      "step": 5165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909282744932669e-05,
+      "loss": 2.8425,
+      "step": 5170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909104271584599e-05,
+      "loss": 2.9488,
+      "step": 5175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908925624458887e-05,
+      "loss": 2.9359,
+      "step": 5180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908746803561861e-05,
+      "loss": 3.0072,
+      "step": 5185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908567808899849e-05,
+      "loss": 2.9537,
+      "step": 5190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908388640479187e-05,
+      "loss": 2.9632,
+      "step": 5195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908209298306217e-05,
+      "loss": 2.9771,
+      "step": 5200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90802978238729e-05,
+      "loss": 2.9897,
+      "step": 5205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907850092728757e-05,
+      "loss": 3.0602,
+      "step": 5210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907670229336983e-05,
+      "loss": 2.9815,
+      "step": 5215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907490192218331e-05,
+      "loss": 3.067,
+      "step": 5220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907309981379176e-05,
+      "loss": 3.0309,
+      "step": 5225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907129596825898e-05,
+      "loss": 3.02,
+      "step": 5230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90694903856488e-05,
+      "loss": 3.1764,
+      "step": 5235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906768306602519e-05,
+      "loss": 2.9672,
+      "step": 5240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906587400945207e-05,
+      "loss": 2.8706,
+      "step": 5245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906406321599349e-05,
+      "loss": 3.052,
+      "step": 5250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906225068571357e-05,
+      "loss": 2.9265,
+      "step": 5255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906043641867646e-05,
+      "loss": 2.9652,
+      "step": 5260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905862041494639e-05,
+      "loss": 2.9992,
+      "step": 5265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905680267458763e-05,
+      "loss": 2.9748,
+      "step": 5270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905498319766454e-05,
+      "loss": 2.9596,
+      "step": 5275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905316198424152e-05,
+      "loss": 3.0264,
+      "step": 5280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905133903438304e-05,
+      "loss": 3.0644,
+      "step": 5285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904951434815363e-05,
+      "loss": 2.9685,
+      "step": 5290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904768792561791e-05,
+      "loss": 2.9185,
+      "step": 5295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904585976684048e-05,
+      "loss": 2.8829,
+      "step": 5300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90440298718861e-05,
+      "loss": 2.9908,
+      "step": 5305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904219824081953e-05,
+      "loss": 2.935,
+      "step": 5310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904036487370562e-05,
+      "loss": 2.9567,
+      "step": 5315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903852977060924e-05,
+      "loss": 3.0683,
+      "step": 5320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903669293159538e-05,
+      "loss": 2.8993,
+      "step": 5325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903485435672904e-05,
+      "loss": 3.0462,
+      "step": 5330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903301404607532e-05,
+      "loss": 2.9016,
+      "step": 5335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903117199969938e-05,
+      "loss": 2.9129,
+      "step": 5340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902932821766639e-05,
+      "loss": 3.0508,
+      "step": 5345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902748270004163e-05,
+      "loss": 2.9182,
+      "step": 5350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902563544689044e-05,
+      "loss": 3.0232,
+      "step": 5355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902378645827821e-05,
+      "loss": 2.9287,
+      "step": 5360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902193573427038e-05,
+      "loss": 2.9448,
+      "step": 5365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902008327493248e-05,
+      "loss": 3.025,
+      "step": 5370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90182290803301e-05,
+      "loss": 3.0479,
+      "step": 5375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901637315052882e-05,
+      "loss": 3.0048,
+      "step": 5380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901451548559438e-05,
+      "loss": 3.0323,
+      "step": 5385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901265608559253e-05,
+      "loss": 2.9611,
+      "step": 5390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901079495058911e-05,
+      "loss": 2.9239,
+      "step": 5395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.900893208064998e-05,
+      "loss": 2.9323,
+      "step": 5400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.900706747584109e-05,
+      "loss": 2.9718,
+      "step": 5405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.900520113622845e-05,
+      "loss": 3.0162,
+      "step": 5410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90033330618781e-05,
+      "loss": 2.9336,
+      "step": 5415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90014632528562e-05,
+      "loss": 2.9882,
+      "step": 5420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.899959170922893e-05,
+      "loss": 2.9688,
+      "step": 5425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.899771843106254e-05,
+      "loss": 3.0982,
+      "step": 5430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899584341842334e-05,
+      "loss": 2.9658,
+      "step": 5435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89939666713777e-05,
+      "loss": 2.9167,
+      "step": 5440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899208818999208e-05,
+      "loss": 3.018,
+      "step": 5445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899020797433293e-05,
+      "loss": 2.9772,
+      "step": 5450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898832602446686e-05,
+      "loss": 3.0288,
+      "step": 5455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898644234046046e-05,
+      "loss": 2.9445,
+      "step": 5460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898455692238042e-05,
+      "loss": 2.9733,
+      "step": 5465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898266977029346e-05,
+      "loss": 2.9712,
+      "step": 5470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89807808842664e-05,
+      "loss": 2.9473,
+      "step": 5475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897889026436612e-05,
+      "loss": 3.1029,
+      "step": 5480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897699791065952e-05,
+      "loss": 3.1108,
+      "step": 5485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897510382321361e-05,
+      "loss": 2.9241,
+      "step": 5490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897320800209542e-05,
+      "loss": 3.0635,
+      "step": 5495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897131044737207e-05,
+      "loss": 3.0169,
+      "step": 5500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896941115911074e-05,
+      "loss": 2.9665,
+      "step": 5505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896751013737863e-05,
+      "loss": 2.9493,
+      "step": 5510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896560738224308e-05,
+      "loss": 3.0361,
+      "step": 5515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89637028937714e-05,
+      "loss": 2.9028,
+      "step": 5520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896179667203104e-05,
+      "loss": 2.9507,
+      "step": 5525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895988871708948e-05,
+      "loss": 2.9857,
+      "step": 5530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895797902901424e-05,
+      "loss": 2.9025,
+      "step": 5535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895606760787293e-05,
+      "loss": 2.8885,
+      "step": 5540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895415445373321e-05,
+      "loss": 3.0084,
+      "step": 5545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89522395666628e-05,
+      "loss": 3.1155,
+      "step": 5550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89503229467295e-05,
+      "loss": 3.0207,
+      "step": 5555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894840459400114e-05,
+      "loss": 3.04,
+      "step": 5560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894648450854564e-05,
+      "loss": 3.0282,
+      "step": 5565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894456269043096e-05,
+      "loss": 2.9565,
+      "step": 5570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894263913972513e-05,
+      "loss": 2.9674,
+      "step": 5575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894071385649627e-05,
+      "loss": 2.8976,
+      "step": 5580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893878684081248e-05,
+      "loss": 3.0132,
+      "step": 5585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893685809274202e-05,
+      "loss": 3.0009,
+      "step": 5590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893492761235315e-05,
+      "loss": 3.105,
+      "step": 5595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893299539971421e-05,
+      "loss": 2.891,
+      "step": 5600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893106145489359e-05,
+      "loss": 2.8761,
+      "step": 5605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892912577795976e-05,
+      "loss": 3.0163,
+      "step": 5610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892718836898124e-05,
+      "loss": 3.0194,
+      "step": 5615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892524922802661e-05,
+      "loss": 2.9912,
+      "step": 5620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89233083551645e-05,
+      "loss": 3.0731,
+      "step": 5625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892136575046364e-05,
+      "loss": 2.9931,
+      "step": 5630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891942141399279e-05,
+      "loss": 2.9991,
+      "step": 5635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891747534582076e-05,
+      "loss": 2.9629,
+      "step": 5640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891552754601644e-05,
+      "loss": 2.9123,
+      "step": 5645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891357801464882e-05,
+      "loss": 3.1166,
+      "step": 5650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891162675178687e-05,
+      "loss": 2.9968,
+      "step": 5655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890967375749967e-05,
+      "loss": 2.9323,
+      "step": 5660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890771903185639e-05,
+      "loss": 2.8654,
+      "step": 5665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890576257492617e-05,
+      "loss": 3.0387,
+      "step": 5670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890380438677828e-05,
+      "loss": 2.9115,
+      "step": 5675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890184446748206e-05,
+      "loss": 2.975,
+      "step": 5680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88998828171069e-05,
+      "loss": 2.9556,
+      "step": 5685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889791943572219e-05,
+      "loss": 3.0586,
+      "step": 5690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88959543233975e-05,
+      "loss": 3.0661,
+      "step": 5695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889398748020231e-05,
+      "loss": 2.9294,
+      "step": 5700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889201890620631e-05,
+      "loss": 2.9886,
+      "step": 5705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889004860147917e-05,
+      "loss": 2.9814,
+      "step": 5710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888807656609063e-05,
+      "loss": 3.0104,
+      "step": 5715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888610280011051e-05,
+      "loss": 2.817,
+      "step": 5720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888412730360864e-05,
+      "loss": 2.8664,
+      "step": 5725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.8882150076655e-05,
+      "loss": 2.9757,
+      "step": 5730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888017111931955e-05,
+      "loss": 2.9426,
+      "step": 5735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887819043167238e-05,
+      "loss": 3.0162,
+      "step": 5740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887620801378357e-05,
+      "loss": 3.0291,
+      "step": 5745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88742238657233e-05,
+      "loss": 3.0385,
+      "step": 5750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887223798756181e-05,
+      "loss": 2.9749,
+      "step": 5755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887025037936941e-05,
+      "loss": 2.9856,
+      "step": 5760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886826104121645e-05,
+      "loss": 2.9452,
+      "step": 5765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886626997317335e-05,
+      "loss": 3.024,
+      "step": 5770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886427717531058e-05,
+      "loss": 3.0157,
+      "step": 5775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88622826476987e-05,
+      "loss": 2.9045,
+      "step": 5780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886028639040832e-05,
+      "loss": 2.987,
+      "step": 5785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88582884035101e-05,
+      "loss": 3.0116,
+      "step": 5790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885628868707475e-05,
+      "loss": 2.9719,
+      "step": 5795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885428724117307e-05,
+      "loss": 2.9187,
+      "step": 5800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885228406587592e-05,
+      "loss": 2.9484,
+      "step": 5805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88502791612542e-05,
+      "loss": 2.9793,
+      "step": 5810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884827252737888e-05,
+      "loss": 2.899,
+      "step": 5815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.8846264164321e-05,
+      "loss": 2.9837,
+      "step": 5820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884425407215164e-05,
+      "loss": 2.9476,
+      "step": 5825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884224225094198e-05,
+      "loss": 2.9683,
+      "step": 5830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884022870076321e-05,
+      "loss": 2.9799,
+      "step": 5835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883821342168662e-05,
+      "loss": 2.9731,
+      "step": 5840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883619641378354e-05,
+      "loss": 2.9583,
+      "step": 5845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88341776771254e-05,
+      "loss": 2.9044,
+      "step": 5850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883215721178362e-05,
+      "loss": 2.9485,
+      "step": 5855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883013501782976e-05,
+      "loss": 2.9676,
+      "step": 5860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882811109533537e-05,
+      "loss": 2.9649,
+      "step": 5865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882608544437212e-05,
+      "loss": 2.9391,
+      "step": 5870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88240580650117e-05,
+      "loss": 2.9498,
+      "step": 5875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88220289573259e-05,
+      "loss": 2.957,
+      "step": 5880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881999812138652e-05,
+      "loss": 3.0578,
+      "step": 5885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881796555726546e-05,
+      "loss": 2.9233,
+      "step": 5890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881593126503467e-05,
+      "loss": 2.9748,
+      "step": 5895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881389524476618e-05,
+      "loss": 2.9654,
+      "step": 5900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881185749653206e-05,
+      "loss": 2.9864,
+      "step": 5905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880981802040441e-05,
+      "loss": 2.9672,
+      "step": 5910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880777681645546e-05,
+      "loss": 2.971,
+      "step": 5915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880573388475745e-05,
+      "loss": 2.8765,
+      "step": 5920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88036892253827e-05,
+      "loss": 2.9548,
+      "step": 5925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880164283840358e-05,
+      "loss": 3.0301,
+      "step": 5930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879959472389257e-05,
+      "loss": 2.9565,
+      "step": 5935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879754488192212e-05,
+      "loss": 2.9403,
+      "step": 5940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879549331256483e-05,
+      "loss": 3.0611,
+      "step": 5945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879344001589331e-05,
+      "loss": 2.8974,
+      "step": 5950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879138499198024e-05,
+      "loss": 2.9005,
+      "step": 5955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878932824089838e-05,
+      "loss": 3.0343,
+      "step": 5960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878726976272052e-05,
+      "loss": 2.962,
+      "step": 5965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878520955751953e-05,
+      "loss": 2.9909,
+      "step": 5970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878314762536836e-05,
+      "loss": 2.9877,
+      "step": 5975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878108396633996e-05,
+      "loss": 2.955,
+      "step": 5980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877901858050742e-05,
+      "loss": 2.9339,
+      "step": 5985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877695146794385e-05,
+      "loss": 2.9217,
+      "step": 5990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877488262872242e-05,
+      "loss": 3.0209,
+      "step": 5995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877281206291634e-05,
+      "loss": 2.9231,
+      "step": 6000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877073977059892e-05,
+      "loss": 3.0652,
+      "step": 6005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876866575184352e-05,
+      "loss": 3.0289,
+      "step": 6010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876659000672357e-05,
+      "loss": 2.9779,
+      "step": 6015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876451253531254e-05,
+      "loss": 2.9875,
+      "step": 6020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876243333768397e-05,
+      "loss": 2.967,
+      "step": 6025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876035241391146e-05,
+      "loss": 2.965,
+      "step": 6030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875826976406867e-05,
+      "loss": 2.9517,
+      "step": 6035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875618538822933e-05,
+      "loss": 2.9673,
+      "step": 6040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875409928646724e-05,
+      "loss": 2.936,
+      "step": 6045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875201145885621e-05,
+      "loss": 3.0495,
+      "step": 6050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874992190547016e-05,
+      "loss": 2.9878,
+      "step": 6055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874783062638309e-05,
+      "loss": 3.0693,
+      "step": 6060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874573762166899e-05,
+      "loss": 3.0078,
+      "step": 6065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874364289140195e-05,
+      "loss": 2.9784,
+      "step": 6070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874154643565615e-05,
+      "loss": 3.0062,
+      "step": 6075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873944825450578e-05,
+      "loss": 2.9491,
+      "step": 6080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873734834802512e-05,
+      "loss": 3.0337,
+      "step": 6085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873524671628854e-05,
+      "loss": 3.0528,
+      "step": 6090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873314335937036e-05,
+      "loss": 3.0805,
+      "step": 6095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873103827734511e-05,
+      "loss": 3.0031,
+      "step": 6100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872893147028726e-05,
+      "loss": 2.9492,
+      "step": 6105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87268229382714e-05,
+      "loss": 3.0755,
+      "step": 6110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872471268137218e-05,
+      "loss": 2.9789,
+      "step": 6115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872260069966431e-05,
+      "loss": 2.8737,
+      "step": 6120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872048699322251e-05,
+      "loss": 3.0131,
+      "step": 6125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871837156212167e-05,
+      "loss": 2.9708,
+      "step": 6130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871625440643661e-05,
+      "loss": 2.9807,
+      "step": 6135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871413552624232e-05,
+      "loss": 2.9396,
+      "step": 6140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871201492161378e-05,
+      "loss": 2.9865,
+      "step": 6145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870989259262605e-05,
+      "loss": 2.9968,
+      "step": 6150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87077685393543e-05,
+      "loss": 3.0007,
+      "step": 6155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870564276187367e-05,
+      "loss": 3.0313,
+      "step": 6160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870351526025945e-05,
+      "loss": 2.9183,
+      "step": 6165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870138603458694e-05,
+      "loss": 3.047,
+      "step": 6170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.86992550849315e-05,
+      "loss": 2.9216,
+      "step": 6175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869712241136857e-05,
+      "loss": 3.0631,
+      "step": 6180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869498801397364e-05,
+      "loss": 2.9309,
+      "step": 6185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869285189282229e-05,
+      "loss": 3.0807,
+      "step": 6190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869071404799011e-05,
+      "loss": 3.0529,
+      "step": 6195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.868857447955277e-05,
+      "loss": 2.9266,
+      "step": 6200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.868643318758605e-05,
+      "loss": 3.0233,
+      "step": 6205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.868429017216572e-05,
+      "loss": 2.9595,
+      "step": 6210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.868214543336762e-05,
+      "loss": 3.0317,
+      "step": 6215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867999897126772e-05,
+      "loss": 2.9257,
+      "step": 6220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867785078594199e-05,
+      "loss": 2.8842,
+      "step": 6225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867570087746644e-05,
+      "loss": 3.1383,
+      "step": 6230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867354924591722e-05,
+      "loss": 3.05,
+      "step": 6235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867139589137047e-05,
+      "loss": 3.0391,
+      "step": 6240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.86692408139024e-05,
+      "loss": 3.0109,
+      "step": 6245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.866708401358933e-05,
+      "loss": 2.9906,
+      "step": 6250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.866492549050761e-05,
+      "loss": 2.9405,
+      "step": 6255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.866276524473363e-05,
+      "loss": 3.0502,
+      "step": 6260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.866060327634386e-05,
+      "loss": 2.9057,
+      "step": 6265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865843958541485e-05,
+      "loss": 2.9816,
+      "step": 6270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865627417202316e-05,
+      "loss": 3.0321,
+      "step": 6275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865410703624549e-05,
+      "loss": 2.9682,
+      "step": 6280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865193817815852e-05,
+      "loss": 3.0773,
+      "step": 6285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864976759783903e-05,
+      "loss": 2.9808,
+      "step": 6290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864759529536387e-05,
+      "loss": 2.9296,
+      "step": 6295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864542127080993e-05,
+      "loss": 2.9871,
+      "step": 6300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864324552425416e-05,
+      "loss": 2.9531,
+      "step": 6305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86410680557736e-05,
+      "loss": 3.0028,
+      "step": 6310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863888886544532e-05,
+      "loss": 2.9291,
+      "step": 6315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863670795334644e-05,
+      "loss": 2.9317,
+      "step": 6320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863452531955419e-05,
+      "loss": 2.9465,
+      "step": 6325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863234096414582e-05,
+      "loss": 2.9379,
+      "step": 6330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863015488719867e-05,
+      "loss": 2.9835,
+      "step": 6335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86279670887901e-05,
+      "loss": 3.0339,
+      "step": 6340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862577756899758e-05,
+      "loss": 2.9747,
+      "step": 6345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86235863278986e-05,
+      "loss": 3.0312,
+      "step": 6350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862139336557074e-05,
+      "loss": 2.9641,
+      "step": 6355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861919868209162e-05,
+      "loss": 2.9765,
+      "step": 6360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861700227753891e-05,
+      "loss": 2.8491,
+      "step": 6365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861480415199041e-05,
+      "loss": 2.977,
+      "step": 6370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86126043055239e-05,
+      "loss": 3.01,
+      "step": 6375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861040273821727e-05,
+      "loss": 2.9983,
+      "step": 6380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860819945014842e-05,
+      "loss": 2.885,
+      "step": 6385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860599444139537e-05,
+      "loss": 2.9958,
+      "step": 6390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860378771203617e-05,
+      "loss": 2.9787,
+      "step": 6395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860157926214895e-05,
+      "loss": 2.9915,
+      "step": 6400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859936909181184e-05,
+      "loss": 2.9977,
+      "step": 6405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859715720110315e-05,
+      "loss": 2.9861,
+      "step": 6410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859494359010111e-05,
+      "loss": 2.8848,
+      "step": 6415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859272825888413e-05,
+      "loss": 2.9383,
+      "step": 6420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85905112075306e-05,
+      "loss": 3.0358,
+      "step": 6425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858829243611902e-05,
+      "loss": 3.0264,
+      "step": 6430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85860719447279e-05,
+      "loss": 3.0721,
+      "step": 6435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85838497334359e-05,
+      "loss": 3.0171,
+      "step": 6440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858162580232164e-05,
+      "loss": 3.0035,
+      "step": 6445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857940015146385e-05,
+      "loss": 2.9014,
+      "step": 6450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857717278094134e-05,
+      "loss": 2.9834,
+      "step": 6455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857494369083292e-05,
+      "loss": 3.0346,
+      "step": 6460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857271288121752e-05,
+      "loss": 2.9905,
+      "step": 6465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857048035217412e-05,
+      "loss": 2.967,
+      "step": 6470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856824610378173e-05,
+      "loss": 2.9588,
+      "step": 6475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856601013611943e-05,
+      "loss": 3.0224,
+      "step": 6480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856377244926641e-05,
+      "loss": 3.0073,
+      "step": 6485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856153304330185e-05,
+      "loss": 3.0598,
+      "step": 6490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855929191830502e-05,
+      "loss": 2.991,
+      "step": 6495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855704907435528e-05,
+      "loss": 2.9643,
+      "step": 6500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855480451153201e-05,
+      "loss": 3.0143,
+      "step": 6505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855255822991465e-05,
+      "loss": 3.009,
+      "step": 6510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855031022958274e-05,
+      "loss": 2.9806,
+      "step": 6515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854806051061585e-05,
+      "loss": 2.9861,
+      "step": 6520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854580907309362e-05,
+      "loss": 2.9377,
+      "step": 6525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854355591709574e-05,
+      "loss": 2.9598,
+      "step": 6530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854130104270199e-05,
+      "loss": 3.0045,
+      "step": 6535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853904444999216e-05,
+      "loss": 3.0057,
+      "step": 6540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853678613904613e-05,
+      "loss": 2.896,
+      "step": 6545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853452610994389e-05,
+      "loss": 3.0166,
+      "step": 6550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853226436276539e-05,
+      "loss": 3.0658,
+      "step": 6555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853000089759072e-05,
+      "loss": 3.0017,
+      "step": 6560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85277357145e-05,
+      "loss": 2.929,
+      "step": 6565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852546881357343e-05,
+      "loss": 2.9655,
+      "step": 6570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852320019489123e-05,
+      "loss": 2.9415,
+      "step": 6575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852092985853372e-05,
+      "loss": 3.0555,
+      "step": 6580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851865780458125e-05,
+      "loss": 2.9671,
+      "step": 6585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851638403311427e-05,
+      "loss": 2.9994,
+      "step": 6590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851410854421328e-05,
+      "loss": 3.0317,
+      "step": 6595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85118313379588e-05,
+      "loss": 2.981,
+      "step": 6600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850955241443146e-05,
+      "loss": 2.9149,
+      "step": 6605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850727177371193e-05,
+      "loss": 2.913,
+      "step": 6610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850498941588094e-05,
+      "loss": 2.9829,
+      "step": 6615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850270534101926e-05,
+      "loss": 2.9365,
+      "step": 6620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850041954920779e-05,
+      "loss": 2.9669,
+      "step": 6625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849813204052743e-05,
+      "loss": 2.9264,
+      "step": 6630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849584281505913e-05,
+      "loss": 3.0822,
+      "step": 6635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849355187288393e-05,
+      "loss": 3.0115,
+      "step": 6640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849125921408297e-05,
+      "loss": 3.0293,
+      "step": 6645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848896483873736e-05,
+      "loss": 2.9732,
+      "step": 6650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848666874692834e-05,
+      "loss": 3.0084,
+      "step": 6655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848437093873719e-05,
+      "loss": 2.9674,
+      "step": 6660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848207141424524e-05,
+      "loss": 3.1193,
+      "step": 6665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847977017353392e-05,
+      "loss": 2.9473,
+      "step": 6670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847746721668465e-05,
+      "loss": 2.9681,
+      "step": 6675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847516254377897e-05,
+      "loss": 3.0176,
+      "step": 6680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847285615489847e-05,
+      "loss": 2.9296,
+      "step": 6685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847054805012478e-05,
+      "loss": 2.9323,
+      "step": 6690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846823822953963e-05,
+      "loss": 3.0002,
+      "step": 6695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846592669322476e-05,
+      "loss": 2.9083,
+      "step": 6700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846361344126202e-05,
+      "loss": 2.9861,
+      "step": 6705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846129847373327e-05,
+      "loss": 2.9644,
+      "step": 6710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845898179072046e-05,
+      "loss": 3.0218,
+      "step": 6715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845666339230564e-05,
+      "loss": 3.0047,
+      "step": 6720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845434327857081e-05,
+      "loss": 2.9162,
+      "step": 6725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845202144959816e-05,
+      "loss": 3.0294,
+      "step": 6730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844969790546986e-05,
+      "loss": 2.9964,
+      "step": 6735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844737264626816e-05,
+      "loss": 2.9892,
+      "step": 6740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844504567207536e-05,
+      "loss": 3.0144,
+      "step": 6745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844271698297385e-05,
+      "loss": 2.8873,
+      "step": 6750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844038657904606e-05,
+      "loss": 3.0167,
+      "step": 6755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843805446037448e-05,
+      "loss": 3.0989,
+      "step": 6760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843572062704166e-05,
+      "loss": 2.9792,
+      "step": 6765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843338507913023e-05,
+      "loss": 2.9647,
+      "step": 6770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843104781672285e-05,
+      "loss": 3.0193,
+      "step": 6775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842870883990228e-05,
+      "loss": 2.9333,
+      "step": 6780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84263681487513e-05,
+      "loss": 3.026,
+      "step": 6785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842402574335274e-05,
+      "loss": 3.0688,
+      "step": 6790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842168162378958e-05,
+      "loss": 2.9119,
+      "step": 6795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841933579014478e-05,
+      "loss": 3.0554,
+      "step": 6800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841698824250134e-05,
+      "loss": 3.0366,
+      "step": 6805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841463898094241e-05,
+      "loss": 2.9278,
+      "step": 6810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841228800555111e-05,
+      "loss": 3.0301,
+      "step": 6815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84099353164107e-05,
+      "loss": 3.0218,
+      "step": 6820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840758091360445e-05,
+      "loss": 3.0057,
+      "step": 6825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84052247972157e-05,
+      "loss": 2.9843,
+      "step": 6830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840286696732785e-05,
+      "loss": 2.9932,
+      "step": 6835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840050742402438e-05,
+      "loss": 2.9691,
+      "step": 6840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839814616738881e-05,
+      "loss": 2.923,
+      "step": 6845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83957831975047e-05,
+      "loss": 2.9741,
+      "step": 6850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839341851445573e-05,
+      "loss": 2.9413,
+      "step": 6855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83910521183256e-05,
+      "loss": 3.0855,
+      "step": 6860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838868400919807e-05,
+      "loss": 2.8961,
+      "step": 6865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838631418715698e-05,
+      "loss": 2.9645,
+      "step": 6870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838394265228621e-05,
+      "loss": 2.8382,
+      "step": 6875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83815694046697e-05,
+      "loss": 2.8632,
+      "step": 6880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83791944443915e-05,
+      "loss": 2.9569,
+      "step": 6885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837681777153564e-05,
+      "loss": 2.9858,
+      "step": 6890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837443938618627e-05,
+      "loss": 3.0702,
+      "step": 6895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837205928842756e-05,
+      "loss": 2.9833,
+      "step": 6900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83696774783438e-05,
+      "loss": 2.9571,
+      "step": 6905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836729395601929e-05,
+      "loss": 2.8934,
+      "step": 6910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83649087215384e-05,
+      "loss": 2.9679,
+      "step": 6915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836252177498557e-05,
+      "loss": 2.9555,
+      "step": 6920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836013311644529e-05,
+      "loss": 3.011,
+      "step": 6925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83577427460021e-05,
+      "loss": 3.0351,
+      "step": 6930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835535066374067e-05,
+      "loss": 2.9515,
+      "step": 6935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835295686974561e-05,
+      "loss": 3.0122,
+      "step": 6940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83505613641017e-05,
+      "loss": 2.8599,
+      "step": 6945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834816414689374e-05,
+      "loss": 3.0015,
+      "step": 6950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834576521820657e-05,
+      "loss": 2.9954,
+      "step": 6955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834336457812513e-05,
+      "loss": 3.0209,
+      "step": 6960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834096222673436e-05,
+      "loss": 2.8865,
+      "step": 6965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833855816411935e-05,
+      "loss": 3.0518,
+      "step": 6970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833615239036518e-05,
+      "loss": 3.03,
+      "step": 6975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.8333744905557e-05,
+      "loss": 2.9783,
+      "step": 6980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833133570978006e-05,
+      "loss": 2.9651,
+      "step": 6985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.832892480311963e-05,
+      "loss": 2.9375,
+      "step": 6990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.832651218566105e-05,
+      "loss": 2.8591,
+      "step": 6995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.832409785748973e-05,
+      "loss": 2.8734,
+      "step": 7000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.832168181869113e-05,
+      "loss": 2.9152,
+      "step": 7005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.831926406935077e-05,
+      "loss": 2.968,
+      "step": 7010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.831684460955425e-05,
+      "loss": 3.0607,
+      "step": 7015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83144234393872e-05,
+      "loss": 2.9653,
+      "step": 7020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.831200055893535e-05,
+      "loss": 3.0468,
+      "step": 7025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.830957596828446e-05,
+      "loss": 2.9898,
+      "step": 7030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.830714966752035e-05,
+      "loss": 2.9055,
+      "step": 7035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.830472165672891e-05,
+      "loss": 2.9967,
+      "step": 7040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.830229193599607e-05,
+      "loss": 2.9789,
+      "step": 7045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.82998605054079e-05,
+      "loss": 2.9237,
+      "step": 7050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.82974273650504e-05,
+      "loss": 2.9293,
+      "step": 7055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.829499251500973e-05,
+      "loss": 2.9397,
+      "step": 7060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.82925559553721e-05,
+      "loss": 3.0192,
+      "step": 7065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.829011768622374e-05,
+      "loss": 2.9888,
+      "step": 7070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.828767770765097e-05,
+      "loss": 3.0106,
+      "step": 7075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.828523601974018e-05,
+      "loss": 2.9575,
+      "step": 7080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.828279262257775e-05,
+      "loss": 2.9538,
+      "step": 7085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.828034751625023e-05,
+      "loss": 2.8903,
+      "step": 7090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.827790070084414e-05,
+      "loss": 2.8284,
+      "step": 7095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.827545217644612e-05,
+      "loss": 2.9721,
+      "step": 7100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.827300194314284e-05,
+      "loss": 2.9166,
+      "step": 7105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827055000102101e-05,
+      "loss": 3.0298,
+      "step": 7110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826809635016744e-05,
+      "loss": 2.9313,
+      "step": 7115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826564099066901e-05,
+      "loss": 3.0524,
+      "step": 7120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826318392261262e-05,
+      "loss": 2.8795,
+      "step": 7125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826072514608524e-05,
+      "loss": 2.922,
+      "step": 7130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825826466117391e-05,
+      "loss": 2.8845,
+      "step": 7135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825580246796575e-05,
+      "loss": 2.957,
+      "step": 7140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825333856654791e-05,
+      "loss": 2.9765,
+      "step": 7145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825087295700759e-05,
+      "loss": 2.9909,
+      "step": 7150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824840563943209e-05,
+      "loss": 2.9367,
+      "step": 7155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824593661390872e-05,
+      "loss": 2.9689,
+      "step": 7160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824346588052494e-05,
+      "loss": 2.887,
+      "step": 7165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824099343936817e-05,
+      "loss": 2.9366,
+      "step": 7170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823851929052593e-05,
+      "loss": 2.9691,
+      "step": 7175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823604343408582e-05,
+      "loss": 2.8892,
+      "step": 7180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823356587013548e-05,
+      "loss": 3.0309,
+      "step": 7185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823108659876259e-05,
+      "loss": 3.0107,
+      "step": 7190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822860562005494e-05,
+      "loss": 2.9376,
+      "step": 7195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822612293410036e-05,
+      "loss": 2.9723,
+      "step": 7200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822363854098671e-05,
+      "loss": 3.0362,
+      "step": 7205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822115244080195e-05,
+      "loss": 3.0468,
+      "step": 7210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821866463363409e-05,
+      "loss": 2.9081,
+      "step": 7215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821617511957117e-05,
+      "loss": 2.9598,
+      "step": 7220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821368389870135e-05,
+      "loss": 2.9122,
+      "step": 7225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82111909711128e-05,
+      "loss": 2.96,
+      "step": 7230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820869633689375e-05,
+      "loss": 2.9337,
+      "step": 7235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820619999613256e-05,
+      "loss": 3.0059,
+      "step": 7240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820370194891755e-05,
+      "loss": 2.9548,
+      "step": 7245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820120219533715e-05,
+      "loss": 2.9703,
+      "step": 7250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819870073547988e-05,
+      "loss": 3.0115,
+      "step": 7255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819619756943426e-05,
+      "loss": 2.8936,
+      "step": 7260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81936926972889e-05,
+      "loss": 2.9666,
+      "step": 7265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819118611913251e-05,
+      "loss": 2.8757,
+      "step": 7270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818867783505378e-05,
+      "loss": 2.8782,
+      "step": 7275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818616784514151e-05,
+      "loss": 2.8864,
+      "step": 7280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818365614948455e-05,
+      "loss": 2.9551,
+      "step": 7285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818114274817183e-05,
+      "loss": 2.9773,
+      "step": 7290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817862764129229e-05,
+      "loss": 2.9974,
+      "step": 7295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817611082893498e-05,
+      "loss": 3.0194,
+      "step": 7300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817359231118901e-05,
+      "loss": 2.9216,
+      "step": 7305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817107208814351e-05,
+      "loss": 2.9338,
+      "step": 7310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816855015988768e-05,
+      "loss": 2.9547,
+      "step": 7315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816602652651084e-05,
+      "loss": 2.9247,
+      "step": 7320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816350118810229e-05,
+      "loss": 3.067,
+      "step": 7325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816097414475143e-05,
+      "loss": 2.9101,
+      "step": 7330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81584453965477e-05,
+      "loss": 3.0328,
+      "step": 7335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815591494358067e-05,
+      "loss": 3.0073,
+      "step": 7340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815338278593983e-05,
+      "loss": 2.8884,
+      "step": 7345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815084892371491e-05,
+      "loss": 2.9359,
+      "step": 7350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814831335699554e-05,
+      "loss": 3.0344,
+      "step": 7355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81457760858715e-05,
+      "loss": 2.9557,
+      "step": 7360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814323711043259e-05,
+      "loss": 2.8465,
+      "step": 7365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814069643076872e-05,
+      "loss": 2.9058,
+      "step": 7370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81381540469698e-05,
+      "loss": 2.8899,
+      "step": 7375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813560995912583e-05,
+      "loss": 2.8423,
+      "step": 7380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813306416732688e-05,
+      "loss": 3.0174,
+      "step": 7385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813051667166309e-05,
+      "loss": 2.9411,
+      "step": 7390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812796747222458e-05,
+      "loss": 2.9311,
+      "step": 7395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812541656910164e-05,
+      "loss": 2.8667,
+      "step": 7400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812286396238452e-05,
+      "loss": 2.971,
+      "step": 7405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812030965216365e-05,
+      "loss": 2.877,
+      "step": 7410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81177536385294e-05,
+      "loss": 2.9719,
+      "step": 7415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811519592157225e-05,
+      "loss": 2.9871,
+      "step": 7420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811263650138275e-05,
+      "loss": 2.9913,
+      "step": 7425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811007537805152e-05,
+      "loss": 2.875,
+      "step": 7430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810751255166921e-05,
+      "loss": 2.9748,
+      "step": 7435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810494802232653e-05,
+      "loss": 2.9459,
+      "step": 7440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810238179011426e-05,
+      "loss": 3.0935,
+      "step": 7445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809981385512325e-05,
+      "loss": 2.9589,
+      "step": 7450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809724421744442e-05,
+      "loss": 2.9744,
+      "step": 7455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80946728771687e-05,
+      "loss": 3.0387,
+      "step": 7460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809209983438714e-05,
+      "loss": 2.9314,
+      "step": 7465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80895250891908e-05,
+      "loss": 3.0286,
+      "step": 7470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808694864167086e-05,
+      "loss": 2.9924,
+      "step": 7475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808437049191848e-05,
+      "loss": 2.97,
+      "step": 7480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808179064002494e-05,
+      "loss": 2.8893,
+      "step": 7485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807920908608157e-05,
+      "loss": 2.9598,
+      "step": 7490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807662583017975e-05,
+      "loss": 2.9572,
+      "step": 7495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807404087241095e-05,
+      "loss": 2.9628,
+      "step": 7500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807145421286664e-05,
+      "loss": 2.9767,
+      "step": 7505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806886585163839e-05,
+      "loss": 2.9879,
+      "step": 7510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806627578881784e-05,
+      "loss": 2.8968,
+      "step": 7515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806368402449666e-05,
+      "loss": 2.9338,
+      "step": 7520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806109055876662e-05,
+      "loss": 2.9802,
+      "step": 7525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80584953917195e-05,
+      "loss": 3.033,
+      "step": 7530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805589852344718e-05,
+      "loss": 2.9741,
+      "step": 7535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805329995404159e-05,
+      "loss": 2.9387,
+      "step": 7540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80506996835947e-05,
+      "loss": 2.9173,
+      "step": 7545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804809771219858e-05,
+      "loss": 2.8816,
+      "step": 7550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804549403994533e-05,
+      "loss": 2.8791,
+      "step": 7555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804288866692711e-05,
+      "loss": 2.9336,
+      "step": 7560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804028159323613e-05,
+      "loss": 2.9812,
+      "step": 7565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.803767281896472e-05,
+      "loss": 3.0001,
+      "step": 7570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80350623442052e-05,
+      "loss": 2.9846,
+      "step": 7575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.803245016905e-05,
+      "loss": 3.0387,
+      "step": 7580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802983629359155e-05,
+      "loss": 2.9809,
+      "step": 7585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802722071792243e-05,
+      "loss": 3.0591,
+      "step": 7590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802460344213518e-05,
+      "loss": 2.8632,
+      "step": 7595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802198446632247e-05,
+      "loss": 2.9772,
+      "step": 7600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801936379057702e-05,
+      "loss": 2.996,
+      "step": 7605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80167414149916e-05,
+      "loss": 2.9922,
+      "step": 7610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801411733965901e-05,
+      "loss": 2.9182,
+      "step": 7615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801149156467215e-05,
+      "loss": 3.0142,
+      "step": 7620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8008864090124e-05,
+      "loss": 3.0544,
+      "step": 7625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800623491610755e-05,
+      "loss": 3.0335,
+      "step": 7630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800360404271586e-05,
+      "loss": 3.0053,
+      "step": 7635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800097147004208e-05,
+      "loss": 2.9171,
+      "step": 7640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799833719817936e-05,
+      "loss": 3.1099,
+      "step": 7645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799570122722101e-05,
+      "loss": 2.9593,
+      "step": 7650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799306355726031e-05,
+      "loss": 2.9238,
+      "step": 7655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799042418839062e-05,
+      "loss": 2.8903,
+      "step": 7660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798778312070539e-05,
+      "loss": 3.0443,
+      "step": 7665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798514035429811e-05,
+      "loss": 3.0258,
+      "step": 7670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798249588926232e-05,
+      "loss": 2.9781,
+      "step": 7675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797984972569163e-05,
+      "loss": 2.9319,
+      "step": 7680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797720186367973e-05,
+      "loss": 2.9662,
+      "step": 7685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797455230332034e-05,
+      "loss": 2.9584,
+      "step": 7690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797190104470726e-05,
+      "loss": 2.9137,
+      "step": 7695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.796924808793433e-05,
+      "loss": 2.9149,
+      "step": 7700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.796659343309546e-05,
+      "loss": 3.0281,
+      "step": 7705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.796393708028465e-05,
+      "loss": 2.9653,
+      "step": 7710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79612790295959e-05,
+      "loss": 3.0071,
+      "step": 7715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795861928112332e-05,
+      "loss": 3.0204,
+      "step": 7720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795595783496105e-05,
+      "loss": 2.9882,
+      "step": 7725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795329469120331e-05,
+      "loss": 2.8945,
+      "step": 7730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795062984994439e-05,
+      "loss": 2.9431,
+      "step": 7735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794796331127858e-05,
+      "loss": 2.9633,
+      "step": 7740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794529507530032e-05,
+      "loss": 2.8963,
+      "step": 7745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794262514210402e-05,
+      "loss": 2.9396,
+      "step": 7750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793995351178423e-05,
+      "loss": 2.8672,
+      "step": 7755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793728018443552e-05,
+      "loss": 2.9456,
+      "step": 7760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793460516015249e-05,
+      "loss": 2.8929,
+      "step": 7765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793192843902985e-05,
+      "loss": 2.9892,
+      "step": 7770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792925002116237e-05,
+      "loss": 3.0094,
+      "step": 7775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792656990664486e-05,
+      "loss": 2.9976,
+      "step": 7780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792388809557218e-05,
+      "loss": 3.0277,
+      "step": 7785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792120458803926e-05,
+      "loss": 2.9121,
+      "step": 7790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79185193841411e-05,
+      "loss": 2.9815,
+      "step": 7795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.791583248397278e-05,
+      "loss": 2.9514,
+      "step": 7800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.791314388762936e-05,
+      "loss": 2.9456,
+      "step": 7805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.791045359520605e-05,
+      "loss": 3.0373,
+      "step": 7810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790776160679809e-05,
+      "loss": 2.8422,
+      "step": 7815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790506792250075e-05,
+      "loss": 2.9316,
+      "step": 7820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79023725424094e-05,
+      "loss": 3.0354,
+      "step": 7825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.789967546661945e-05,
+      "loss": 2.9397,
+      "step": 7830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.789697669522638e-05,
+      "loss": 2.9549,
+      "step": 7835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78942762283257e-05,
+      "loss": 2.9547,
+      "step": 7840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.789157406601302e-05,
+      "loss": 3.0462,
+      "step": 7845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.7888870208384e-05,
+      "loss": 2.9536,
+      "step": 7850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.788616465553436e-05,
+      "loss": 2.9392,
+      "step": 7855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.788345740755984e-05,
+      "loss": 2.9478,
+      "step": 7860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78807484645563e-05,
+      "loss": 3.0034,
+      "step": 7865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.787803782661964e-05,
+      "loss": 2.9693,
+      "step": 7870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.787532549384582e-05,
+      "loss": 3.0075,
+      "step": 7875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.787261146633082e-05,
+      "loss": 2.9511,
+      "step": 7880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.786989574417074e-05,
+      "loss": 2.9665,
+      "step": 7885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78671783274617e-05,
+      "loss": 2.9277,
+      "step": 7890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78644592162999e-05,
+      "loss": 3.0654,
+      "step": 7895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.786173841078161e-05,
+      "loss": 2.9504,
+      "step": 7900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78590159110031e-05,
+      "loss": 2.9508,
+      "step": 7905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.785629171706079e-05,
+      "loss": 2.8798,
+      "step": 7910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78535658290511e-05,
+      "loss": 3.0353,
+      "step": 7915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78508382470705e-05,
+      "loss": 2.9072,
+      "step": 7920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.784810897121558e-05,
+      "loss": 3.0761,
+      "step": 7925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.784537800158295e-05,
+      "loss": 2.9893,
+      "step": 7930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.784264533826926e-05,
+      "loss": 2.9484,
+      "step": 7935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.783991098137124e-05,
+      "loss": 2.9013,
+      "step": 7940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783717493098572e-05,
+      "loss": 2.9159,
+      "step": 7945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783443718720953e-05,
+      "loss": 2.9897,
+      "step": 7950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783169775013959e-05,
+      "loss": 2.9631,
+      "step": 7955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782895661987285e-05,
+      "loss": 2.9339,
+      "step": 7960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782621379650636e-05,
+      "loss": 3.0654,
+      "step": 7965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782346928013724e-05,
+      "loss": 2.9271,
+      "step": 7970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78207230708626e-05,
+      "loss": 3.0283,
+      "step": 7975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.781797516877969e-05,
+      "loss": 2.9928,
+      "step": 7980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.781522557398574e-05,
+      "loss": 2.9196,
+      "step": 7985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.781247428657814e-05,
+      "loss": 2.9982,
+      "step": 7990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780972130665424e-05,
+      "loss": 2.873,
+      "step": 7995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78069666343115e-05,
+      "loss": 2.9977,
+      "step": 8000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780421026964744e-05,
+      "loss": 2.9431,
+      "step": 8005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780145221275962e-05,
+      "loss": 2.9325,
+      "step": 8010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77986924637457e-05,
+      "loss": 3.0197,
+      "step": 8015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779593102270334e-05,
+      "loss": 2.9715,
+      "step": 8020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779316788973032e-05,
+      "loss": 2.9514,
+      "step": 8025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779040306492444e-05,
+      "loss": 3.0012,
+      "step": 8030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778763654838356e-05,
+      "loss": 2.9619,
+      "step": 8035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778486834020563e-05,
+      "loss": 3.0213,
+      "step": 8040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778209844048865e-05,
+      "loss": 2.9658,
+      "step": 8045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777932684933064e-05,
+      "loss": 2.8941,
+      "step": 8050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777655356682974e-05,
+      "loss": 2.9866,
+      "step": 8055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77737785930841e-05,
+      "loss": 2.9237,
+      "step": 8060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777100192819198e-05,
+      "loss": 2.9707,
+      "step": 8065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776822357225164e-05,
+      "loss": 2.9717,
+      "step": 8070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776544352536145e-05,
+      "loss": 2.9046,
+      "step": 8075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776266178761981e-05,
+      "loss": 3.0115,
+      "step": 8080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775987835912522e-05,
+      "loss": 3.049,
+      "step": 8085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775709323997614e-05,
+      "loss": 2.9748,
+      "step": 8090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775430643027125e-05,
+      "loss": 3.0229,
+      "step": 8095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775151793010915e-05,
+      "loss": 2.9935,
+      "step": 8100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774872773958856e-05,
+      "loss": 2.9263,
+      "step": 8105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774593585880825e-05,
+      "loss": 2.9356,
+      "step": 8110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774314228786704e-05,
+      "loss": 2.9471,
+      "step": 8115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774034702686384e-05,
+      "loss": 3.0001,
+      "step": 8120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.773755007589758e-05,
+      "loss": 3.013,
+      "step": 8125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.773475143506727e-05,
+      "loss": 2.8706,
+      "step": 8130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7731951104472e-05,
+      "loss": 3.0021,
+      "step": 8135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772914908421089e-05,
+      "loss": 2.9826,
+      "step": 8140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772634537438312e-05,
+      "loss": 2.9539,
+      "step": 8145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772353997508794e-05,
+      "loss": 2.9705,
+      "step": 8150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772073288642466e-05,
+      "loss": 2.8973,
+      "step": 8155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771792410849265e-05,
+      "loss": 3.003,
+      "step": 8160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771511364139134e-05,
+      "loss": 2.9352,
+      "step": 8165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771230148522024e-05,
+      "loss": 2.9567,
+      "step": 8170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770948764007884e-05,
+      "loss": 2.9857,
+      "step": 8175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77066721060668e-05,
+      "loss": 2.9635,
+      "step": 8180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770385488328376e-05,
+      "loss": 3.0756,
+      "step": 8185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770103597182947e-05,
+      "loss": 3.1228,
+      "step": 8190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769821537180368e-05,
+      "loss": 2.9637,
+      "step": 8195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769539308330628e-05,
+      "loss": 2.9166,
+      "step": 8200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769256910643713e-05,
+      "loss": 2.9898,
+      "step": 8205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768974344129624e-05,
+      "loss": 2.9628,
+      "step": 8210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768691608798363e-05,
+      "loss": 2.9218,
+      "step": 8215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768408704659935e-05,
+      "loss": 2.9385,
+      "step": 8220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768125631724358e-05,
+      "loss": 2.9314,
+      "step": 8225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767842390001651e-05,
+      "loss": 2.8971,
+      "step": 8230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767558979501841e-05,
+      "loss": 2.9433,
+      "step": 8235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767275400234961e-05,
+      "loss": 2.9974,
+      "step": 8240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766991652211049e-05,
+      "loss": 2.9576,
+      "step": 8245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766707735440147e-05,
+      "loss": 2.926,
+      "step": 8250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766423649932307e-05,
+      "loss": 2.9749,
+      "step": 8255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766139395697589e-05,
+      "loss": 2.959,
+      "step": 8260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765854972746051e-05,
+      "loss": 3.0066,
+      "step": 8265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765570381087763e-05,
+      "loss": 2.9617,
+      "step": 8270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765285620732798e-05,
+      "loss": 2.9301,
+      "step": 8275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765000691691237e-05,
+      "loss": 3.0191,
+      "step": 8280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.764715593973166e-05,
+      "loss": 3.1297,
+      "step": 8285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.764430327588679e-05,
+      "loss": 3.014,
+      "step": 8290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.764144892547871e-05,
+      "loss": 2.9671,
+      "step": 8295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.763859288860848e-05,
+      "loss": 2.9885,
+      "step": 8300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76357351653772e-05,
+      "loss": 2.9024,
+      "step": 8305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.763287575588604e-05,
+      "loss": 2.8125,
+      "step": 8310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76300146602362e-05,
+      "loss": 3.0133,
+      "step": 8315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.762715187852898e-05,
+      "loss": 3.0114,
+      "step": 8320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76242874108657e-05,
+      "loss": 3.0558,
+      "step": 8325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.762142125734776e-05,
+      "loss": 3.0091,
+      "step": 8330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761855341807664e-05,
+      "loss": 2.9512,
+      "step": 8335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761568389315385e-05,
+      "loss": 2.9426,
+      "step": 8340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761281268268097e-05,
+      "loss": 2.9554,
+      "step": 8345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760993978675961e-05,
+      "loss": 2.9011,
+      "step": 8350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76070652054915e-05,
+      "loss": 2.9928,
+      "step": 8355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76041889389784e-05,
+      "loss": 3.0143,
+      "step": 8360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76013109873221e-05,
+      "loss": 2.9499,
+      "step": 8365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.759843135062451e-05,
+      "loss": 3.0191,
+      "step": 8370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.759555002898754e-05,
+      "loss": 2.9292,
+      "step": 8375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75926670225132e-05,
+      "loss": 2.9885,
+      "step": 8380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758978233130353e-05,
+      "loss": 2.9717,
+      "step": 8385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758689595546067e-05,
+      "loss": 3.0451,
+      "step": 8390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758400789508677e-05,
+      "loss": 3.0046,
+      "step": 8395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758111815028408e-05,
+      "loss": 3.0546,
+      "step": 8400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757822672115489e-05,
+      "loss": 3.0443,
+      "step": 8405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757533360780154e-05,
+      "loss": 3.0135,
+      "step": 8410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757243881032647e-05,
+      "loss": 2.9752,
+      "step": 8415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756954232883214e-05,
+      "loss": 2.9429,
+      "step": 8420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756664416342108e-05,
+      "loss": 2.9997,
+      "step": 8425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756374431419589e-05,
+      "loss": 2.916,
+      "step": 8430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756084278125922e-05,
+      "loss": 3.0725,
+      "step": 8435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.755793956471378e-05,
+      "loss": 2.9575,
+      "step": 8440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.755503466466233e-05,
+      "loss": 2.8759,
+      "step": 8445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75521280812077e-05,
+      "loss": 2.9587,
+      "step": 8450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754921981445282e-05,
+      "loss": 2.9979,
+      "step": 8455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75463098645006e-05,
+      "loss": 2.8715,
+      "step": 8460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754339823145404e-05,
+      "loss": 3.0121,
+      "step": 8465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754048491541626e-05,
+      "loss": 2.9338,
+      "step": 8470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753756991649033e-05,
+      "loss": 2.7856,
+      "step": 8475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753465323477948e-05,
+      "loss": 2.9189,
+      "step": 8480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753173487038695e-05,
+      "loss": 3.0258,
+      "step": 8485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752881482341602e-05,
+      "loss": 3.017,
+      "step": 8490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75258930939701e-05,
+      "loss": 2.9344,
+      "step": 8495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752296968215259e-05,
+      "loss": 2.8305,
+      "step": 8500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752004458806698e-05,
+      "loss": 3.0371,
+      "step": 8505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75171178118168e-05,
+      "loss": 2.8977,
+      "step": 8510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75141893535057e-05,
+      "loss": 3.0173,
+      "step": 8515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75112592132373e-05,
+      "loss": 2.9824,
+      "step": 8520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.750832739111533e-05,
+      "loss": 3.0489,
+      "step": 8525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75053938872436e-05,
+      "loss": 2.9733,
+      "step": 8530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.750245870172592e-05,
+      "loss": 2.95,
+      "step": 8535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749952183466623e-05,
+      "loss": 2.9542,
+      "step": 8540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749658328616846e-05,
+      "loss": 2.9613,
+      "step": 8545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749364305633665e-05,
+      "loss": 3.0176,
+      "step": 8550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749070114527487e-05,
+      "loss": 3.0347,
+      "step": 8555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748775755308726e-05,
+      "loss": 2.8765,
+      "step": 8560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748481227987805e-05,
+      "loss": 3.0245,
+      "step": 8565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748186532575145e-05,
+      "loss": 2.8601,
+      "step": 8570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747891669081182e-05,
+      "loss": 3.0169,
+      "step": 8575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747596637516352e-05,
+      "loss": 3.083,
+      "step": 8580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747301437891099e-05,
+      "loss": 3.002,
+      "step": 8585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747006070215874e-05,
+      "loss": 2.9516,
+      "step": 8590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746710534501131e-05,
+      "loss": 2.9885,
+      "step": 8595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746414830757334e-05,
+      "loss": 2.977,
+      "step": 8600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746118958994947e-05,
+      "loss": 3.0405,
+      "step": 8605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.745822919224448e-05,
+      "loss": 2.9441,
+      "step": 8610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.745526711456312e-05,
+      "loss": 2.9442,
+      "step": 8615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.745230335701028e-05,
+      "loss": 2.9378,
+      "step": 8620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744933791969087e-05,
+      "loss": 2.9446,
+      "step": 8625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744637080270983e-05,
+      "loss": 2.9507,
+      "step": 8630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744340200617223e-05,
+      "loss": 3.0039,
+      "step": 8635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744043153018316e-05,
+      "loss": 2.9706,
+      "step": 8640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743745937484775e-05,
+      "loss": 2.8438,
+      "step": 8645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743448554027124e-05,
+      "loss": 3.0668,
+      "step": 8650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743151002655886e-05,
+      "loss": 2.9735,
+      "step": 8655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742853283381599e-05,
+      "loss": 2.9624,
+      "step": 8660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742555396214799e-05,
+      "loss": 3.0049,
+      "step": 8665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742257341166031e-05,
+      "loss": 3.0399,
+      "step": 8670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.741959118245847e-05,
+      "loss": 3.0715,
+      "step": 8675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.741660727464803e-05,
+      "loss": 2.9858,
+      "step": 8680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.741362168833461e-05,
+      "loss": 2.9645,
+      "step": 8685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.741063442362391e-05,
+      "loss": 2.8697,
+      "step": 8690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.740764548062167e-05,
+      "loss": 2.9975,
+      "step": 8695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.740465485943369e-05,
+      "loss": 2.9161,
+      "step": 8700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.740166256016587e-05,
+      "loss": 2.9546,
+      "step": 8705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.739866858292408e-05,
+      "loss": 2.9863,
+      "step": 8710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.739567292781435e-05,
+      "loss": 3.0704,
+      "step": 8715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.739267559494269e-05,
+      "loss": 2.9609,
+      "step": 8720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.738967658441523e-05,
+      "loss": 3.0318,
+      "step": 8725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.738667589633812e-05,
+      "loss": 2.8748,
+      "step": 8730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.738367353081758e-05,
+      "loss": 2.8573,
+      "step": 8735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.73806694879599e-05,
+      "loss": 2.9681,
+      "step": 8740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.737766376787138e-05,
+      "loss": 2.9129,
+      "step": 8745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.737465637065847e-05,
+      "loss": 2.9136,
+      "step": 8750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.737164729642763e-05,
+      "loss": 3.0414,
+      "step": 8755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.736863654528535e-05,
+      "loss": 2.9738,
+      "step": 8760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.73656241173382e-05,
+      "loss": 2.974,
+      "step": 8765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.736261001269286e-05,
+      "loss": 2.9452,
+      "step": 8770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.735959423145598e-05,
+      "loss": 2.9042,
+      "step": 8775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735657677373436e-05,
+      "loss": 3.0859,
+      "step": 8780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735355763963476e-05,
+      "loss": 3.0262,
+      "step": 8785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735053682926411e-05,
+      "loss": 3.0204,
+      "step": 8790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734751434272931e-05,
+      "loss": 3.0515,
+      "step": 8795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734449018013737e-05,
+      "loss": 2.8391,
+      "step": 8800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734146434159533e-05,
+      "loss": 2.9363,
+      "step": 8805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73384368272103e-05,
+      "loss": 2.9602,
+      "step": 8810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733540763708948e-05,
+      "loss": 2.8481,
+      "step": 8815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733237677134007e-05,
+      "loss": 3.0814,
+      "step": 8820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.732934423006936e-05,
+      "loss": 2.9558,
+      "step": 8825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.732631001338472e-05,
+      "loss": 2.9499,
+      "step": 8830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.732327412139354e-05,
+      "loss": 2.8297,
+      "step": 8835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73202365542033e-05,
+      "loss": 2.872,
+      "step": 8840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.731719731192151e-05,
+      "loss": 2.9609,
+      "step": 8845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.731415639465578e-05,
+      "loss": 2.9715,
+      "step": 8850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.731111380251375e-05,
+      "loss": 2.9491,
+      "step": 8855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73080695356031e-05,
+      "loss": 2.9529,
+      "step": 8860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.730502359403161e-05,
+      "loss": 2.9417,
+      "step": 8865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.730197597790712e-05,
+      "loss": 2.9552,
+      "step": 8870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729892668733749e-05,
+      "loss": 2.9218,
+      "step": 8875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729587572243066e-05,
+      "loss": 3.0138,
+      "step": 8880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729282308329467e-05,
+      "loss": 2.8715,
+      "step": 8885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728976877003752e-05,
+      "loss": 2.94,
+      "step": 8890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728671278276737e-05,
+      "loss": 3.0322,
+      "step": 8895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728365512159241e-05,
+      "loss": 3.007,
+      "step": 8900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728059578662085e-05,
+      "loss": 2.9823,
+      "step": 8905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727753477796099e-05,
+      "loss": 3.0224,
+      "step": 8910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72744720957212e-05,
+      "loss": 3.0404,
+      "step": 8915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727140774000988e-05,
+      "loss": 3.0108,
+      "step": 8920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.726834171093553e-05,
+      "loss": 2.987,
+      "step": 8925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.726527400860666e-05,
+      "loss": 2.9747,
+      "step": 8930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.726220463313187e-05,
+      "loss": 3.0257,
+      "step": 8935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725913358461982e-05,
+      "loss": 2.9164,
+      "step": 8940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725606086317922e-05,
+      "loss": 3.055,
+      "step": 8945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725298646891884e-05,
+      "loss": 2.988,
+      "step": 8950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72499104019475e-05,
+      "loss": 2.9424,
+      "step": 8955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72468326623741e-05,
+      "loss": 3.0335,
+      "step": 8960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.724375325030762e-05,
+      "loss": 3.0183,
+      "step": 8965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.724067216585702e-05,
+      "loss": 2.9692,
+      "step": 8970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723758940913137e-05,
+      "loss": 3.0087,
+      "step": 8975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723450498023983e-05,
+      "loss": 2.9842,
+      "step": 8980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723141887929155e-05,
+      "loss": 3.0085,
+      "step": 8985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.722833110639581e-05,
+      "loss": 2.9972,
+      "step": 8990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.722524166166189e-05,
+      "loss": 2.9581,
+      "step": 8995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.722215054519916e-05,
+      "loss": 2.9263,
+      "step": 9000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721905775711705e-05,
+      "loss": 2.9455,
+      "step": 9005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721596329752502e-05,
+      "loss": 2.8994,
+      "step": 9010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721286716653264e-05,
+      "loss": 3.017,
+      "step": 9015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72097693642495e-05,
+      "loss": 3.0189,
+      "step": 9020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.720666989078524e-05,
+      "loss": 2.9856,
+      "step": 9025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72035687462496e-05,
+      "loss": 2.9524,
+      "step": 9030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.720046593075233e-05,
+      "loss": 2.9905,
+      "step": 9035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719736144440332e-05,
+      "loss": 2.9371,
+      "step": 9040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719425528731241e-05,
+      "loss": 3.0667,
+      "step": 9045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71911474595896e-05,
+      "loss": 3.0182,
+      "step": 9050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718803796134487e-05,
+      "loss": 2.925,
+      "step": 9055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718492679268831e-05,
+      "loss": 2.9487,
+      "step": 9060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718181395373005e-05,
+      "loss": 2.7889,
+      "step": 9065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717869944458028e-05,
+      "loss": 3.0022,
+      "step": 9070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717558326534924e-05,
+      "loss": 3.0069,
+      "step": 9075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717246541614726e-05,
+      "loss": 2.9606,
+      "step": 9080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71693458970847e-05,
+      "loss": 2.886,
+      "step": 9085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.7166224708272e-05,
+      "loss": 2.9256,
+      "step": 9090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71631018498196e-05,
+      "loss": 2.9454,
+      "step": 9095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71599773218381e-05,
+      "loss": 2.8807,
+      "step": 9100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.715685112443809e-05,
+      "loss": 2.868,
+      "step": 9105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.715372325773022e-05,
+      "loss": 2.9598,
+      "step": 9110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.715059372182525e-05,
+      "loss": 2.9888,
+      "step": 9115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714746251683391e-05,
+      "loss": 2.9763,
+      "step": 9120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714432964286708e-05,
+      "loss": 2.9959,
+      "step": 9125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714119510003564e-05,
+      "loss": 2.9421,
+      "step": 9130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.713805888845057e-05,
+      "loss": 3.0319,
+      "step": 9135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.713492100822286e-05,
+      "loss": 3.0024,
+      "step": 9140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.713178145946362e-05,
+      "loss": 3.0281,
+      "step": 9145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712864024228398e-05,
+      "loss": 2.9712,
+      "step": 9150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712549735679513e-05,
+      "loss": 3.0299,
+      "step": 9155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712235280310831e-05,
+      "loss": 3.0322,
+      "step": 9160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711920658133485e-05,
+      "loss": 2.9277,
+      "step": 9165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711605869158612e-05,
+      "loss": 2.9234,
+      "step": 9170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711290913397356e-05,
+      "loss": 3.0221,
+      "step": 9175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710975790860866e-05,
+      "loss": 3.0206,
+      "step": 9180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710660501560295e-05,
+      "loss": 3.0004,
+      "step": 9185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710345045506807e-05,
+      "loss": 2.8857,
+      "step": 9190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710029422711567e-05,
+      "loss": 2.9069,
+      "step": 9195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.709713633185749e-05,
+      "loss": 2.9198,
+      "step": 9200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.709397676940528e-05,
+      "loss": 2.9605,
+      "step": 9205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.709081553987094e-05,
+      "loss": 2.9007,
+      "step": 9210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.708765264336634e-05,
+      "loss": 3.0278,
+      "step": 9215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.708448808000345e-05,
+      "loss": 2.9976,
+      "step": 9220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.708132184989428e-05,
+      "loss": 2.9532,
+      "step": 9225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707815395315094e-05,
+      "loss": 2.9386,
+      "step": 9230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707498438988556e-05,
+      "loss": 2.8927,
+      "step": 9235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707181316021033e-05,
+      "loss": 2.9767,
+      "step": 9240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706864026423752e-05,
+      "loss": 2.9672,
+      "step": 9245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706546570207945e-05,
+      "loss": 2.9075,
+      "step": 9250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706228947384846e-05,
+      "loss": 2.9265,
+      "step": 9255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705911157965702e-05,
+      "loss": 2.8648,
+      "step": 9260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705593201961762e-05,
+      "loss": 2.9375,
+      "step": 9265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705275079384282e-05,
+      "loss": 3.0235,
+      "step": 9270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.704956790244523e-05,
+      "loss": 2.9941,
+      "step": 9275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70463833455375e-05,
+      "loss": 2.9718,
+      "step": 9280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.704319712323238e-05,
+      "loss": 3.0,
+      "step": 9285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.704000923564265e-05,
+      "loss": 3.0631,
+      "step": 9290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703681968288116e-05,
+      "loss": 2.8731,
+      "step": 9295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703362846506084e-05,
+      "loss": 2.8633,
+      "step": 9300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703043558229462e-05,
+      "loss": 3.019,
+      "step": 9305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.702724103469554e-05,
+      "loss": 2.9079,
+      "step": 9310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70240448223767e-05,
+      "loss": 2.9738,
+      "step": 9315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.702084694545122e-05,
+      "loss": 2.9891,
+      "step": 9320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70176474040323e-05,
+      "loss": 2.9436,
+      "step": 9325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70144461982332e-05,
+      "loss": 2.9302,
+      "step": 9330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.701124332816728e-05,
+      "loss": 2.9455,
+      "step": 9335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700803879394788e-05,
+      "loss": 2.9513,
+      "step": 9340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700483259568845e-05,
+      "loss": 2.8978,
+      "step": 9345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700162473350247e-05,
+      "loss": 3.0098,
+      "step": 9350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69984152075035e-05,
+      "loss": 2.9373,
+      "step": 9355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.699520401780517e-05,
+      "loss": 2.9497,
+      "step": 9360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.699199116452115e-05,
+      "loss": 2.9808,
+      "step": 9365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.698877664776518e-05,
+      "loss": 3.0079,
+      "step": 9370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.698556046765102e-05,
+      "loss": 2.8821,
+      "step": 9375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.698234262429253e-05,
+      "loss": 3.0152,
+      "step": 9380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697912311780364e-05,
+      "loss": 2.9175,
+      "step": 9385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697590194829831e-05,
+      "loss": 2.8899,
+      "step": 9390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697267911589054e-05,
+      "loss": 2.8588,
+      "step": 9395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696945462069444e-05,
+      "loss": 3.0438,
+      "step": 9400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696622846282417e-05,
+      "loss": 2.9339,
+      "step": 9405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696300064239388e-05,
+      "loss": 3.0437,
+      "step": 9410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69597711595179e-05,
+      "loss": 2.8604,
+      "step": 9415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.695654001431049e-05,
+      "loss": 2.9244,
+      "step": 9420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.695330720688606e-05,
+      "loss": 3.0582,
+      "step": 9425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.695007273735904e-05,
+      "loss": 3.0318,
+      "step": 9430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694683660584393e-05,
+      "loss": 2.9843,
+      "step": 9435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694359881245529e-05,
+      "loss": 3.0493,
+      "step": 9440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694035935730772e-05,
+      "loss": 2.9449,
+      "step": 9445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.693711824051593e-05,
+      "loss": 2.929,
+      "step": 9450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.693387546219459e-05,
+      "loss": 2.8788,
+      "step": 9455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.693063102245855e-05,
+      "loss": 2.967,
+      "step": 9460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692738492142263e-05,
+      "loss": 2.9047,
+      "step": 9465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692413715920176e-05,
+      "loss": 2.875,
+      "step": 9470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692088773591088e-05,
+      "loss": 2.934,
+      "step": 9475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.691763665166503e-05,
+      "loss": 2.9567,
+      "step": 9480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69143839065793e-05,
+      "loss": 2.9832,
+      "step": 9485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.691112950076884e-05,
+      "loss": 2.9951,
+      "step": 9490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690787343434884e-05,
+      "loss": 2.9899,
+      "step": 9495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690461570743456e-05,
+      "loss": 2.9628,
+      "step": 9500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690135632014132e-05,
+      "loss": 2.9444,
+      "step": 9505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.68980952725845e-05,
+      "loss": 2.9371,
+      "step": 9510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.689483256487957e-05,
+      "loss": 3.0347,
+      "step": 9515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.689156819714198e-05,
+      "loss": 2.9111,
+      "step": 9520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.68883021694873e-05,
+      "loss": 2.9489,
+      "step": 9525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.688503448203117e-05,
+      "loss": 2.9021,
+      "step": 9530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.688176513488923e-05,
+      "loss": 2.9818,
+      "step": 9535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.687849412817721e-05,
+      "loss": 2.9354,
+      "step": 9540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.687522146201093e-05,
+      "loss": 2.9832,
+      "step": 9545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.687194713650623e-05,
+      "loss": 2.9587,
+      "step": 9550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.686867115177899e-05,
+      "loss": 2.9805,
+      "step": 9555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.686539350794521e-05,
+      "loss": 3.0833,
+      "step": 9560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.68621142051209e-05,
+      "loss": 2.9232,
+      "step": 9565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.685883324342215e-05,
+      "loss": 2.9124,
+      "step": 9570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.685555062296512e-05,
+      "loss": 2.9769,
+      "step": 9575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.685226634386598e-05,
+      "loss": 3.0218,
+      "step": 9580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.684898040624099e-05,
+      "loss": 2.9796,
+      "step": 9585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.68456928102065e-05,
+      "loss": 3.0088,
+      "step": 9590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.684240355587886e-05,
+      "loss": 2.9734,
+      "step": 9595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.683911264337451e-05,
+      "loss": 3.0035,
+      "step": 9600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.683582007280997e-05,
+      "loss": 2.9451,
+      "step": 9605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.683252584430176e-05,
+      "loss": 2.9464,
+      "step": 9610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.682922995796652e-05,
+      "loss": 2.8908,
+      "step": 9615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68259324139209e-05,
+      "loss": 2.8892,
+      "step": 9620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.682263321228164e-05,
+      "loss": 2.8309,
+      "step": 9625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.681933235316553e-05,
+      "loss": 2.9009,
+      "step": 9630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.681602983668942e-05,
+      "loss": 2.9438,
+      "step": 9635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.681272566297022e-05,
+      "loss": 2.9529,
+      "step": 9640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680941983212488e-05,
+      "loss": 2.9011,
+      "step": 9645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680611234427042e-05,
+      "loss": 2.9112,
+      "step": 9650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680280319952393e-05,
+      "loss": 2.9349,
+      "step": 9655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.679949239800255e-05,
+      "loss": 2.8398,
+      "step": 9660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.679617993982348e-05,
+      "loss": 2.9282,
+      "step": 9665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6792865825104e-05,
+      "loss": 3.0203,
+      "step": 9670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.678955005396139e-05,
+      "loss": 2.9769,
+      "step": 9675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.678623262651303e-05,
+      "loss": 3.0246,
+      "step": 9680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.678291354287638e-05,
+      "loss": 3.0238,
+      "step": 9685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.677959280316891e-05,
+      "loss": 2.9675,
+      "step": 9690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.677627040750819e-05,
+      "loss": 2.9994,
+      "step": 9695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67729463560118e-05,
+      "loss": 2.9455,
+      "step": 9700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676962064879743e-05,
+      "loss": 2.9237,
+      "step": 9705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676629328598281e-05,
+      "loss": 2.9101,
+      "step": 9710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676296426768572e-05,
+      "loss": 2.9768,
+      "step": 9715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6759633594024e-05,
+      "loss": 2.9144,
+      "step": 9720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675630126511555e-05,
+      "loss": 2.9053,
+      "step": 9725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675296728107833e-05,
+      "loss": 2.9538,
+      "step": 9730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.674963164203036e-05,
+      "loss": 2.9485,
+      "step": 9735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.674629434808975e-05,
+      "loss": 2.8457,
+      "step": 9740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67429553993746e-05,
+      "loss": 3.0441,
+      "step": 9745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673961479600311e-05,
+      "loss": 2.9602,
+      "step": 9750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673627253809354e-05,
+      "loss": 2.9475,
+      "step": 9755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673292862576422e-05,
+      "loss": 2.8889,
+      "step": 9760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.672958305913349e-05,
+      "loss": 3.0064,
+      "step": 9765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67262358383198e-05,
+      "loss": 2.9799,
+      "step": 9770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.672288696344164e-05,
+      "loss": 2.8744,
+      "step": 9775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671953643461754e-05,
+      "loss": 2.8636,
+      "step": 9780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671618425196613e-05,
+      "loss": 2.9084,
+      "step": 9785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671283041560605e-05,
+      "loss": 2.9196,
+      "step": 9790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670947492565602e-05,
+      "loss": 2.8358,
+      "step": 9795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670611778223486e-05,
+      "loss": 2.9748,
+      "step": 9800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670275898546138e-05,
+      "loss": 2.9553,
+      "step": 9805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669939853545447e-05,
+      "loss": 2.9789,
+      "step": 9810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66960364323331e-05,
+      "loss": 2.9226,
+      "step": 9815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669267267621629e-05,
+      "loss": 2.8723,
+      "step": 9820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.668930726722311e-05,
+      "loss": 3.0605,
+      "step": 9825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.668594020547268e-05,
+      "loss": 2.8778,
+      "step": 9830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.668257149108421e-05,
+      "loss": 2.9514,
+      "step": 9835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.667920112417694e-05,
+      "loss": 2.9375,
+      "step": 9840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.667582910487016e-05,
+      "loss": 2.9074,
+      "step": 9845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.667245543328329e-05,
+      "loss": 2.9355,
+      "step": 9850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66690801095357e-05,
+      "loss": 2.875,
+      "step": 9855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.666570313374688e-05,
+      "loss": 2.9024,
+      "step": 9860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66623245060364e-05,
+      "loss": 3.036,
+      "step": 9865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665894422652384e-05,
+      "loss": 2.8108,
+      "step": 9870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665556229532886e-05,
+      "loss": 2.994,
+      "step": 9875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665217871257117e-05,
+      "loss": 2.8767,
+      "step": 9880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.664879347837057e-05,
+      "loss": 2.9789,
+      "step": 9885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.664540659284688e-05,
+      "loss": 3.0205,
+      "step": 9890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.664201805611998e-05,
+      "loss": 2.9774,
+      "step": 9895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.663862786830987e-05,
+      "loss": 2.9605,
+      "step": 9900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.663523602953648e-05,
+      "loss": 2.897,
+      "step": 9905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.663184253991994e-05,
+      "loss": 2.9897,
+      "step": 9910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662844739958035e-05,
+      "loss": 2.963,
+      "step": 9915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662505060863792e-05,
+      "loss": 2.923,
+      "step": 9920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662165216721286e-05,
+      "loss": 2.9088,
+      "step": 9925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661825207542548e-05,
+      "loss": 2.9587,
+      "step": 9930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661485033339615e-05,
+      "loss": 2.9534,
+      "step": 9935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661144694124529e-05,
+      "loss": 2.9971,
+      "step": 9940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660804189909339e-05,
+      "loss": 2.9327,
+      "step": 9945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660463520706094e-05,
+      "loss": 3.0014,
+      "step": 9950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660122686526856e-05,
+      "loss": 3.0016,
+      "step": 9955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.659781687383691e-05,
+      "loss": 2.9176,
+      "step": 9960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.659440523288669e-05,
+      "loss": 2.8411,
+      "step": 9965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.659099194253868e-05,
+      "loss": 2.9729,
+      "step": 9970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.65875770029137e-05,
+      "loss": 2.9882,
+      "step": 9975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.658416041413263e-05,
+      "loss": 2.97,
+      "step": 9980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.658074217631643e-05,
+      "loss": 2.9697,
+      "step": 9985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.657732228958609e-05,
+      "loss": 2.9642,
+      "step": 9990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.657390075406265e-05,
+      "loss": 2.9173,
+      "step": 9995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.657047756986728e-05,
+      "loss": 3.0046,
+      "step": 10000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.65670527371211e-05,
+      "loss": 2.9397,
+      "step": 10005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.65636262559454e-05,
+      "loss": 2.9097,
+      "step": 10010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.656019812646144e-05,
+      "loss": 2.886,
+      "step": 10015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655676834879057e-05,
+      "loss": 2.9873,
+      "step": 10020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655333692305423e-05,
+      "loss": 2.9274,
+      "step": 10025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.654990384937386e-05,
+      "loss": 3.0064,
+      "step": 10030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6546469127871e-05,
+      "loss": 3.0218,
+      "step": 10035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.654303275866724e-05,
+      "loss": 2.9937,
+      "step": 10040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.653959474188422e-05,
+      "loss": 2.903,
+      "step": 10045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.653615507764364e-05,
+      "loss": 2.9818,
+      "step": 10050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.653271376606725e-05,
+      "loss": 3.0299,
+      "step": 10055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.65292708072769e-05,
+      "loss": 2.9274,
+      "step": 10060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.652582620139444e-05,
+      "loss": 2.9533,
+      "step": 10065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.652237994854182e-05,
+      "loss": 2.9561,
+      "step": 10070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651893204884102e-05,
+      "loss": 3.0066,
+      "step": 10075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651548250241411e-05,
+      "loss": 2.8202,
+      "step": 10080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651203130938319e-05,
+      "loss": 2.8828,
+      "step": 10085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.650857846987043e-05,
+      "loss": 3.0235,
+      "step": 10090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.650512398399806e-05,
+      "loss": 2.9219,
+      "step": 10095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.650166785188836e-05,
+      "loss": 2.9311,
+      "step": 10100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649821007366369e-05,
+      "loss": 2.9827,
+      "step": 10105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649475064944643e-05,
+      "loss": 2.9665,
+      "step": 10110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649128957935905e-05,
+      "loss": 2.9635,
+      "step": 10115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.648782686352406e-05,
+      "loss": 2.9078,
+      "step": 10120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.648436250206406e-05,
+      "loss": 2.982,
+      "step": 10125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.648089649510166e-05,
+      "loss": 2.9848,
+      "step": 10130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.647742884275957e-05,
+      "loss": 2.9451,
+      "step": 10135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.647395954516053e-05,
+      "loss": 2.9327,
+      "step": 10140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.647048860242736e-05,
+      "loss": 2.8868,
+      "step": 10145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646701601468293e-05,
+      "loss": 2.9939,
+      "step": 10150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646354178205015e-05,
+      "loss": 2.8958,
+      "step": 10155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646006590465202e-05,
+      "loss": 3.0119,
+      "step": 10160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.645658838261158e-05,
+      "loss": 2.9317,
+      "step": 10165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.645310921605191e-05,
+      "loss": 2.8715,
+      "step": 10170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64496284050962e-05,
+      "loss": 2.9369,
+      "step": 10175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.644614594986766e-05,
+      "loss": 2.966,
+      "step": 10180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.644266185048956e-05,
+      "loss": 2.9225,
+      "step": 10185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643917610708522e-05,
+      "loss": 3.0437,
+      "step": 10190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643568871977806e-05,
+      "loss": 2.9569,
+      "step": 10195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64321996886915e-05,
+      "loss": 2.9521,
+      "step": 10200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.642870901394909e-05,
+      "loss": 2.9747,
+      "step": 10205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.642521669567436e-05,
+      "loss": 2.9426,
+      "step": 10210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.642172273399094e-05,
+      "loss": 2.9306,
+      "step": 10215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.641822712902254e-05,
+      "loss": 2.9955,
+      "step": 10220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.641472988089284e-05,
+      "loss": 3.0048,
+      "step": 10225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.641123098972571e-05,
+      "loss": 2.9972,
+      "step": 10230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.640773045564497e-05,
+      "loss": 2.9015,
+      "step": 10235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.640422827877454e-05,
+      "loss": 3.0197,
+      "step": 10240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64007244592384e-05,
+      "loss": 2.9746,
+      "step": 10245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639721899716058e-05,
+      "loss": 2.814,
+      "step": 10250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639371189266516e-05,
+      "loss": 2.9885,
+      "step": 10255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63902031458763e-05,
+      "loss": 2.9781,
+      "step": 10260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63866927569182e-05,
+      "loss": 2.9985,
+      "step": 10265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.638318072591515e-05,
+      "loss": 2.9153,
+      "step": 10270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637966705299142e-05,
+      "loss": 3.0184,
+      "step": 10275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637615173827143e-05,
+      "loss": 2.9873,
+      "step": 10280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637263478187961e-05,
+      "loss": 2.99,
+      "step": 10285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.636911618394045e-05,
+      "loss": 2.9023,
+      "step": 10290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.636559594457852e-05,
+      "loss": 2.8215,
+      "step": 10295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.636207406391842e-05,
+      "loss": 2.8545,
+      "step": 10300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635855054208482e-05,
+      "loss": 2.9526,
+      "step": 10305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635502537920246e-05,
+      "loss": 2.9254,
+      "step": 10310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635149857539614e-05,
+      "loss": 2.9983,
+      "step": 10315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.634797013079066e-05,
+      "loss": 2.8816,
+      "step": 10320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.634444004551097e-05,
+      "loss": 2.9596,
+      "step": 10325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6340908319682e-05,
+      "loss": 2.9546,
+      "step": 10330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.633737495342879e-05,
+      "loss": 2.8975,
+      "step": 10335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.633383994687642e-05,
+      "loss": 3.0765,
+      "step": 10340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.633030330015e-05,
+      "loss": 2.8514,
+      "step": 10345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.632676501337474e-05,
+      "loss": 2.9299,
+      "step": 10350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.632322508667591e-05,
+      "loss": 2.873,
+      "step": 10355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63196835201788e-05,
+      "loss": 2.8843,
+      "step": 10360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.631614031400875e-05,
+      "loss": 2.8877,
+      "step": 10365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.631259546829125e-05,
+      "loss": 2.8916,
+      "step": 10370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.630904898315174e-05,
+      "loss": 2.9248,
+      "step": 10375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.630550085871577e-05,
+      "loss": 2.9669,
+      "step": 10380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.630195109510895e-05,
+      "loss": 2.9545,
+      "step": 10385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.629839969245693e-05,
+      "loss": 2.8973,
+      "step": 10390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.629484665088542e-05,
+      "loss": 3.0241,
+      "step": 10395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.62912919705202e-05,
+      "loss": 2.8807,
+      "step": 10400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.628773565148712e-05,
+      "loss": 2.8427,
+      "step": 10405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.628417769391205e-05,
+      "loss": 2.8694,
+      "step": 10410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.628061809792093e-05,
+      "loss": 2.974,
+      "step": 10415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.627705686363979e-05,
+      "loss": 2.847,
+      "step": 10420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.627349399119468e-05,
+      "loss": 2.9998,
+      "step": 10425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.626992948071173e-05,
+      "loss": 3.0274,
+      "step": 10430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.626636333231712e-05,
+      "loss": 2.9306,
+      "step": 10435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.626279554613707e-05,
+      "loss": 2.8799,
+      "step": 10440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.62592261222979e-05,
+      "loss": 2.8065,
+      "step": 10445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.625565506092596e-05,
+      "loss": 2.8915,
+      "step": 10450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.625208236214766e-05,
+      "loss": 2.9509,
+      "step": 10455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.624850802608946e-05,
+      "loss": 2.9087,
+      "step": 10460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.624493205287791e-05,
+      "loss": 3.003,
+      "step": 10465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.624135444263955e-05,
+      "loss": 2.8823,
+      "step": 10470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623777519550107e-05,
+      "loss": 2.9608,
+      "step": 10475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623419431158917e-05,
+      "loss": 3.0731,
+      "step": 10480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623061179103059e-05,
+      "loss": 3.0245,
+      "step": 10485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.622702763395215e-05,
+      "loss": 3.0109,
+      "step": 10490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.622344184048074e-05,
+      "loss": 2.8377,
+      "step": 10495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.621985441074328e-05,
+      "loss": 3.0497,
+      "step": 10500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.621626534486677e-05,
+      "loss": 2.9294,
+      "step": 10505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.621267464297825e-05,
+      "loss": 2.8612,
+      "step": 10510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620908230520485e-05,
+      "loss": 2.9862,
+      "step": 10515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620548833167369e-05,
+      "loss": 2.9725,
+      "step": 10520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620189272251204e-05,
+      "loss": 2.882,
+      "step": 10525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.619829547784716e-05,
+      "loss": 2.9664,
+      "step": 10530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61946965978064e-05,
+      "loss": 2.9849,
+      "step": 10535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.619109608251713e-05,
+      "loss": 2.9498,
+      "step": 10540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.618749393210684e-05,
+      "loss": 2.96,
+      "step": 10545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.618389014670302e-05,
+      "loss": 3.0452,
+      "step": 10550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.618028472643326e-05,
+      "loss": 2.9861,
+      "step": 10555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.617667767142516e-05,
+      "loss": 3.0085,
+      "step": 10560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.617306898180642e-05,
+      "loss": 2.9343,
+      "step": 10565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61694586577048e-05,
+      "loss": 3.05,
+      "step": 10570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.616584669924808e-05,
+      "loss": 2.9718,
+      "step": 10575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.616223310656414e-05,
+      "loss": 2.9254,
+      "step": 10580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615861787978086e-05,
+      "loss": 2.9672,
+      "step": 10585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615500101902627e-05,
+      "loss": 2.9219,
+      "step": 10590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615138252442837e-05,
+      "loss": 2.9969,
+      "step": 10595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.614776239611525e-05,
+      "loss": 3.0482,
+      "step": 10600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.614414063421507e-05,
+      "loss": 3.0019,
+      "step": 10605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.614051723885602e-05,
+      "loss": 3.0107,
+      "step": 10610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.613689221016637e-05,
+      "loss": 2.9527,
+      "step": 10615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.613326554827447e-05,
+      "loss": 3.0047,
+      "step": 10620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612963725330868e-05,
+      "loss": 3.0292,
+      "step": 10625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612600732539743e-05,
+      "loss": 3.0028,
+      "step": 10630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612237576466923e-05,
+      "loss": 2.963,
+      "step": 10635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611874257125262e-05,
+      "loss": 2.9764,
+      "step": 10640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611510774527624e-05,
+      "loss": 2.9977,
+      "step": 10645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611147128686872e-05,
+      "loss": 2.9526,
+      "step": 10650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61078331961588e-05,
+      "loss": 2.8174,
+      "step": 10655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.610419347327528e-05,
+      "loss": 2.9545,
+      "step": 10660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.6100552118347e-05,
+      "loss": 3.0629,
+      "step": 10665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.609690913150285e-05,
+      "loss": 2.8855,
+      "step": 10670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60932645128718e-05,
+      "loss": 2.9628,
+      "step": 10675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608961826258285e-05,
+      "loss": 2.9517,
+      "step": 10680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608597038076508e-05,
+      "loss": 2.9665,
+      "step": 10685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608232086754763e-05,
+      "loss": 2.951,
+      "step": 10690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60786697230597e-05,
+      "loss": 2.9394,
+      "step": 10695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.607501694743048e-05,
+      "loss": 2.9704,
+      "step": 10700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.607136254078935e-05,
+      "loss": 3.0385,
+      "step": 10705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.606770650326561e-05,
+      "loss": 2.943,
+      "step": 10710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.606404883498874e-05,
+      "loss": 2.9094,
+      "step": 10715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.606038953608818e-05,
+      "loss": 2.973,
+      "step": 10720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.605672860669346e-05,
+      "loss": 2.9597,
+      "step": 10725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.605306604693418e-05,
+      "loss": 2.9375,
+      "step": 10730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.604940185694e-05,
+      "loss": 2.8892,
+      "step": 10735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.604573603684064e-05,
+      "loss": 2.9684,
+      "step": 10740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.604206858676584e-05,
+      "loss": 2.903,
+      "step": 10745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.603839950684543e-05,
+      "loss": 3.0293,
+      "step": 10750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.603472879720932e-05,
+      "loss": 2.893,
+      "step": 10755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.603105645798742e-05,
+      "loss": 2.8762,
+      "step": 10760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.602738248930971e-05,
+      "loss": 2.8948,
+      "step": 10765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60237068913063e-05,
+      "loss": 2.9558,
+      "step": 10770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.602002966410726e-05,
+      "loss": 2.9419,
+      "step": 10775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.601635080784277e-05,
+      "loss": 2.954,
+      "step": 10780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.601267032264307e-05,
+      "loss": 2.9737,
+      "step": 10785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.600898820863843e-05,
+      "loss": 3.0484,
+      "step": 10790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.600530446595921e-05,
+      "loss": 2.9162,
+      "step": 10795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60016190947358e-05,
+      "loss": 2.9489,
+      "step": 10800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.599793209509865e-05,
+      "loss": 2.931,
+      "step": 10805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.599424346717829e-05,
+      "loss": 2.9434,
+      "step": 10810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.599055321110528e-05,
+      "loss": 2.8426,
+      "step": 10815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59868613270103e-05,
+      "loss": 2.9964,
+      "step": 10820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.598316781502395e-05,
+      "loss": 2.9757,
+      "step": 10825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.597947267527706e-05,
+      "loss": 2.9937,
+      "step": 10830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.597577590790039e-05,
+      "loss": 3.0021,
+      "step": 10835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.597207751302482e-05,
+      "loss": 2.8458,
+      "step": 10840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.596837749078126e-05,
+      "loss": 2.9854,
+      "step": 10845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59646758413007e-05,
+      "loss": 2.9545,
+      "step": 10850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.596097256471414e-05,
+      "loss": 2.9416,
+      "step": 10855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595726766115273e-05,
+      "loss": 2.9512,
+      "step": 10860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595356113074757e-05,
+      "loss": 3.014,
+      "step": 10865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.594985297362989e-05,
+      "loss": 2.9515,
+      "step": 10870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.594614318993096e-05,
+      "loss": 2.9847,
+      "step": 10875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.594243177978208e-05,
+      "loss": 2.9617,
+      "step": 10880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.593871874331466e-05,
+      "loss": 2.9277,
+      "step": 10885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.593500408066012e-05,
+      "loss": 2.8615,
+      "step": 10890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.593128779194995e-05,
+      "loss": 2.9742,
+      "step": 10895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59275698773157e-05,
+      "loss": 2.8845,
+      "step": 10900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.592385033688902e-05,
+      "loss": 2.9618,
+      "step": 10905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.592012917080153e-05,
+      "loss": 3.0029,
+      "step": 10910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.591640637918499e-05,
+      "loss": 2.953,
+      "step": 10915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.591268196217115e-05,
+      "loss": 2.9566,
+      "step": 10920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.590895591989189e-05,
+      "loss": 2.9557,
+      "step": 10925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.590522825247909e-05,
+      "loss": 2.9766,
+      "step": 10930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.590149896006469e-05,
+      "loss": 2.9124,
+      "step": 10935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.589776804278072e-05,
+      "loss": 2.9215,
+      "step": 10940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.589403550075925e-05,
+      "loss": 2.872,
+      "step": 10945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.589030133413242e-05,
+      "loss": 2.8562,
+      "step": 10950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.588656554303238e-05,
+      "loss": 2.9113,
+      "step": 10955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58828281275914e-05,
+      "loss": 2.937,
+      "step": 10960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.587908908794178e-05,
+      "loss": 2.8777,
+      "step": 10965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58753484242159e-05,
+      "loss": 2.9143,
+      "step": 10970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.587160613654614e-05,
+      "loss": 2.9787,
+      "step": 10975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.586786222506497e-05,
+      "loss": 2.9121,
+      "step": 10980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.586411668990493e-05,
+      "loss": 3.0078,
+      "step": 10985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.586036953119863e-05,
+      "loss": 2.8846,
+      "step": 10990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.585662074907871e-05,
+      "loss": 2.9637,
+      "step": 10995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.585287034367785e-05,
+      "loss": 2.9047,
+      "step": 11000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.584911831512884e-05,
+      "loss": 2.9681,
+      "step": 11005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.584536466356447e-05,
+      "loss": 2.961,
+      "step": 11010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.584160938911764e-05,
+      "loss": 3.0343,
+      "step": 11015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583785249192125e-05,
+      "loss": 2.9752,
+      "step": 11020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583409397210832e-05,
+      "loss": 3.1216,
+      "step": 11025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58303338298119e-05,
+      "loss": 2.916,
+      "step": 11030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.582657206516508e-05,
+      "loss": 2.98,
+      "step": 11035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.582280867830103e-05,
+      "loss": 3.012,
+      "step": 11040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.581904366935298e-05,
+      "loss": 2.8859,
+      "step": 11045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.581527703845419e-05,
+      "loss": 2.9032,
+      "step": 11050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.5811508785738e-05,
+      "loss": 2.9285,
+      "step": 11055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580773891133778e-05,
+      "loss": 2.9499,
+      "step": 11060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580396741538705e-05,
+      "loss": 3.0187,
+      "step": 11065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580019429801924e-05,
+      "loss": 3.0168,
+      "step": 11070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.579641955936795e-05,
+      "loss": 2.9141,
+      "step": 11075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.579264319956682e-05,
+      "loss": 2.9785,
+      "step": 11080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.578886521874948e-05,
+      "loss": 2.9656,
+      "step": 11085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.578508561704972e-05,
+      "loss": 2.8679,
+      "step": 11090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.578130439460129e-05,
+      "loss": 2.8907,
+      "step": 11095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.577752155153807e-05,
+      "loss": 2.83,
+      "step": 11100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.577373708799397e-05,
+      "loss": 2.9112,
+      "step": 11105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.576995100410292e-05,
+      "loss": 3.0272,
+      "step": 11110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.576616329999902e-05,
+      "loss": 3.0158,
+      "step": 11115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.576237397581625e-05,
+      "loss": 2.9241,
+      "step": 11120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.575858303168883e-05,
+      "loss": 2.9853,
+      "step": 11125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.575479046775091e-05,
+      "loss": 2.98,
+      "step": 11130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.575099628413677e-05,
+      "loss": 2.866,
+      "step": 11135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.574720048098071e-05,
+      "loss": 2.9486,
+      "step": 11140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.57434030584171e-05,
+      "loss": 2.929,
+      "step": 11145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573960401658035e-05,
+      "loss": 2.9696,
+      "step": 11150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573580335560497e-05,
+      "loss": 2.9917,
+      "step": 11155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573200107562548e-05,
+      "loss": 2.9047,
+      "step": 11160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.572819717677648e-05,
+      "loss": 2.9603,
+      "step": 11165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.572439165919262e-05,
+      "loss": 2.9859,
+      "step": 11170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.572058452300862e-05,
+      "loss": 2.858,
+      "step": 11175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.571677576835926e-05,
+      "loss": 3.0072,
+      "step": 11180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.571296539537935e-05,
+      "loss": 2.9168,
+      "step": 11185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.570915340420377e-05,
+      "loss": 3.0137,
+      "step": 11190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.570533979496748e-05,
+      "loss": 2.9069,
+      "step": 11195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.570152456780545e-05,
+      "loss": 2.9403,
+      "step": 11200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.569770772285276e-05,
+      "loss": 2.9184,
+      "step": 11205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.569388926024452e-05,
+      "loss": 3.0545,
+      "step": 11210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.569006918011589e-05,
+      "loss": 2.964,
+      "step": 11215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.56862474826021e-05,
+      "loss": 3.0111,
+      "step": 11220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.568242416783844e-05,
+      "loss": 2.95,
+      "step": 11225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.567859923596024e-05,
+      "loss": 2.8678,
+      "step": 11230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.56747726871029e-05,
+      "loss": 3.0433,
+      "step": 11235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.567094452140191e-05,
+      "loss": 2.9153,
+      "step": 11240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.566711473899274e-05,
+      "loss": 2.8971,
+      "step": 11245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.566328334001098e-05,
+      "loss": 2.9095,
+      "step": 11250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.565945032459225e-05,
+      "loss": 2.9303,
+      "step": 11255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.565561569287225e-05,
+      "loss": 2.9435,
+      "step": 11260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.565177944498669e-05,
+      "loss": 2.8846,
+      "step": 11265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.564794158107143e-05,
+      "loss": 2.8897,
+      "step": 11270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.564410210126224e-05,
+      "loss": 2.8801,
+      "step": 11275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.564026100569512e-05,
+      "loss": 2.9142,
+      "step": 11280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5636418294506e-05,
+      "loss": 2.9464,
+      "step": 11285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56325739678309e-05,
+      "loss": 2.9837,
+      "step": 11290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.562872802580591e-05,
+      "loss": 3.0515,
+      "step": 11295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56248804685672e-05,
+      "loss": 2.9771,
+      "step": 11300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.562103129625095e-05,
+      "loss": 2.9959,
+      "step": 11305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56171805089934e-05,
+      "loss": 3.0384,
+      "step": 11310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56133281069309e-05,
+      "loss": 2.9707,
+      "step": 11315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56094740901998e-05,
+      "loss": 2.9783,
+      "step": 11320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.560561845893652e-05,
+      "loss": 3.0212,
+      "step": 11325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.560176121327756e-05,
+      "loss": 2.8416,
+      "step": 11330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559790235335947e-05,
+      "loss": 2.9729,
+      "step": 11335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559404187931883e-05,
+      "loss": 3.0101,
+      "step": 11340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559017979129231e-05,
+      "loss": 3.0344,
+      "step": 11345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.558631608941663e-05,
+      "loss": 2.9926,
+      "step": 11350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.558245077382855e-05,
+      "loss": 2.8227,
+      "step": 11355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557858384466489e-05,
+      "loss": 3.125,
+      "step": 11360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557471530206254e-05,
+      "loss": 2.9584,
+      "step": 11365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557084514615847e-05,
+      "loss": 2.9801,
+      "step": 11370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.556697337708966e-05,
+      "loss": 2.9355,
+      "step": 11375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.556309999499315e-05,
+      "loss": 2.897,
+      "step": 11380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.555922500000608e-05,
+      "loss": 2.9384,
+      "step": 11385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.555534839226561e-05,
+      "loss": 2.9601,
+      "step": 11390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.555147017190897e-05,
+      "loss": 3.0301,
+      "step": 11395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.554759033907345e-05,
+      "loss": 2.8904,
+      "step": 11400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.554370889389639e-05,
+      "loss": 3.0675,
+      "step": 11405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.553982583651517e-05,
+      "loss": 2.9181,
+      "step": 11410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.553594116706728e-05,
+      "loss": 2.931,
+      "step": 11415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.553205488569021e-05,
+      "loss": 2.9669,
+      "step": 11420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.552816699252153e-05,
+      "loss": 2.9709,
+      "step": 11425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.55242774876989e-05,
+      "loss": 2.961,
+      "step": 11430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.552038637135996e-05,
+      "loss": 3.053,
+      "step": 11435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.551649364364248e-05,
+      "loss": 2.9701,
+      "step": 11440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.551259930468425e-05,
+      "loss": 2.999,
+      "step": 11445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.550870335462311e-05,
+      "loss": 3.0088,
+      "step": 11450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.550480579359702e-05,
+      "loss": 2.8475,
+      "step": 11455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.550090662174391e-05,
+      "loss": 2.9266,
+      "step": 11460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.54970058392018e-05,
+      "loss": 2.9649,
+      "step": 11465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.549310344610882e-05,
+      "loss": 2.9259,
+      "step": 11470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548919944260306e-05,
+      "loss": 2.9425,
+      "step": 11475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548529382882275e-05,
+      "loss": 2.9508,
+      "step": 11480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548138660490613e-05,
+      "loss": 2.9526,
+      "step": 11485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.547747777099152e-05,
+      "loss": 3.0379,
+      "step": 11490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.547356732721728e-05,
+      "loss": 2.9454,
+      "step": 11495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.546965527372184e-05,
+      "loss": 3.012,
+      "step": 11500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.54657416106437e-05,
+      "loss": 2.936,
+      "step": 11505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.546182633812137e-05,
+      "loss": 2.9823,
+      "step": 11510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545790945629347e-05,
+      "loss": 2.922,
+      "step": 11515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545399096529865e-05,
+      "loss": 2.9656,
+      "step": 11520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545007086527562e-05,
+      "loss": 2.9582,
+      "step": 11525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.544614915636313e-05,
+      "loss": 2.9376,
+      "step": 11530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.544222583870002e-05,
+      "loss": 2.9246,
+      "step": 11535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.543830091242519e-05,
+      "loss": 2.8623,
+      "step": 11540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.543437437767754e-05,
+      "loss": 2.9283,
+      "step": 11545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.54304462345961e-05,
+      "loss": 2.9456,
+      "step": 11550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.542651648331989e-05,
+      "loss": 2.9478,
+      "step": 11555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.542258512398805e-05,
+      "loss": 3.0419,
+      "step": 11560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.541865215673973e-05,
+      "loss": 3.0549,
+      "step": 11565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.541471758171415e-05,
+      "loss": 2.9801,
+      "step": 11570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.54107813990506e-05,
+      "loss": 2.8942,
+      "step": 11575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.540684360888844e-05,
+      "loss": 3.0352,
+      "step": 11580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5402904211367e-05,
+      "loss": 2.9058,
+      "step": 11585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53989632066258e-05,
+      "loss": 3.0735,
+      "step": 11590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53950205948043e-05,
+      "loss": 2.9245,
+      "step": 11595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.539107637604208e-05,
+      "loss": 3.0214,
+      "step": 11600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.538713055047878e-05,
+      "loss": 2.997,
+      "step": 11605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.538318311825404e-05,
+      "loss": 2.8918,
+      "step": 11610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.537923407950764e-05,
+      "loss": 2.9111,
+      "step": 11615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.537528343437933e-05,
+      "loss": 3.0139,
+      "step": 11620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5371331183009e-05,
+      "loss": 2.936,
+      "step": 11625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.536737732553652e-05,
+      "loss": 2.9634,
+      "step": 11630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.536342186210188e-05,
+      "loss": 3.0212,
+      "step": 11635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.535946479284507e-05,
+      "loss": 2.9653,
+      "step": 11640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53555061179062e-05,
+      "loss": 2.9448,
+      "step": 11645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.535154583742538e-05,
+      "loss": 2.9139,
+      "step": 11650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53475839515428e-05,
+      "loss": 2.949,
+      "step": 11655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.534362046039872e-05,
+      "loss": 2.8976,
+      "step": 11660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.533965536413343e-05,
+      "loss": 2.9508,
+      "step": 11665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.533568866288732e-05,
+      "loss": 2.9656,
+      "step": 11670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.533172035680075e-05,
+      "loss": 2.8662,
+      "step": 11675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.532775044601428e-05,
+      "loss": 2.9256,
+      "step": 11680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.532377893066835e-05,
+      "loss": 3.0195,
+      "step": 11685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.531980581090361e-05,
+      "loss": 2.9721,
+      "step": 11690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.531583108686069e-05,
+      "loss": 2.9977,
+      "step": 11695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.531185475868027e-05,
+      "loss": 2.8687,
+      "step": 11700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.530787682650313e-05,
+      "loss": 2.8974,
+      "step": 11705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.530389729047007e-05,
+      "loss": 3.0038,
+      "step": 11710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.529991615072197e-05,
+      "loss": 2.899,
+      "step": 11715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.529593340739977e-05,
+      "loss": 2.9499,
+      "step": 11720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.529194906064445e-05,
+      "loss": 2.9578,
+      "step": 11725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.528796311059704e-05,
+      "loss": 2.8976,
+      "step": 11730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.528397555739863e-05,
+      "loss": 3.0045,
+      "step": 11735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.527998640119042e-05,
+      "loss": 2.9756,
+      "step": 11740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.527599564211357e-05,
+      "loss": 2.8481,
+      "step": 11745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.527200328030938e-05,
+      "loss": 2.9339,
+      "step": 11750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.526800931591917e-05,
+      "loss": 2.8367,
+      "step": 11755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.526401374908431e-05,
+      "loss": 2.9977,
+      "step": 11760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.526001657994627e-05,
+      "loss": 2.991,
+      "step": 11765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.525601780864652e-05,
+      "loss": 2.8688,
+      "step": 11770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.525201743532661e-05,
+      "loss": 3.0178,
+      "step": 11775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.524801546012815e-05,
+      "loss": 2.9825,
+      "step": 11780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.524401188319285e-05,
+      "loss": 2.9125,
+      "step": 11785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.524000670466238e-05,
+      "loss": 2.9296,
+      "step": 11790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.523599992467853e-05,
+      "loss": 2.9446,
+      "step": 11795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.523199154338314e-05,
+      "loss": 2.9219,
+      "step": 11800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.522798156091812e-05,
+      "loss": 2.9795,
+      "step": 11805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.522396997742539e-05,
+      "loss": 2.9208,
+      "step": 11810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.521995679304698e-05,
+      "loss": 2.9692,
+      "step": 11815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.521594200792494e-05,
+      "loss": 2.9056,
+      "step": 11820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52119256222014e-05,
+      "loss": 2.954,
+      "step": 11825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.520790763601852e-05,
+      "loss": 3.0107,
+      "step": 11830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.520388804951857e-05,
+      "loss": 2.9953,
+      "step": 11835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.519986686284379e-05,
+      "loss": 3.0361,
+      "step": 11840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.519584407613655e-05,
+      "loss": 2.975,
+      "step": 11845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.519181968953926e-05,
+      "loss": 2.9021,
+      "step": 11850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.518779370319436e-05,
+      "loss": 2.9718,
+      "step": 11855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.518376611724439e-05,
+      "loss": 2.8964,
+      "step": 11860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.517973693183191e-05,
+      "loss": 2.9153,
+      "step": 11865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.517570614709954e-05,
+      "loss": 3.0329,
+      "step": 11870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.517167376318999e-05,
+      "loss": 2.8839,
+      "step": 11875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.516763978024597e-05,
+      "loss": 2.8532,
+      "step": 11880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.516360419841031e-05,
+      "loss": 2.78,
+      "step": 11885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.515956701782584e-05,
+      "loss": 2.8629,
+      "step": 11890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51555282386355e-05,
+      "loss": 2.9511,
+      "step": 11895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.515148786098224e-05,
+      "loss": 2.9345,
+      "step": 11900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.514744588500909e-05,
+      "loss": 2.9284,
+      "step": 11905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.514340231085913e-05,
+      "loss": 3.0014,
+      "step": 11910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51393571386755e-05,
+      "loss": 2.9942,
+      "step": 11915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51353103686014e-05,
+      "loss": 2.8754,
+      "step": 11920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.513126200078009e-05,
+      "loss": 3.0172,
+      "step": 11925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.512721203535484e-05,
+      "loss": 3.0148,
+      "step": 11930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.512316047246907e-05,
+      "loss": 3.0001,
+      "step": 11935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.511910731226614e-05,
+      "loss": 2.9591,
+      "step": 11940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51150525548896e-05,
+      "loss": 2.9354,
+      "step": 11945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.511099620048292e-05,
+      "loss": 2.9757,
+      "step": 11950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51069382491897e-05,
+      "loss": 2.9381,
+      "step": 11955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.510287870115364e-05,
+      "loss": 3.0618,
+      "step": 11960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.509881755651839e-05,
+      "loss": 3.0257,
+      "step": 11965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.509475481542774e-05,
+      "loss": 3.0196,
+      "step": 11970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.509069047802548e-05,
+      "loss": 2.8449,
+      "step": 11975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.50866245444555e-05,
+      "loss": 2.9877,
+      "step": 11980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.508255701486173e-05,
+      "loss": 2.9029,
+      "step": 11985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.507848788938818e-05,
+      "loss": 2.9615,
+      "step": 11990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.507441716817884e-05,
+      "loss": 2.8868,
+      "step": 11995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.507034485137786e-05,
+      "loss": 2.931,
+      "step": 12000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.506627093912937e-05,
+      "loss": 2.9331,
+      "step": 12005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.50621954315776e-05,
+      "loss": 2.8123,
+      "step": 12010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.505811832886679e-05,
+      "loss": 2.9148,
+      "step": 12015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.505403963114128e-05,
+      "loss": 3.0597,
+      "step": 12020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.504995933854546e-05,
+      "loss": 2.8804,
+      "step": 12025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.504587745122378e-05,
+      "loss": 2.9361,
+      "step": 12030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.50417939693207e-05,
+      "loss": 2.8622,
+      "step": 12035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.50377088929808e-05,
+      "loss": 2.9971,
+      "step": 12040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.503362222234866e-05,
+      "loss": 2.9923,
+      "step": 12045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.502953395756897e-05,
+      "loss": 2.8899,
+      "step": 12050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.502544409878645e-05,
+      "loss": 3.0247,
+      "step": 12055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.502135264614587e-05,
+      "loss": 2.9,
+      "step": 12060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.501725959979206e-05,
+      "loss": 3.0564,
+      "step": 12065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.501316495986992e-05,
+      "loss": 2.9832,
+      "step": 12070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.500906872652437e-05,
+      "loss": 2.9373,
+      "step": 12075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.500497089990046e-05,
+      "loss": 2.9317,
+      "step": 12080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.500087148014321e-05,
+      "loss": 2.8736,
+      "step": 12085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.499677046739774e-05,
+      "loss": 2.8879,
+      "step": 12090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.499266786180922e-05,
+      "loss": 2.8787,
+      "step": 12095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.498856366352291e-05,
+      "loss": 3.0734,
+      "step": 12100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.498445787268406e-05,
+      "loss": 2.9384,
+      "step": 12105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.498035048943803e-05,
+      "loss": 2.935,
+      "step": 12110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.497624151393021e-05,
+      "loss": 2.8652,
+      "step": 12115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.497213094630604e-05,
+      "loss": 3.0103,
+      "step": 12120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.496801878671107e-05,
+      "loss": 2.994,
+      "step": 12125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.496390503529083e-05,
+      "loss": 2.9422,
+      "step": 12130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.495978969219095e-05,
+      "loss": 2.9865,
+      "step": 12135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.495567275755713e-05,
+      "loss": 2.8924,
+      "step": 12140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.495155423153506e-05,
+      "loss": 3.0391,
+      "step": 12145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.494743411427058e-05,
+      "loss": 2.9925,
+      "step": 12150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.494331240590952e-05,
+      "loss": 2.9974,
+      "step": 12155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.493918910659778e-05,
+      "loss": 2.9083,
+      "step": 12160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.493506421648133e-05,
+      "loss": 3.0241,
+      "step": 12165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.493093773570616e-05,
+      "loss": 2.951,
+      "step": 12170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.492680966441839e-05,
+      "loss": 2.9637,
+      "step": 12175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.492268000276412e-05,
+      "loss": 2.9431,
+      "step": 12180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.491854875088954e-05,
+      "loss": 2.8453,
+      "step": 12185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.491441590894089e-05,
+      "loss": 2.8756,
+      "step": 12190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.491028147706447e-05,
+      "loss": 3.0043,
+      "step": 12195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.490614545540665e-05,
+      "loss": 2.8306,
+      "step": 12200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.490200784411382e-05,
+      "loss": 2.9297,
+      "step": 12205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.489786864333246e-05,
+      "loss": 2.9641,
+      "step": 12210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.489372785320909e-05,
+      "loss": 2.9808,
+      "step": 12215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.488958547389028e-05,
+      "loss": 2.8707,
+      "step": 12220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48854415055227e-05,
+      "loss": 3.0164,
+      "step": 12225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.4881295948253e-05,
+      "loss": 2.8926,
+      "step": 12230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.487714880222794e-05,
+      "loss": 2.9694,
+      "step": 12235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.487300006759435e-05,
+      "loss": 2.89,
+      "step": 12240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.486884974449906e-05,
+      "loss": 2.934,
+      "step": 12245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.486469783308902e-05,
+      "loss": 2.9602,
+      "step": 12250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.486054433351116e-05,
+      "loss": 2.8523,
+      "step": 12255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485638924591256e-05,
+      "loss": 2.8915,
+      "step": 12260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485223257044027e-05,
+      "loss": 2.9775,
+      "step": 12265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.484807430724146e-05,
+      "loss": 2.985,
+      "step": 12270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.484391445646328e-05,
+      "loss": 2.9048,
+      "step": 12275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.483975301825305e-05,
+      "loss": 2.8968,
+      "step": 12280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.483558999275803e-05,
+      "loss": 2.9833,
+      "step": 12285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48314253801256e-05,
+      "loss": 2.9205,
+      "step": 12290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.482725918050321e-05,
+      "loss": 2.8476,
+      "step": 12295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48230913940383e-05,
+      "loss": 2.9443,
+      "step": 12300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.481892202087843e-05,
+      "loss": 3.0147,
+      "step": 12305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.481475106117119e-05,
+      "loss": 2.9392,
+      "step": 12310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48105785150642e-05,
+      "loss": 2.9259,
+      "step": 12315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.480640438270522e-05,
+      "loss": 2.894,
+      "step": 12320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.480222866424195e-05,
+      "loss": 2.8478,
+      "step": 12325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.479805135982226e-05,
+      "loss": 2.9873,
+      "step": 12330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.479387246959399e-05,
+      "loss": 3.0168,
+      "step": 12335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.478969199370507e-05,
+      "loss": 2.9394,
+      "step": 12340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.478550993230349e-05,
+      "loss": 2.9597,
+      "step": 12345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.47813262855373e-05,
+      "loss": 2.9403,
+      "step": 12350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.477714105355458e-05,
+      "loss": 2.9165,
+      "step": 12355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.47729542365035e-05,
+      "loss": 2.8846,
+      "step": 12360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476876583453226e-05,
+      "loss": 2.8923,
+      "step": 12365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476457584778913e-05,
+      "loss": 2.9833,
+      "step": 12370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476038427642242e-05,
+      "loss": 2.974,
+      "step": 12375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.475619112058051e-05,
+      "loss": 2.9206,
+      "step": 12380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.475199638041185e-05,
+      "loss": 2.9356,
+      "step": 12385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.474780005606494e-05,
+      "loss": 2.9343,
+      "step": 12390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.474360214768828e-05,
+      "loss": 3.0449,
+      "step": 12395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.473940265543051e-05,
+      "loss": 2.8749,
+      "step": 12400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.473520157944027e-05,
+      "loss": 2.9223,
+      "step": 12405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.473099891986628e-05,
+      "loss": 2.899,
+      "step": 12410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.47267946768573e-05,
+      "loss": 2.8453,
+      "step": 12415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.472258885056217e-05,
+      "loss": 2.8634,
+      "step": 12420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.471838144112979e-05,
+      "loss": 2.8911,
+      "step": 12425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.471417244870905e-05,
+      "loss": 2.8736,
+      "step": 12430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.470996187344899e-05,
+      "loss": 2.9065,
+      "step": 12435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.470574971549864e-05,
+      "loss": 2.9276,
+      "step": 12440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.47015359750071e-05,
+      "loss": 3.0055,
+      "step": 12445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.469732065212355e-05,
+      "loss": 2.8781,
+      "step": 12450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.46931037469972e-05,
+      "loss": 2.9634,
+      "step": 12455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.46888852597773e-05,
+      "loss": 3.0115,
+      "step": 12460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.468466519061323e-05,
+      "loss": 2.942,
+      "step": 12465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.468044353965434e-05,
+      "loss": 2.9882,
+      "step": 12470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.467622030705008e-05,
+      "loss": 2.9846,
+      "step": 12475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.467199549294996e-05,
+      "loss": 2.9632,
+      "step": 12480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.466776909750352e-05,
+      "loss": 2.944,
+      "step": 12485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.466354112086037e-05,
+      "loss": 2.8535,
+      "step": 12490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465931156317018e-05,
+      "loss": 2.8925,
+      "step": 12495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465508042458269e-05,
+      "loss": 2.9621,
+      "step": 12500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465084770524766e-05,
+      "loss": 3.0206,
+      "step": 12505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.464661340531492e-05,
+      "loss": 2.839,
+      "step": 12510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.464237752493436e-05,
+      "loss": 2.897,
+      "step": 12515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.463814006425595e-05,
+      "loss": 3.0097,
+      "step": 12520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.463390102342968e-05,
+      "loss": 2.9142,
+      "step": 12525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.462966040260559e-05,
+      "loss": 2.9721,
+      "step": 12530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.46254182019338e-05,
+      "loss": 2.9096,
+      "step": 12535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.462117442156451e-05,
+      "loss": 2.8626,
+      "step": 12540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.46169290616479e-05,
+      "loss": 2.9449,
+      "step": 12545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.461268212233429e-05,
+      "loss": 2.9944,
+      "step": 12550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.4608433603774e-05,
+      "loss": 2.9532,
+      "step": 12555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.460418350611744e-05,
+      "loss": 2.9497,
+      "step": 12560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.459993182951502e-05,
+      "loss": 2.9441,
+      "step": 12565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.459567857411729e-05,
+      "loss": 2.8659,
+      "step": 12570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.459142374007478e-05,
+      "loss": 3.0314,
+      "step": 12575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.458716732753812e-05,
+      "loss": 2.9243,
+      "step": 12580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.458290933665798e-05,
+      "loss": 2.9722,
+      "step": 12585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.45786497675851e-05,
+      "loss": 3.0246,
+      "step": 12590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.457438862047026e-05,
+      "loss": 2.8945,
+      "step": 12595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.457012589546428e-05,
+      "loss": 2.948,
+      "step": 12600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.45658615927181e-05,
+      "loss": 2.9144,
+      "step": 12605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.456159571238263e-05,
+      "loss": 2.872,
+      "step": 12610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.455732825460887e-05,
+      "loss": 3.0049,
+      "step": 12615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.455305921954794e-05,
+      "loss": 2.9112,
+      "step": 12620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.454878860735092e-05,
+      "loss": 2.9741,
+      "step": 12625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.454451641816898e-05,
+      "loss": 2.9202,
+      "step": 12630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.454024265215337e-05,
+      "loss": 3.0153,
+      "step": 12635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.453596730945537e-05,
+      "loss": 2.9979,
+      "step": 12640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.453169039022633e-05,
+      "loss": 2.8485,
+      "step": 12645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.452741189461764e-05,
+      "loss": 2.9984,
+      "step": 12650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.452313182278076e-05,
+      "loss": 3.009,
+      "step": 12655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.451885017486718e-05,
+      "loss": 2.9752,
+      "step": 12660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.451456695102852e-05,
+      "loss": 2.8246,
+      "step": 12665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.451028215141634e-05,
+      "loss": 2.8355,
+      "step": 12670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.450599577618235e-05,
+      "loss": 2.96,
+      "step": 12675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.450170782547829e-05,
+      "loss": 2.9807,
+      "step": 12680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.449741829945593e-05,
+      "loss": 3.052,
+      "step": 12685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44931271982671e-05,
+      "loss": 2.9549,
+      "step": 12690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.448883452206377e-05,
+      "loss": 2.9875,
+      "step": 12695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.448454027099782e-05,
+      "loss": 2.9009,
+      "step": 12700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44802444452213e-05,
+      "loss": 2.8599,
+      "step": 12705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.447594704488627e-05,
+      "loss": 2.9037,
+      "step": 12710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.447164807014484e-05,
+      "loss": 2.9599,
+      "step": 12715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.446734752114923e-05,
+      "loss": 2.8636,
+      "step": 12720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.446304539805163e-05,
+      "loss": 3.0279,
+      "step": 12725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.445874170100436e-05,
+      "loss": 2.884,
+      "step": 12730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.445443643015977e-05,
+      "loss": 2.9922,
+      "step": 12735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.445012958567023e-05,
+      "loss": 3.0131,
+      "step": 12740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.444582116768822e-05,
+      "loss": 2.9258,
+      "step": 12745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.444151117636626e-05,
+      "loss": 2.9659,
+      "step": 12750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.443719961185691e-05,
+      "loss": 3.014,
+      "step": 12755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44328864743128e-05,
+      "loss": 2.9355,
+      "step": 12760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44285717638866e-05,
+      "loss": 3.0109,
+      "step": 12765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.442425548073106e-05,
+      "loss": 2.9046,
+      "step": 12770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.441993762499897e-05,
+      "loss": 2.9861,
+      "step": 12775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.441561819684317e-05,
+      "loss": 2.9773,
+      "step": 12780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.441129719641657e-05,
+      "loss": 2.9684,
+      "step": 12785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.440697462387213e-05,
+      "loss": 2.8615,
+      "step": 12790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.440265047936287e-05,
+      "loss": 2.9504,
+      "step": 12795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.439832476304184e-05,
+      "loss": 2.8459,
+      "step": 12800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.439399747506219e-05,
+      "loss": 2.8815,
+      "step": 12805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.43896686155771e-05,
+      "loss": 2.9243,
+      "step": 12810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.438533818473979e-05,
+      "loss": 2.8826,
+      "step": 12815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.438100618270356e-05,
+      "loss": 2.9484,
+      "step": 12820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.437667260962175e-05,
+      "loss": 2.9489,
+      "step": 12825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.43723374656478e-05,
+      "loss": 3.0826,
+      "step": 12830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.436800075093512e-05,
+      "loss": 3.0654,
+      "step": 12835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.436366246563728e-05,
+      "loss": 2.9088,
+      "step": 12840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.43593226099078e-05,
+      "loss": 2.9897,
+      "step": 12845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.435498118390033e-05,
+      "loss": 2.9787,
+      "step": 12850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.435063818776857e-05,
+      "loss": 2.9164,
+      "step": 12855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.434629362166623e-05,
+      "loss": 2.9984,
+      "step": 12860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.434194748574711e-05,
+      "loss": 3.0061,
+      "step": 12865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.433759978016507e-05,
+      "loss": 2.9362,
+      "step": 12870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.4333250505074e-05,
+      "loss": 2.923,
+      "step": 12875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.432889966062788e-05,
+      "loss": 2.8977,
+      "step": 12880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.43245472469807e-05,
+      "loss": 2.918,
+      "step": 12885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.432019326428654e-05,
+      "loss": 2.9773,
+      "step": 12890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.431583771269954e-05,
+      "loss": 2.8567,
+      "step": 12895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.431148059237387e-05,
+      "loss": 2.9211,
+      "step": 12900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.430712190346378e-05,
+      "loss": 2.8655,
+      "step": 12905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.430276164612354e-05,
+      "loss": 3.0275,
+      "step": 12910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.429839982050752e-05,
+      "loss": 2.9412,
+      "step": 12915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.429403642677013e-05,
+      "loss": 3.012,
+      "step": 12920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.42896714650658e-05,
+      "loss": 2.9119,
+      "step": 12925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.428530493554907e-05,
+      "loss": 2.9385,
+      "step": 12930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.428093683837451e-05,
+      "loss": 2.9853,
+      "step": 12935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.427656717369672e-05,
+      "loss": 2.9597,
+      "step": 12940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.427219594167044e-05,
+      "loss": 3.0161,
+      "step": 12945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.426782314245036e-05,
+      "loss": 3.0063,
+      "step": 12950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.426344877619126e-05,
+      "loss": 3.0151,
+      "step": 12955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.425907284304803e-05,
+      "loss": 2.997,
+      "step": 12960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.425469534317556e-05,
+      "loss": 2.895,
+      "step": 12965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.42503162767288e-05,
+      "loss": 2.9543,
+      "step": 12970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.424593564386276e-05,
+      "loss": 2.9913,
+      "step": 12975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.424155344473253e-05,
+      "loss": 2.9498,
+      "step": 12980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.423716967949323e-05,
+      "loss": 2.8911,
+      "step": 12985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.423278434830004e-05,
+      "loss": 2.7783,
+      "step": 12990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.422839745130819e-05,
+      "loss": 2.8738,
+      "step": 12995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.422400898867298e-05,
+      "loss": 3.0093,
+      "step": 13000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.421961896054974e-05,
+      "loss": 3.0031,
+      "step": 13005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.42152273670939e-05,
+      "loss": 2.9118,
+      "step": 13010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.42108342084609e-05,
+      "loss": 2.9531,
+      "step": 13015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.420643948480625e-05,
+      "loss": 2.8018,
+      "step": 13020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.420204319628554e-05,
+      "loss": 2.9038,
+      "step": 13025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.419764534305438e-05,
+      "loss": 2.8706,
+      "step": 13030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.419324592526844e-05,
+      "loss": 3.0088,
+      "step": 13035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.418884494308346e-05,
+      "loss": 2.9388,
+      "step": 13040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.418444239665526e-05,
+      "loss": 2.9827,
+      "step": 13045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.418003828613966e-05,
+      "loss": 2.9755,
+      "step": 13050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.417563261169255e-05,
+      "loss": 2.9766,
+      "step": 13055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.417122537346992e-05,
+      "loss": 2.9645,
+      "step": 13060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.416681657162774e-05,
+      "loss": 2.9705,
+      "step": 13065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.416240620632212e-05,
+      "loss": 2.95,
+      "step": 13070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.415799427770916e-05,
+      "loss": 2.7722,
+      "step": 13075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.415358078594503e-05,
+      "loss": 2.9781,
+      "step": 13080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.4149165731186e-05,
+      "loss": 3.0449,
+      "step": 13085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41447491135883e-05,
+      "loss": 2.9464,
+      "step": 13090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.414033093330833e-05,
+      "loss": 2.9763,
+      "step": 13095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.413591119050246e-05,
+      "loss": 2.8937,
+      "step": 13100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.413148988532717e-05,
+      "loss": 2.8842,
+      "step": 13105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.412706701793892e-05,
+      "loss": 3.0113,
+      "step": 13110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.412264258849433e-05,
+      "loss": 2.9024,
+      "step": 13115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.411821659714999e-05,
+      "loss": 2.872,
+      "step": 13120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.411378904406259e-05,
+      "loss": 3.0028,
+      "step": 13125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.410935992938885e-05,
+      "loss": 3.0077,
+      "step": 13130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.410492925328556e-05,
+      "loss": 2.9317,
+      "step": 13135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.410049701590956e-05,
+      "loss": 3.065,
+      "step": 13140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.409606321741775e-05,
+      "loss": 2.966,
+      "step": 13145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.40916278579671e-05,
+      "loss": 2.9493,
+      "step": 13150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.408719093771458e-05,
+      "loss": 3.01,
+      "step": 13155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.408275245681728e-05,
+      "loss": 3.0655,
+      "step": 13160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.407831241543231e-05,
+      "loss": 2.9912,
+      "step": 13165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.407387081371685e-05,
+      "loss": 2.9117,
+      "step": 13170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.406942765182812e-05,
+      "loss": 3.0014,
+      "step": 13175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.406498292992341e-05,
+      "loss": 2.9446,
+      "step": 13180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.406053664816004e-05,
+      "loss": 2.9695,
+      "step": 13185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.405608880669543e-05,
+      "loss": 2.9584,
+      "step": 13190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.4051639405687e-05,
+      "loss": 2.9064,
+      "step": 13195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.40471884452923e-05,
+      "loss": 3.0222,
+      "step": 13200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.404273592566885e-05,
+      "loss": 2.942,
+      "step": 13205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.403828184697429e-05,
+      "loss": 2.9268,
+      "step": 13210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.403382620936628e-05,
+      "loss": 2.9933,
+      "step": 13215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.402936901300252e-05,
+      "loss": 2.8509,
+      "step": 13220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.402491025804084e-05,
+      "loss": 2.9068,
+      "step": 13225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.402044994463903e-05,
+      "loss": 3.0376,
+      "step": 13230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.4015988072955e-05,
+      "loss": 3.0071,
+      "step": 13235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.401152464314671e-05,
+      "loss": 3.0485,
+      "step": 13240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.400705965537214e-05,
+      "loss": 2.8625,
+      "step": 13245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.400259310978935e-05,
+      "loss": 2.9582,
+      "step": 13250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.399812500655647e-05,
+      "loss": 2.9261,
+      "step": 13255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.399365534583164e-05,
+      "loss": 2.9475,
+      "step": 13260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.398918412777308e-05,
+      "loss": 2.9082,
+      "step": 13265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.398471135253912e-05,
+      "loss": 2.9161,
+      "step": 13270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.398023702028802e-05,
+      "loss": 2.9754,
+      "step": 13275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.397576113117822e-05,
+      "loss": 2.9688,
+      "step": 13280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.397128368536812e-05,
+      "loss": 2.914,
+      "step": 13285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.396680468301627e-05,
+      "loss": 2.9823,
+      "step": 13290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.396232412428116e-05,
+      "loss": 3.0003,
+      "step": 13295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.395784200932145e-05,
+      "loss": 2.9626,
+      "step": 13300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.395335833829579e-05,
+      "loss": 2.9329,
+      "step": 13305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.394887311136286e-05,
+      "loss": 2.9363,
+      "step": 13310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.39443863286815e-05,
+      "loss": 3.0002,
+      "step": 13315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.393989799041048e-05,
+      "loss": 3.0182,
+      "step": 13320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.39354080967087e-05,
+      "loss": 2.9162,
+      "step": 13325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.39309166477351e-05,
+      "loss": 2.9222,
+      "step": 13330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.392642364364868e-05,
+      "loss": 2.9429,
+      "step": 13335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.392192908460848e-05,
+      "loss": 2.9784,
+      "step": 13340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.391743297077362e-05,
+      "loss": 2.9456,
+      "step": 13345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.391293530230322e-05,
+      "loss": 2.9183,
+      "step": 13350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.390843607935654e-05,
+      "loss": 3.0466,
+      "step": 13355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.390393530209281e-05,
+      "loss": 2.9896,
+      "step": 13360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.389943297067139e-05,
+      "loss": 2.9705,
+      "step": 13365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.389492908525161e-05,
+      "loss": 2.9595,
+      "step": 13370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.389042364599294e-05,
+      "loss": 3.0631,
+      "step": 13375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.388591665305487e-05,
+      "loss": 3.0112,
+      "step": 13380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.388140810659692e-05,
+      "loss": 2.9693,
+      "step": 13385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.387689800677869e-05,
+      "loss": 2.8591,
+      "step": 13390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.387238635375986e-05,
+      "loss": 2.9268,
+      "step": 13395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.386787314770011e-05,
+      "loss": 2.9927,
+      "step": 13400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.386335838875921e-05,
+      "loss": 2.8285,
+      "step": 13405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.385884207709701e-05,
+      "loss": 2.9221,
+      "step": 13410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.385432421287333e-05,
+      "loss": 3.0173,
+      "step": 13415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.384980479624815e-05,
+      "loss": 2.8814,
+      "step": 13420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.38452838273814e-05,
+      "loss": 2.9246,
+      "step": 13425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.384076130643316e-05,
+      "loss": 2.8912,
+      "step": 13430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.38362372335635e-05,
+      "loss": 2.8998,
+      "step": 13435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.383171160893258e-05,
+      "loss": 2.9777,
+      "step": 13440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.382718443270059e-05,
+      "loss": 2.8937,
+      "step": 13445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.382265570502782e-05,
+      "loss": 2.9361,
+      "step": 13450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.381812542607453e-05,
+      "loss": 3.0277,
+      "step": 13455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.381359359600113e-05,
+      "loss": 2.9214,
+      "step": 13460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.380906021496804e-05,
+      "loss": 2.8607,
+      "step": 13465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.380452528313573e-05,
+      "loss": 2.8195,
+      "step": 13470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.379998880066471e-05,
+      "loss": 2.9119,
+      "step": 13475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.379545076771559e-05,
+      "loss": 2.8277,
+      "step": 13480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.3790911184449e-05,
+      "loss": 2.9869,
+      "step": 13485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.378637005102566e-05,
+      "loss": 2.9783,
+      "step": 13490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.378182736760631e-05,
+      "loss": 2.9149,
+      "step": 13495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.377728313435175e-05,
+      "loss": 2.9838,
+      "step": 13500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.377273735142285e-05,
+      "loss": 2.9378,
+      "step": 13505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.37681900189805e-05,
+      "loss": 2.9773,
+      "step": 13510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.376364113718572e-05,
+      "loss": 2.9349,
+      "step": 13515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.375909070619949e-05,
+      "loss": 2.9028,
+      "step": 13520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.375453872618291e-05,
+      "loss": 2.9124,
+      "step": 13525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.374998519729713e-05,
+      "loss": 3.1045,
+      "step": 13530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.374543011970331e-05,
+      "loss": 3.0096,
+      "step": 13535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.374087349356273e-05,
+      "loss": 2.9488,
+      "step": 13540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.373631531903667e-05,
+      "loss": 2.9331,
+      "step": 13545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.373175559628648e-05,
+      "loss": 3.0448,
+      "step": 13550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.372719432547356e-05,
+      "loss": 2.8743,
+      "step": 13555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.372263150675942e-05,
+      "loss": 2.8849,
+      "step": 13560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.371806714030555e-05,
+      "loss": 2.904,
+      "step": 13565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.371350122627351e-05,
+      "loss": 2.9051,
+      "step": 13570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.370893376482495e-05,
+      "loss": 2.7986,
+      "step": 13575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.370436475612154e-05,
+      "loss": 3.0232,
+      "step": 13580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.369979420032504e-05,
+      "loss": 2.9499,
+      "step": 13585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.369522209759722e-05,
+      "loss": 3.0117,
+      "step": 13590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.369064844809996e-05,
+      "loss": 2.9082,
+      "step": 13595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.368607325199512e-05,
+      "loss": 3.0064,
+      "step": 13600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.368149650944468e-05,
+      "loss": 2.8824,
+      "step": 13605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.367691822061066e-05,
+      "loss": 2.9579,
+      "step": 13610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.36723383856551e-05,
+      "loss": 2.949,
+      "step": 13615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.366775700474015e-05,
+      "loss": 2.9366,
+      "step": 13620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.3663174078028e-05,
+      "loss": 2.8789,
+      "step": 13625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.365858960568082e-05,
+      "loss": 2.9052,
+      "step": 13630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.365400358786096e-05,
+      "loss": 3.0491,
+      "step": 13635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.364941602473072e-05,
+      "loss": 2.9668,
+      "step": 13640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.364482691645252e-05,
+      "loss": 3.0164,
+      "step": 13645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.36402362631888e-05,
+      "loss": 2.8438,
+      "step": 13650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.363564406510205e-05,
+      "loss": 2.9588,
+      "step": 13655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.363105032235486e-05,
+      "loss": 3.0156,
+      "step": 13660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.362645503510983e-05,
+      "loss": 2.9443,
+      "step": 13665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.362185820352961e-05,
+      "loss": 2.883,
+      "step": 13670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.361725982777696e-05,
+      "loss": 2.9835,
+      "step": 13675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.361265990801462e-05,
+      "loss": 2.9397,
+      "step": 13680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.360805844440546e-05,
+      "loss": 2.9104,
+      "step": 13685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.360345543711233e-05,
+      "loss": 3.0341,
+      "step": 13690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.359885088629821e-05,
+      "loss": 2.9115,
+      "step": 13695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.359424479212606e-05,
+      "loss": 2.993,
+      "step": 13700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.358963715475895e-05,
+      "loss": 2.8337,
+      "step": 13705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.358502797435999e-05,
+      "loss": 2.9328,
+      "step": 13710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.358041725109234e-05,
+      "loss": 2.9539,
+      "step": 13715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.357580498511922e-05,
+      "loss": 3.0128,
+      "step": 13720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.357119117660387e-05,
+      "loss": 2.9717,
+      "step": 13725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.356657582570966e-05,
+      "loss": 3.031,
+      "step": 13730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.356195893259992e-05,
+      "loss": 2.9898,
+      "step": 13735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.355734049743813e-05,
+      "loss": 2.9944,
+      "step": 13740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.355272052038775e-05,
+      "loss": 2.8582,
+      "step": 13745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.354809900161232e-05,
+      "loss": 2.9426,
+      "step": 13750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.354347594127544e-05,
+      "loss": 3.0563,
+      "step": 13755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.353885133954079e-05,
+      "loss": 2.8878,
+      "step": 13760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.353422519657205e-05,
+      "loss": 2.9917,
+      "step": 13765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.352959751253296e-05,
+      "loss": 2.8796,
+      "step": 13770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.35249682875874e-05,
+      "loss": 2.8915,
+      "step": 13775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.352033752189917e-05,
+      "loss": 2.9395,
+      "step": 13780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.351570521563224e-05,
+      "loss": 2.9959,
+      "step": 13785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.351107136895058e-05,
+      "loss": 2.9893,
+      "step": 13790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.350643598201821e-05,
+      "loss": 2.9525,
+      "step": 13795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.350179905499922e-05,
+      "loss": 2.9039,
+      "step": 13800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.34971605880578e-05,
+      "loss": 2.9076,
+      "step": 13805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.349252058135807e-05,
+      "loss": 2.9896,
+      "step": 13810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.348787903506432e-05,
+      "loss": 2.8044,
+      "step": 13815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.348323594934088e-05,
+      "loss": 2.928,
+      "step": 13820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.347859132435209e-05,
+      "loss": 2.9215,
+      "step": 13825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.347394516026233e-05,
+      "loss": 2.9881,
+      "step": 13830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.346929745723612e-05,
+      "loss": 2.9475,
+      "step": 13835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.346464821543796e-05,
+      "loss": 2.9181,
+      "step": 13840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.345999743503245e-05,
+      "loss": 2.956,
+      "step": 13845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.34553451161842e-05,
+      "loss": 2.9635,
+      "step": 13850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.345069125905791e-05,
+      "loss": 2.9695,
+      "step": 13855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.344603586381832e-05,
+      "loss": 2.8047,
+      "step": 13860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.34413789306302e-05,
+      "loss": 2.9932,
+      "step": 13865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.343672045965846e-05,
+      "loss": 2.9754,
+      "step": 13870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.343206045106796e-05,
+      "loss": 2.9318,
+      "step": 13875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.342739890502365e-05,
+      "loss": 2.9421,
+      "step": 13880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.34227358216906e-05,
+      "loss": 3.0255,
+      "step": 13885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.341807120123383e-05,
+      "loss": 3.0154,
+      "step": 13890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.341340504381847e-05,
+      "loss": 2.9246,
+      "step": 13895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.340873734960971e-05,
+      "loss": 2.9831,
+      "step": 13900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.340406811877278e-05,
+      "loss": 2.9363,
+      "step": 13905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.339939735147296e-05,
+      "loss": 2.9674,
+      "step": 13910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.339472504787559e-05,
+      "loss": 2.9225,
+      "step": 13915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.339005120814608e-05,
+      "loss": 2.9599,
+      "step": 13920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.338537583244987e-05,
+      "loss": 3.0277,
+      "step": 13925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.338069892095243e-05,
+      "loss": 2.965,
+      "step": 13930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.33760204738194e-05,
+      "loss": 3.0186,
+      "step": 13935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.33713404912163e-05,
+      "loss": 2.9302,
+      "step": 13940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.336665897330886e-05,
+      "loss": 2.8358,
+      "step": 13945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.336197592026279e-05,
+      "loss": 2.9854,
+      "step": 13950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.335729133224385e-05,
+      "loss": 2.9197,
+      "step": 13955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.335260520941787e-05,
+      "loss": 2.9328,
+      "step": 13960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.334791755195073e-05,
+      "loss": 2.9591,
+      "step": 13965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.334322836000841e-05,
+      "loss": 2.9394,
+      "step": 13970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.333853763375686e-05,
+      "loss": 3.0204,
+      "step": 13975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.333384537336214e-05,
+      "loss": 2.938,
+      "step": 13980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.332915157899035e-05,
+      "loss": 3.0389,
+      "step": 13985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.332445625080765e-05,
+      "loss": 2.8141,
+      "step": 13990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.331975938898024e-05,
+      "loss": 2.9602,
+      "step": 13995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.33150609936744e-05,
+      "loss": 2.9077,
+      "step": 14000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.331036106505643e-05,
+      "loss": 2.886,
+      "step": 14005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.330565960329274e-05,
+      "loss": 2.8787,
+      "step": 14010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.330095660854972e-05,
+      "loss": 2.9073,
+      "step": 14015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.329625208099386e-05,
+      "loss": 2.9442,
+      "step": 14020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.329154602079169e-05,
+      "loss": 2.9473,
+      "step": 14025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.328683842810982e-05,
+      "loss": 2.9865,
+      "step": 14030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.328212930311488e-05,
+      "loss": 2.7903,
+      "step": 14035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.327741864597357e-05,
+      "loss": 2.8672,
+      "step": 14040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.327270645685265e-05,
+      "loss": 2.8683,
+      "step": 14045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.326799273591891e-05,
+      "loss": 2.9498,
+      "step": 14050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.326327748333925e-05,
+      "loss": 2.9405,
+      "step": 14055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.325856069928054e-05,
+      "loss": 2.9567,
+      "step": 14060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.325384238390976e-05,
+      "loss": 2.9127,
+      "step": 14065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.324912253739396e-05,
+      "loss": 2.9406,
+      "step": 14070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.324440115990018e-05,
+      "loss": 2.9431,
+      "step": 14075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.323967825159559e-05,
+      "loss": 2.9111,
+      "step": 14080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.323495381264735e-05,
+      "loss": 2.9137,
+      "step": 14085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.323022784322271e-05,
+      "loss": 2.894,
+      "step": 14090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.322550034348896e-05,
+      "loss": 2.8935,
+      "step": 14095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.322077131361346e-05,
+      "loss": 2.879,
+      "step": 14100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.32160407537636e-05,
+      "loss": 3.0195,
+      "step": 14105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.321130866410686e-05,
+      "loss": 2.9287,
+      "step": 14110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.320657504481072e-05,
+      "loss": 3.0074,
+      "step": 14115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.320183989604276e-05,
+      "loss": 2.9227,
+      "step": 14120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.319710321797061e-05,
+      "loss": 2.8256,
+      "step": 14125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.319236501076193e-05,
+      "loss": 2.9668,
+      "step": 14130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.318762527458444e-05,
+      "loss": 2.912,
+      "step": 14135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.318288400960596e-05,
+      "loss": 2.8595,
+      "step": 14140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.317814121599428e-05,
+      "loss": 3.0032,
+      "step": 14145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.317339689391733e-05,
+      "loss": 2.9396,
+      "step": 14150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.316865104354302e-05,
+      "loss": 2.8762,
+      "step": 14155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.316390366503938e-05,
+      "loss": 2.9351,
+      "step": 14160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.315915475857444e-05,
+      "loss": 3.0779,
+      "step": 14165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.31544043243163e-05,
+      "loss": 2.9766,
+      "step": 14170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.314965236243314e-05,
+      "loss": 2.9511,
+      "step": 14175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.314489887309319e-05,
+      "loss": 2.8955,
+      "step": 14180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.314014385646468e-05,
+      "loss": 2.955,
+      "step": 14185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.313538731271597e-05,
+      "loss": 2.908,
+      "step": 14190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.313062924201541e-05,
+      "loss": 2.9756,
+      "step": 14195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.312586964453142e-05,
+      "loss": 2.9552,
+      "step": 14200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.312110852043254e-05,
+      "loss": 3.0811,
+      "step": 14205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.311634586988725e-05,
+      "loss": 2.9374,
+      "step": 14210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.311158169306417e-05,
+      "loss": 2.9746,
+      "step": 14215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.310681599013196e-05,
+      "loss": 2.828,
+      "step": 14220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.31020487612593e-05,
+      "loss": 2.974,
+      "step": 14225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.309728000661496e-05,
+      "loss": 2.839,
+      "step": 14230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.309250972636773e-05,
+      "loss": 2.9631,
+      "step": 14235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.30877379206865e-05,
+      "loss": 2.9547,
+      "step": 14240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.308296458974016e-05,
+      "loss": 2.9694,
+      "step": 14245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.307818973369769e-05,
+      "loss": 2.9364,
+      "step": 14250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.307341335272813e-05,
+      "loss": 2.9864,
+      "step": 14255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.306863544700054e-05,
+      "loss": 2.9555,
+      "step": 14260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.306385601668406e-05,
+      "loss": 2.8305,
+      "step": 14265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.305907506194789e-05,
+      "loss": 2.9645,
+      "step": 14270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.305429258296126e-05,
+      "loss": 2.9699,
+      "step": 14275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.304950857989345e-05,
+      "loss": 2.9332,
+      "step": 14280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.304472305291385e-05,
+      "loss": 2.9045,
+      "step": 14285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.303993600219183e-05,
+      "loss": 2.9763,
+      "step": 14290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.303514742789684e-05,
+      "loss": 2.9229,
+      "step": 14295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.303035733019841e-05,
+      "loss": 2.9579,
+      "step": 14300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.30255657092661e-05,
+      "loss": 2.9914,
+      "step": 14305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.302077256526956e-05,
+      "loss": 2.9082,
+      "step": 14310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.301597789837839e-05,
+      "loss": 2.9591,
+      "step": 14315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.301118170876238e-05,
+      "loss": 2.9183,
+      "step": 14320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.300638399659129e-05,
+      "loss": 2.9638,
+      "step": 14325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.300158476203495e-05,
+      "loss": 2.967,
+      "step": 14330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.299678400526327e-05,
+      "loss": 2.9902,
+      "step": 14335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.299198172644617e-05,
+      "loss": 3.0147,
+      "step": 14340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.298717792575363e-05,
+      "loss": 2.9168,
+      "step": 14345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.298237260335573e-05,
+      "loss": 2.9072,
+      "step": 14350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.297756575942257e-05,
+      "loss": 2.8257,
+      "step": 14355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.29727573941243e-05,
+      "loss": 2.8491,
+      "step": 14360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.296794750763113e-05,
+      "loss": 2.974,
+      "step": 14365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.296313610011334e-05,
+      "loss": 2.9074,
+      "step": 14370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.295832317174124e-05,
+      "loss": 2.9172,
+      "step": 14375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.29535087226852e-05,
+      "loss": 2.9104,
+      "step": 14380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.294869275311565e-05,
+      "loss": 2.9259,
+      "step": 14385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.294387526320306e-05,
+      "loss": 3.009,
+      "step": 14390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.293905625311797e-05,
+      "loss": 2.9427,
+      "step": 14395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.293423572303098e-05,
+      "loss": 2.9523,
+      "step": 14400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.292941367311274e-05,
+      "loss": 2.8794,
+      "step": 14405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.292459010353392e-05,
+      "loss": 2.9788,
+      "step": 14410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.291976501446526e-05,
+      "loss": 2.9354,
+      "step": 14415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.291493840607761e-05,
+      "loss": 2.8922,
+      "step": 14420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.29101102785418e-05,
+      "loss": 2.7938,
+      "step": 14425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.290528063202873e-05,
+      "loss": 2.9077,
+      "step": 14430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.29004494667094e-05,
+      "loss": 2.9586,
+      "step": 14435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.289561678275478e-05,
+      "loss": 2.9314,
+      "step": 14440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.289078258033598e-05,
+      "loss": 2.877,
+      "step": 14445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.28859468596241e-05,
+      "loss": 2.9165,
+      "step": 14450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.288110962079036e-05,
+      "loss": 2.9387,
+      "step": 14455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.287627086400597e-05,
+      "loss": 3.0056,
+      "step": 14460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.28714305894422e-05,
+      "loss": 2.8697,
+      "step": 14465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.286658879727041e-05,
+      "loss": 2.9728,
+      "step": 14470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.286174548766199e-05,
+      "loss": 2.9267,
+      "step": 14475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.28569006607884e-05,
+      "loss": 2.8657,
+      "step": 14480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.285205431682111e-05,
+      "loss": 3.0193,
+      "step": 14485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.284720645593172e-05,
+      "loss": 2.9511,
+      "step": 14490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.284235707829182e-05,
+      "loss": 2.9682,
+      "step": 14495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.283750618407307e-05,
+      "loss": 2.9607,
+      "step": 14500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.283265377344718e-05,
+      "loss": 2.9192,
+      "step": 14505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.282779984658593e-05,
+      "loss": 2.9571,
+      "step": 14510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.282294440366115e-05,
+      "loss": 2.8831,
+      "step": 14515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.281808744484472e-05,
+      "loss": 2.9769,
+      "step": 14520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.281322897030855e-05,
+      "loss": 2.9284,
+      "step": 14525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.280836898022465e-05,
+      "loss": 2.9662,
+      "step": 14530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.280350747476504e-05,
+      "loss": 2.8315,
+      "step": 14535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.279864445410182e-05,
+      "loss": 2.9263,
+      "step": 14540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.279377991840714e-05,
+      "loss": 3.0032,
+      "step": 14545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.27889138678532e-05,
+      "loss": 2.9528,
+      "step": 14550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.278404630261226e-05,
+      "loss": 2.9251,
+      "step": 14555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.27791772228566e-05,
+      "loss": 2.9082,
+      "step": 14560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.277430662875861e-05,
+      "loss": 2.9489,
+      "step": 14565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.27694345204907e-05,
+      "loss": 2.9801,
+      "step": 14570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.276456089822531e-05,
+      "loss": 2.9306,
+      "step": 14575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.2759685762135e-05,
+      "loss": 3.0334,
+      "step": 14580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.275480911239232e-05,
+      "loss": 2.9639,
+      "step": 14585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.27499309491699e-05,
+      "loss": 2.8602,
+      "step": 14590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.274505127264044e-05,
+      "loss": 2.9027,
+      "step": 14595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.274017008297666e-05,
+      "loss": 2.9472,
+      "step": 14600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.273528738035137e-05,
+      "loss": 2.9809,
+      "step": 14605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.273040316493738e-05,
+      "loss": 2.9602,
+      "step": 14610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.27255174369076e-05,
+      "loss": 2.957,
+      "step": 14615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.2720630196435e-05,
+      "loss": 2.9763,
+      "step": 14620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.271574144369257e-05,
+      "loss": 2.956,
+      "step": 14625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.271085117885335e-05,
+      "loss": 2.946,
+      "step": 14630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.270595940209047e-05,
+      "loss": 2.9657,
+      "step": 14635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.27010661135771e-05,
+      "loss": 2.9812,
+      "step": 14640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.269617131348645e-05,
+      "loss": 2.9085,
+      "step": 14645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.269127500199178e-05,
+      "loss": 3.01,
+      "step": 14650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.268637717926643e-05,
+      "loss": 2.9573,
+      "step": 14655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.268147784548377e-05,
+      "loss": 2.9977,
+      "step": 14660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.267657700081722e-05,
+      "loss": 2.8105,
+      "step": 14665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.267167464544031e-05,
+      "loss": 2.9152,
+      "step": 14670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.266677077952652e-05,
+      "loss": 2.9092,
+      "step": 14675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.266186540324948e-05,
+      "loss": 2.8733,
+      "step": 14680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.265695851678285e-05,
+      "loss": 2.9132,
+      "step": 14685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.265205012030026e-05,
+      "loss": 2.8413,
+      "step": 14690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.264714021397553e-05,
+      "loss": 2.9683,
+      "step": 14695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.264222879798246e-05,
+      "loss": 2.8097,
+      "step": 14700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.263731587249487e-05,
+      "loss": 2.9387,
+      "step": 14705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.263240143768671e-05,
+      "loss": 2.9919,
+      "step": 14710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.262748549373192e-05,
+      "loss": 2.9614,
+      "step": 14715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.262256804080455e-05,
+      "loss": 2.9194,
+      "step": 14720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.261764907907864e-05,
+      "loss": 2.9319,
+      "step": 14725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.261272860872834e-05,
+      "loss": 2.9528,
+      "step": 14730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.260780662992781e-05,
+      "loss": 2.9097,
+      "step": 14735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.260288314285131e-05,
+      "loss": 2.9862,
+      "step": 14740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.259795814767311e-05,
+      "loss": 2.9695,
+      "step": 14745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.259303164456755e-05,
+      "loss": 2.974,
+      "step": 14750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.258810363370902e-05,
+      "loss": 2.9084,
+      "step": 14755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.258317411527198e-05,
+      "loss": 2.9159,
+      "step": 14760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.257824308943093e-05,
+      "loss": 2.9982,
+      "step": 14765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.257331055636041e-05,
+      "loss": 2.9918,
+      "step": 14770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.256837651623502e-05,
+      "loss": 2.9774,
+      "step": 14775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.256344096922946e-05,
+      "loss": 2.9003,
+      "step": 14780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.255850391551841e-05,
+      "loss": 2.8171,
+      "step": 14785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.255356535527664e-05,
+      "loss": 2.9106,
+      "step": 14790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.254862528867896e-05,
+      "loss": 2.9817,
+      "step": 14795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.254368371590027e-05,
+      "loss": 2.948,
+      "step": 14800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.25387406371155e-05,
+      "loss": 2.9334,
+      "step": 14805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.253379605249961e-05,
+      "loss": 2.9221,
+      "step": 14810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.252884996222764e-05,
+      "loss": 2.9545,
+      "step": 14815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.252390236647467e-05,
+      "loss": 2.9166,
+      "step": 14820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.251895326541585e-05,
+      "loss": 2.9297,
+      "step": 14825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.251400265922636e-05,
+      "loss": 2.8812,
+      "step": 14830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.250905054808147e-05,
+      "loss": 2.942,
+      "step": 14835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.250409693215647e-05,
+      "loss": 2.8587,
+      "step": 14840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.249914181162671e-05,
+      "loss": 2.9501,
+      "step": 14845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.24941851866676e-05,
+      "loss": 2.9758,
+      "step": 14850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.248922705745457e-05,
+      "loss": 2.9134,
+      "step": 14855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.248426742416318e-05,
+      "loss": 2.8819,
+      "step": 14860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.247930628696898e-05,
+      "loss": 2.9942,
+      "step": 14865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.247434364604758e-05,
+      "loss": 2.8746,
+      "step": 14870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.246937950157465e-05,
+      "loss": 3.0048,
+      "step": 14875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.246441385372595e-05,
+      "loss": 3.0513,
+      "step": 14880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.245944670267721e-05,
+      "loss": 2.8801,
+      "step": 14885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.24544780486043e-05,
+      "loss": 2.996,
+      "step": 14890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.244950789168309e-05,
+      "loss": 2.8584,
+      "step": 14895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.244453623208952e-05,
+      "loss": 3.0324,
+      "step": 14900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.243956306999958e-05,
+      "loss": 2.7713,
+      "step": 14905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.243458840558931e-05,
+      "loss": 3.0168,
+      "step": 14910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.242961223903484e-05,
+      "loss": 2.987,
+      "step": 14915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.242463457051227e-05,
+      "loss": 2.8883,
+      "step": 14920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.241965540019786e-05,
+      "loss": 2.764,
+      "step": 14925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.241467472826782e-05,
+      "loss": 2.9565,
+      "step": 14930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.240969255489848e-05,
+      "loss": 2.9353,
+      "step": 14935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.240470888026622e-05,
+      "loss": 2.9262,
+      "step": 14940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.239972370454744e-05,
+      "loss": 3.0255,
+      "step": 14945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.239473702791862e-05,
+      "loss": 2.9799,
+      "step": 14950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.238974885055627e-05,
+      "loss": 2.9385,
+      "step": 14955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.238475917263697e-05,
+      "loss": 2.9515,
+      "step": 14960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.237976799433736e-05,
+      "loss": 2.9151,
+      "step": 14965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.237477531583411e-05,
+      "loss": 2.8619,
+      "step": 14970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.236978113730398e-05,
+      "loss": 2.9448,
+      "step": 14975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.236478545892373e-05,
+      "loss": 2.9953,
+      "step": 14980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.235978828087021e-05,
+      "loss": 2.8972,
+      "step": 14985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.235478960332034e-05,
+      "loss": 2.9712,
+      "step": 14990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.234978942645103e-05,
+      "loss": 2.8493,
+      "step": 14995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.234478775043931e-05,
+      "loss": 2.8906,
+      "step": 15000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.233978457546221e-05,
+      "loss": 2.961,
+      "step": 15005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.233477990169687e-05,
+      "loss": 2.9949,
+      "step": 15010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.232977372932042e-05,
+      "loss": 2.9403,
+      "step": 15015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.232476605851009e-05,
+      "loss": 2.9869,
+      "step": 15020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.231975688944314e-05,
+      "loss": 2.8974,
+      "step": 15025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.231474622229691e-05,
+      "loss": 2.9566,
+      "step": 15030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.230973405724874e-05,
+      "loss": 3.0092,
+      "step": 15035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.230472039447609e-05,
+      "loss": 2.8841,
+      "step": 15040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.22997052341564e-05,
+      "loss": 2.9425,
+      "step": 15045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.229468857646724e-05,
+      "loss": 2.9114,
+      "step": 15050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.228967042158616e-05,
+      "loss": 2.9676,
+      "step": 15055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.228465076969082e-05,
+      "loss": 2.9215,
+      "step": 15060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.227962962095891e-05,
+      "loss": 2.9756,
+      "step": 15065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.227460697556817e-05,
+      "loss": 2.9554,
+      "step": 15070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.22695828336964e-05,
+      "loss": 2.9057,
+      "step": 15075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.226455719552145e-05,
+      "loss": 2.9325,
+      "step": 15080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.225953006122122e-05,
+      "loss": 2.8408,
+      "step": 15085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.225450143097368e-05,
+      "loss": 2.8454,
+      "step": 15090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.22494713049568e-05,
+      "loss": 2.8853,
+      "step": 15095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.224443968334868e-05,
+      "loss": 2.9163,
+      "step": 15100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.223940656632741e-05,
+      "loss": 2.9231,
+      "step": 15105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.223437195407118e-05,
+      "loss": 3.0233,
+      "step": 15110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.222933584675819e-05,
+      "loss": 2.853,
+      "step": 15115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.222429824456673e-05,
+      "loss": 2.8628,
+      "step": 15120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.221925914767511e-05,
+      "loss": 2.912,
+      "step": 15125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.221421855626173e-05,
+      "loss": 2.8777,
+      "step": 15130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.2209176470505e-05,
+      "loss": 2.9955,
+      "step": 15135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.220413289058343e-05,
+      "loss": 2.9233,
+      "step": 15140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.219908781667552e-05,
+      "loss": 2.9397,
+      "step": 15145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.21940412489599e-05,
+      "loss": 2.9256,
+      "step": 15150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.21889931876152e-05,
+      "loss": 2.932,
+      "step": 15155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.218394363282011e-05,
+      "loss": 2.9302,
+      "step": 15160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.217889258475338e-05,
+      "loss": 3.0037,
+      "step": 15165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.217384004359382e-05,
+      "loss": 2.9361,
+      "step": 15170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.216878600952029e-05,
+      "loss": 2.9368,
+      "step": 15175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.216373048271169e-05,
+      "loss": 2.8353,
+      "step": 15180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.215867346334697e-05,
+      "loss": 2.9201,
+      "step": 15185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.215361495160515e-05,
+      "loss": 2.9546,
+      "step": 15190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.214855494766531e-05,
+      "loss": 2.9288,
+      "step": 15195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.214349345170657e-05,
+      "loss": 2.8494,
+      "step": 15200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.213843046390807e-05,
+      "loss": 2.8985,
+      "step": 15205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.213336598444907e-05,
+      "loss": 2.9264,
+      "step": 15210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.212830001350882e-05,
+      "loss": 3.0494,
+      "step": 15215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.212323255126668e-05,
+      "loss": 2.9286,
+      "step": 15220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.211816359790202e-05,
+      "loss": 3.0427,
+      "step": 15225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.211309315359427e-05,
+      "loss": 2.9139,
+      "step": 15230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.210802121852292e-05,
+      "loss": 2.8337,
+      "step": 15235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.210294779286753e-05,
+      "loss": 3.011,
+      "step": 15240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.209787287680765e-05,
+      "loss": 2.9036,
+      "step": 15245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.209279647052298e-05,
+      "loss": 3.0997,
+      "step": 15250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.20877185741932e-05,
+      "loss": 2.9286,
+      "step": 15255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.208263918799805e-05,
+      "loss": 2.937,
+      "step": 15260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.207755831211736e-05,
+      "loss": 2.9338,
+      "step": 15265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.207247594673097e-05,
+      "loss": 3.0022,
+      "step": 15270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.206739209201878e-05,
+      "loss": 2.9064,
+      "step": 15275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.206230674816077e-05,
+      "loss": 2.9378,
+      "step": 15280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.205721991533696e-05,
+      "loss": 2.9773,
+      "step": 15285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.20521315937274e-05,
+      "loss": 3.0411,
+      "step": 15290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.204704178351225e-05,
+      "loss": 3.0704,
+      "step": 15295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.204195048487166e-05,
+      "loss": 3.0279,
+      "step": 15300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.203685769798585e-05,
+      "loss": 2.9279,
+      "step": 15305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.203176342303509e-05,
+      "loss": 2.9578,
+      "step": 15310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.202666766019974e-05,
+      "loss": 2.9179,
+      "step": 15315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.202157040966017e-05,
+      "loss": 2.9106,
+      "step": 15320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.201647167159683e-05,
+      "loss": 2.9848,
+      "step": 15325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.201137144619021e-05,
+      "loss": 2.9375,
+      "step": 15330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.200626973362083e-05,
+      "loss": 2.9654,
+      "step": 15335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.200116653406932e-05,
+      "loss": 2.8364,
+      "step": 15340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.19960618477163e-05,
+      "loss": 2.9179,
+      "step": 15345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.199095567474247e-05,
+      "loss": 3.0031,
+      "step": 15350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.198584801532861e-05,
+      "loss": 2.9023,
+      "step": 15355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.19807388696555e-05,
+      "loss": 2.922,
+      "step": 15360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.197562823790403e-05,
+      "loss": 2.9864,
+      "step": 15365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.197051612025507e-05,
+      "loss": 2.8911,
+      "step": 15370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.196540251688962e-05,
+      "loss": 2.9754,
+      "step": 15375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.196028742798866e-05,
+      "loss": 2.8905,
+      "step": 15380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.195517085373331e-05,
+      "loss": 2.8404,
+      "step": 15385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.195005279430464e-05,
+      "loss": 2.9181,
+      "step": 15390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.194493324988387e-05,
+      "loss": 2.9441,
+      "step": 15395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.193981222065219e-05,
+      "loss": 2.973,
+      "step": 15400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.19346897067909e-05,
+      "loss": 2.9155,
+      "step": 15405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.192956570848133e-05,
+      "loss": 2.8937,
+      "step": 15410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.192444022590485e-05,
+      "loss": 2.9179,
+      "step": 15415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.191931325924293e-05,
+      "loss": 2.9219,
+      "step": 15420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.191418480867702e-05,
+      "loss": 2.9941,
+      "step": 15425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.190905487438867e-05,
+      "loss": 2.8933,
+      "step": 15430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.190392345655952e-05,
+      "loss": 2.932,
+      "step": 15435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.189879055537116e-05,
+      "loss": 2.9119,
+      "step": 15440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.189365617100532e-05,
+      "loss": 3.0242,
+      "step": 15445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.188852030364373e-05,
+      "loss": 2.9125,
+      "step": 15450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.188338295346823e-05,
+      "loss": 2.9319,
+      "step": 15455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.187824412066063e-05,
+      "loss": 2.8813,
+      "step": 15460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.18731038054029e-05,
+      "loss": 2.9736,
+      "step": 15465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.186796200787696e-05,
+      "loss": 2.9609,
+      "step": 15470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.186281872826483e-05,
+      "loss": 3.0618,
+      "step": 15475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.185767396674858e-05,
+      "loss": 2.9837,
+      "step": 15480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.185252772351035e-05,
+      "loss": 2.8929,
+      "step": 15485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.184737999873226e-05,
+      "loss": 2.8522,
+      "step": 15490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.184223079259657e-05,
+      "loss": 2.9638,
+      "step": 15495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.183708010528557e-05,
+      "loss": 2.9101,
+      "step": 15500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.183192793698159e-05,
+      "loss": 2.8322,
+      "step": 15505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.182677428786696e-05,
+      "loss": 2.8466,
+      "step": 15510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.182161915812419e-05,
+      "loss": 2.9365,
+      "step": 15515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.18164625479357e-05,
+      "loss": 2.91,
+      "step": 15520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.181130445748406e-05,
+      "loss": 2.8573,
+      "step": 15525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.180614488695185e-05,
+      "loss": 2.841,
+      "step": 15530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.180098383652174e-05,
+      "loss": 2.9108,
+      "step": 15535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.179582130637641e-05,
+      "loss": 2.9807,
+      "step": 15540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.17906572966986e-05,
+      "loss": 3.0578,
+      "step": 15545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.178549180767112e-05,
+      "loss": 2.8689,
+      "step": 15550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.178032483947679e-05,
+      "loss": 2.9895,
+      "step": 15555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.177515639229857e-05,
+      "loss": 2.8971,
+      "step": 15560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.17699864663194e-05,
+      "loss": 2.952,
+      "step": 15565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.176481506172228e-05,
+      "loss": 2.9863,
+      "step": 15570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.175964217869028e-05,
+      "loss": 2.9929,
+      "step": 15575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.175446781740653e-05,
+      "loss": 2.9391,
+      "step": 15580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.174929197805415e-05,
+      "loss": 2.9593,
+      "step": 15585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.174411466081641e-05,
+      "loss": 2.9581,
+      "step": 15590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.173893586587656e-05,
+      "loss": 2.8483,
+      "step": 15595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.173375559341792e-05,
+      "loss": 2.913,
+      "step": 15600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.172857384362389e-05,
+      "loss": 2.9207,
+      "step": 15605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.172339061667787e-05,
+      "loss": 2.984,
+      "step": 15610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.171820591276337e-05,
+      "loss": 2.8685,
+      "step": 15615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.17130197320639e-05,
+      "loss": 2.987,
+      "step": 15620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.170783207476304e-05,
+      "loss": 2.964,
+      "step": 15625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.170264294104445e-05,
+      "loss": 2.9056,
+      "step": 15630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.169745233109181e-05,
+      "loss": 2.9786,
+      "step": 15635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.169226024508888e-05,
+      "loss": 2.9742,
+      "step": 15640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.168706668321943e-05,
+      "loss": 2.8908,
+      "step": 15645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.168187164566731e-05,
+      "loss": 2.9141,
+      "step": 15650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.167667513261645e-05,
+      "loss": 2.9388,
+      "step": 15655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.167147714425076e-05,
+      "loss": 2.9453,
+      "step": 15660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.166627768075426e-05,
+      "loss": 2.9419,
+      "step": 15665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.166107674231101e-05,
+      "loss": 2.9878,
+      "step": 15670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.165587432910512e-05,
+      "loss": 2.9843,
+      "step": 15675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.165067044132074e-05,
+      "loss": 2.9961,
+      "step": 15680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.164546507914209e-05,
+      "loss": 2.9593,
+      "step": 15685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.164025824275344e-05,
+      "loss": 2.8919,
+      "step": 15690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.16350499323391e-05,
+      "loss": 2.9093,
+      "step": 15695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.162984014808344e-05,
+      "loss": 2.8986,
+      "step": 15700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.16246288901709e-05,
+      "loss": 2.9402,
+      "step": 15705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.161941615878592e-05,
+      "loss": 3.0102,
+      "step": 15710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.161420195411304e-05,
+      "loss": 2.8924,
+      "step": 15715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.160898627633685e-05,
+      "loss": 2.9894,
+      "step": 15720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.160376912564196e-05,
+      "loss": 3.0052,
+      "step": 15725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.159855050221307e-05,
+      "loss": 2.9993,
+      "step": 15730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.159333040623489e-05,
+      "loss": 2.8759,
+      "step": 15735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.158810883789224e-05,
+      "loss": 2.8303,
+      "step": 15740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.158288579736993e-05,
+      "loss": 2.9309,
+      "step": 15745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.157766128485287e-05,
+      "loss": 2.9212,
+      "step": 15750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.157243530052598e-05,
+      "loss": 3.04,
+      "step": 15755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.156720784457429e-05,
+      "loss": 2.8737,
+      "step": 15760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.156197891718281e-05,
+      "loss": 2.9661,
+      "step": 15765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.155674851853667e-05,
+      "loss": 3.0051,
+      "step": 15770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.1551516648821e-05,
+      "loss": 3.0096,
+      "step": 15775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.1546283308221e-05,
+      "loss": 2.8828,
+      "step": 15780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.154104849692196e-05,
+      "loss": 2.8534,
+      "step": 15785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.153581221510914e-05,
+      "loss": 2.8749,
+      "step": 15790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.153057446296793e-05,
+      "loss": 2.8648,
+      "step": 15795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.152533524068373e-05,
+      "loss": 2.9858,
+      "step": 15800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.152009454844201e-05,
+      "loss": 2.9356,
+      "step": 15805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.151485238642828e-05,
+      "loss": 2.8929,
+      "step": 15810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.150960875482812e-05,
+      "loss": 2.9797,
+      "step": 15815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.150436365382713e-05,
+      "loss": 3.0396,
+      "step": 15820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.1499117083611e-05,
+      "loss": 2.9477,
+      "step": 15825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.149386904436543e-05,
+      "loss": 2.9867,
+      "step": 15830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.148861953627622e-05,
+      "loss": 2.8888,
+      "step": 15835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.148336855952919e-05,
+      "loss": 2.9714,
+      "step": 15840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.14781161143102e-05,
+      "loss": 2.891,
+      "step": 15845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.147286220080523e-05,
+      "loss": 2.9494,
+      "step": 15850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.146760681920022e-05,
+      "loss": 2.9977,
+      "step": 15855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.146234996968122e-05,
+      "loss": 2.9797,
+      "step": 15860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.145709165243431e-05,
+      "loss": 2.9311,
+      "step": 15865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.145183186764564e-05,
+      "loss": 3.0542,
+      "step": 15870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.144657061550139e-05,
+      "loss": 2.8873,
+      "step": 15875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.144130789618781e-05,
+      "loss": 2.9519,
+      "step": 15880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.143604370989121e-05,
+      "loss": 2.9664,
+      "step": 15885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.143077805679791e-05,
+      "loss": 3.0324,
+      "step": 15890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.142551093709432e-05,
+      "loss": 2.9724,
+      "step": 15895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.14202423509669e-05,
+      "loss": 2.9993,
+      "step": 15900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.141497229860215e-05,
+      "loss": 2.9841,
+      "step": 15905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.140970078018661e-05,
+      "loss": 2.9007,
+      "step": 15910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.14044277959069e-05,
+      "loss": 3.0222,
+      "step": 15915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.139915334594968e-05,
+      "loss": 2.952,
+      "step": 15920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.139387743050165e-05,
+      "loss": 2.8399,
+      "step": 15925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.138860004974958e-05,
+      "loss": 3.0717,
+      "step": 15930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.13833212038803e-05,
+      "loss": 2.9897,
+      "step": 15935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.137804089308065e-05,
+      "loss": 2.9657,
+      "step": 15940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.137275911753755e-05,
+      "loss": 2.8923,
+      "step": 15945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.136747587743798e-05,
+      "loss": 2.9557,
+      "step": 15950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.136219117296896e-05,
+      "loss": 2.9011,
+      "step": 15955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.135690500431756e-05,
+      "loss": 2.9298,
+      "step": 15960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.13516173716709e-05,
+      "loss": 3.0951,
+      "step": 15965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.134632827521619e-05,
+      "loss": 2.9125,
+      "step": 15970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.134103771514063e-05,
+      "loss": 2.9048,
+      "step": 15975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.133574569163148e-05,
+      "loss": 2.8726,
+      "step": 15980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.13304522048761e-05,
+      "loss": 2.9678,
+      "step": 15985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.13251572550619e-05,
+      "loss": 2.9156,
+      "step": 15990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.131986084237628e-05,
+      "loss": 2.9371,
+      "step": 15995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.131456296700671e-05,
+      "loss": 2.9469,
+      "step": 16000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.130926362914078e-05,
+      "loss": 3.0657,
+      "step": 16005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.130396282896605e-05,
+      "loss": 2.9067,
+      "step": 16010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.129866056667018e-05,
+      "loss": 2.9266,
+      "step": 16015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.129335684244084e-05,
+      "loss": 2.9034,
+      "step": 16020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.12880516564658e-05,
+      "loss": 2.8654,
+      "step": 16025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.128274500893285e-05,
+      "loss": 2.9973,
+      "step": 16030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.127743690002984e-05,
+      "loss": 2.9576,
+      "step": 16035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.127212732994467e-05,
+      "loss": 2.9746,
+      "step": 16040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.126681629886531e-05,
+      "loss": 2.9641,
+      "step": 16045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.126150380697976e-05,
+      "loss": 2.998,
+      "step": 16050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.125618985447605e-05,
+      "loss": 2.9941,
+      "step": 16055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.125087444154231e-05,
+      "loss": 3.0132,
+      "step": 16060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.12455575683667e-05,
+      "loss": 3.0265,
+      "step": 16065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.124023923513745e-05,
+      "loss": 2.9752,
+      "step": 16070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.12349194420428e-05,
+      "loss": 3.0433,
+      "step": 16075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.122959818927104e-05,
+      "loss": 3.037,
+      "step": 16080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.122427547701061e-05,
+      "loss": 2.9674,
+      "step": 16085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.121895130544987e-05,
+      "loss": 2.9102,
+      "step": 16090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.12136256747773e-05,
+      "loss": 2.9204,
+      "step": 16095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.120829858518143e-05,
+      "loss": 2.9481,
+      "step": 16100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.120297003685084e-05,
+      "loss": 2.9934,
+      "step": 16105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.119764002997416e-05,
+      "loss": 2.8896,
+      "step": 16110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.119230856474003e-05,
+      "loss": 2.9381,
+      "step": 16115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.118697564133723e-05,
+      "loss": 2.9422,
+      "step": 16120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.118164125995451e-05,
+      "loss": 2.8557,
+      "step": 16125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.11763054207807e-05,
+      "loss": 3.0063,
+      "step": 16130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.117096812400472e-05,
+      "loss": 2.9338,
+      "step": 16135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.116562936981546e-05,
+      "loss": 2.9154,
+      "step": 16140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.116028915840191e-05,
+      "loss": 2.869,
+      "step": 16145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.115494748995313e-05,
+      "loss": 2.9537,
+      "step": 16150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.11496043646582e-05,
+      "loss": 2.8548,
+      "step": 16155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.114425978270629e-05,
+      "loss": 2.9457,
+      "step": 16160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.113891374428654e-05,
+      "loss": 2.9328,
+      "step": 16165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.113356624958823e-05,
+      "loss": 2.9252,
+      "step": 16170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.112821729880066e-05,
+      "loss": 3.0027,
+      "step": 16175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.112286689211315e-05,
+      "loss": 2.909,
+      "step": 16180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.111751502971511e-05,
+      "loss": 2.9483,
+      "step": 16185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.111216171179602e-05,
+      "loss": 2.9102,
+      "step": 16190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.110680693854534e-05,
+      "loss": 2.9518,
+      "step": 16195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.110145071015264e-05,
+      "loss": 2.9563,
+      "step": 16200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.109609302680753e-05,
+      "loss": 2.8498,
+      "step": 16205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.109073388869967e-05,
+      "loss": 2.9832,
+      "step": 16210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.108537329601876e-05,
+      "loss": 3.0397,
+      "step": 16215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.108001124895458e-05,
+      "loss": 2.9087,
+      "step": 16220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.10746477476969e-05,
+      "loss": 2.9723,
+      "step": 16225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.106928279243562e-05,
+      "loss": 2.9147,
+      "step": 16230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.106391638336064e-05,
+      "loss": 3.0307,
+      "step": 16235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.105854852066192e-05,
+      "loss": 2.954,
+      "step": 16240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.10531792045295e-05,
+      "loss": 2.9295,
+      "step": 16245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.104780843515341e-05,
+      "loss": 2.9228,
+      "step": 16250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.104243621272382e-05,
+      "loss": 2.9379,
+      "step": 16255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.103706253743086e-05,
+      "loss": 2.9454,
+      "step": 16260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.103168740946478e-05,
+      "loss": 3.0764,
+      "step": 16265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.102631082901582e-05,
+      "loss": 2.9542,
+      "step": 16270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.102093279627435e-05,
+      "loss": 2.9812,
+      "step": 16275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.101555331143073e-05,
+      "loss": 2.9148,
+      "step": 16280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.101017237467535e-05,
+      "loss": 2.9443,
+      "step": 16285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.100478998619876e-05,
+      "loss": 2.9996,
+      "step": 16290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.099940614619144e-05,
+      "loss": 2.963,
+      "step": 16295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.099402085484399e-05,
+      "loss": 2.9166,
+      "step": 16300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.098863411234704e-05,
+      "loss": 2.9471,
+      "step": 16305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.09832459188913e-05,
+      "loss": 2.9407,
+      "step": 16310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.097785627466746e-05,
+      "loss": 2.8941,
+      "step": 16315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.097246517986634e-05,
+      "loss": 2.9082,
+      "step": 16320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.096707263467877e-05,
+      "loss": 3.0274,
+      "step": 16325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.096167863929565e-05,
+      "loss": 2.9285,
+      "step": 16330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.095628319390791e-05,
+      "loss": 2.9701,
+      "step": 16335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.095088629870656e-05,
+      "loss": 2.9642,
+      "step": 16340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.094548795388264e-05,
+      "loss": 2.884,
+      "step": 16345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.094008815962723e-05,
+      "loss": 2.9274,
+      "step": 16350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.09346869161315e-05,
+      "loss": 2.9995,
+      "step": 16355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.092928422358662e-05,
+      "loss": 2.9775,
+      "step": 16360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.092388008218387e-05,
+      "loss": 3.0064,
+      "step": 16365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.091847449211454e-05,
+      "loss": 3.0057,
+      "step": 16370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.091306745357e-05,
+      "loss": 2.9254,
+      "step": 16375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.090765896674161e-05,
+      "loss": 2.9234,
+      "step": 16380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.090224903182086e-05,
+      "loss": 2.844,
+      "step": 16385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.089683764899924e-05,
+      "loss": 2.9695,
+      "step": 16390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.089142481846833e-05,
+      "loss": 3.0263,
+      "step": 16395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.088601054041972e-05,
+      "loss": 2.9186,
+      "step": 16400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.088059481504508e-05,
+      "loss": 3.0024,
+      "step": 16405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.087517764253611e-05,
+      "loss": 2.8495,
+      "step": 16410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.086975902308459e-05,
+      "loss": 2.9066,
+      "step": 16415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.086433895688231e-05,
+      "loss": 3.0331,
+      "step": 16420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.085891744412116e-05,
+      "loss": 2.9599,
+      "step": 16425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.085349448499304e-05,
+      "loss": 2.8832,
+      "step": 16430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.084807007968993e-05,
+      "loss": 2.8673,
+      "step": 16435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.084264422840384e-05,
+      "loss": 2.9723,
+      "step": 16440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.083721693132684e-05,
+      "loss": 2.8303,
+      "step": 16445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.083178818865106e-05,
+      "loss": 2.9703,
+      "step": 16450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.082635800056864e-05,
+      "loss": 2.9864,
+      "step": 16455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.082092636727186e-05,
+      "loss": 2.9555,
+      "step": 16460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.081549328895296e-05,
+      "loss": 2.9068,
+      "step": 16465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.081005876580425e-05,
+      "loss": 2.983,
+      "step": 16470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.080462279801813e-05,
+      "loss": 2.8808,
+      "step": 16475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.079918538578704e-05,
+      "loss": 2.8618,
+      "step": 16480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.079374652930343e-05,
+      "loss": 2.9595,
+      "step": 16485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.078830622875986e-05,
+      "loss": 2.9763,
+      "step": 16490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.078286448434889e-05,
+      "loss": 2.8517,
+      "step": 16495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.077742129626315e-05,
+      "loss": 2.9641,
+      "step": 16500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.077197666469534e-05,
+      "loss": 2.9133,
+      "step": 16505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.076653058983818e-05,
+      "loss": 2.9693,
+      "step": 16510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.076108307188446e-05,
+      "loss": 2.9123,
+      "step": 16515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.075563411102703e-05,
+      "loss": 2.8845,
+      "step": 16520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.075018370745877e-05,
+      "loss": 2.9993,
+      "step": 16525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.074473186137261e-05,
+      "loss": 2.9985,
+      "step": 16530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.073927857296155e-05,
+      "loss": 3.009,
+      "step": 16535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.073382384241863e-05,
+      "loss": 2.9393,
+      "step": 16540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.072836766993692e-05,
+      "loss": 2.8442,
+      "step": 16545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.072291005570961e-05,
+      "loss": 2.9677,
+      "step": 16550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.071745099992987e-05,
+      "loss": 3.0675,
+      "step": 16555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.071199050279093e-05,
+      "loss": 2.92,
+      "step": 16560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.07065285644861e-05,
+      "loss": 2.8906,
+      "step": 16565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.070106518520874e-05,
+      "loss": 2.8647,
+      "step": 16570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.069560036515222e-05,
+      "loss": 3.0474,
+      "step": 16575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.069013410451002e-05,
+      "loss": 2.8825,
+      "step": 16580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.068466640347562e-05,
+      "loss": 2.9653,
+      "step": 16585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.067919726224257e-05,
+      "loss": 2.9431,
+      "step": 16590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.067372668100449e-05,
+      "loss": 2.9825,
+      "step": 16595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.066825465995501e-05,
+      "loss": 2.9477,
+      "step": 16600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.066278119928786e-05,
+      "loss": 3.0237,
+      "step": 16605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.065730629919678e-05,
+      "loss": 2.988,
+      "step": 16610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.065182995987558e-05,
+      "loss": 2.9333,
+      "step": 16615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.064635218151812e-05,
+      "loss": 2.9623,
+      "step": 16620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.064087296431829e-05,
+      "loss": 2.779,
+      "step": 16625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.063539230847008e-05,
+      "loss": 3.047,
+      "step": 16630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.062991021416747e-05,
+      "loss": 2.901,
+      "step": 16635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.062442668160455e-05,
+      "loss": 2.9215,
+      "step": 16640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.06189417109754e-05,
+      "loss": 3.0107,
+      "step": 16645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.061345530247421e-05,
+      "loss": 2.8772,
+      "step": 16650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.060796745629518e-05,
+      "loss": 2.9458,
+      "step": 16655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.060247817263259e-05,
+      "loss": 2.9815,
+      "step": 16660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.059698745168072e-05,
+      "loss": 2.8055,
+      "step": 16665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.059149529363397e-05,
+      "loss": 2.9414,
+      "step": 16670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.058600169868674e-05,
+      "loss": 2.9378,
+      "step": 16675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.058050666703353e-05,
+      "loss": 3.0014,
+      "step": 16680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.05750101988688e-05,
+      "loss": 2.8864,
+      "step": 16685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.056951229438718e-05,
+      "loss": 2.8602,
+      "step": 16690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.056401295378324e-05,
+      "loss": 3.0097,
+      "step": 16695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.055851217725168e-05,
+      "loss": 2.8609,
+      "step": 16700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.055300996498724e-05,
+      "loss": 2.873,
+      "step": 16705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.054750631718464e-05,
+      "loss": 2.9089,
+      "step": 16710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.054200123403875e-05,
+      "loss": 2.8971,
+      "step": 16715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.053649471574443e-05,
+      "loss": 2.8274,
+      "step": 16720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.053098676249661e-05,
+      "loss": 2.8743,
+      "step": 16725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.052547737449024e-05,
+      "loss": 2.8736,
+      "step": 16730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.051996655192039e-05,
+      "loss": 2.9149,
+      "step": 16735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.051445429498211e-05,
+      "loss": 2.962,
+      "step": 16740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.050894060387053e-05,
+      "loss": 2.863,
+      "step": 16745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.050342547878084e-05,
+      "loss": 2.8875,
+      "step": 16750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.049790891990826e-05,
+      "loss": 2.9166,
+      "step": 16755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.049239092744807e-05,
+      "loss": 2.9127,
+      "step": 16760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.048687150159562e-05,
+      "loss": 3.0275,
+      "step": 16765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.048135064254628e-05,
+      "loss": 2.8919,
+      "step": 16770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.047582835049546e-05,
+      "loss": 2.8521,
+      "step": 16775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.047030462563869e-05,
+      "loss": 2.8557,
+      "step": 16780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.046477946817148e-05,
+      "loss": 2.9571,
+      "step": 16785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.04592528782894e-05,
+      "loss": 2.8348,
+      "step": 16790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.045372485618813e-05,
+      "loss": 2.8856,
+      "step": 16795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.04481954020633e-05,
+      "loss": 2.9133,
+      "step": 16800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.04426645161107e-05,
+      "loss": 2.901,
+      "step": 16805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.043713219852608e-05,
+      "loss": 2.9672,
+      "step": 16810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.043159844950529e-05,
+      "loss": 2.9069,
+      "step": 16815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.042606326924423e-05,
+      "loss": 2.9028,
+      "step": 16820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.042052665793884e-05,
+      "loss": 3.0447,
+      "step": 16825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.041498861578511e-05,
+      "loss": 2.927,
+      "step": 16830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.040944914297906e-05,
+      "loss": 2.9692,
+      "step": 16835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.04039082397168e-05,
+      "loss": 2.9495,
+      "step": 16840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.039836590619446e-05,
+      "loss": 2.9806,
+      "step": 16845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.039282214260826e-05,
+      "loss": 3.0325,
+      "step": 16850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.038727694915443e-05,
+      "loss": 2.9325,
+      "step": 16855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.038173032602925e-05,
+      "loss": 2.9864,
+      "step": 16860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.037618227342907e-05,
+      "loss": 2.8332,
+      "step": 16865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.037063279155032e-05,
+      "loss": 2.953,
+      "step": 16870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.03650818805894e-05,
+      "loss": 2.9812,
+      "step": 16875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.035952954074285e-05,
+      "loss": 2.8863,
+      "step": 16880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.035397577220717e-05,
+      "loss": 2.9116,
+      "step": 16885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.034842057517898e-05,
+      "loss": 3.019,
+      "step": 16890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.034286394985494e-05,
+      "loss": 3.0278,
+      "step": 16895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.033730589643175e-05,
+      "loss": 2.837,
+      "step": 16900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.033174641510614e-05,
+      "loss": 2.9285,
+      "step": 16905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.032618550607493e-05,
+      "loss": 2.8812,
+      "step": 16910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.032062316953495e-05,
+      "loss": 2.9459,
+      "step": 16915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.031505940568311e-05,
+      "loss": 2.9917,
+      "step": 16920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.030949421471637e-05,
+      "loss": 2.7366,
+      "step": 16925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.030392759683174e-05,
+      "loss": 2.9335,
+      "step": 16930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.029835955222626e-05,
+      "loss": 2.9666,
+      "step": 16935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.029279008109702e-05,
+      "loss": 2.7521,
+      "step": 16940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.028721918364118e-05,
+      "loss": 2.9929,
+      "step": 16945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.028164686005596e-05,
+      "loss": 2.955,
+      "step": 16950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.02760731105386e-05,
+      "loss": 3.0066,
+      "step": 16955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.027049793528643e-05,
+      "loss": 2.9482,
+      "step": 16960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.026492133449679e-05,
+      "loss": 2.8989,
+      "step": 16965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.025934330836708e-05,
+      "loss": 2.9259,
+      "step": 16970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.025376385709477e-05,
+      "loss": 2.842,
+      "step": 16975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.024818298087734e-05,
+      "loss": 3.0241,
+      "step": 16980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.024260067991237e-05,
+      "loss": 2.985,
+      "step": 16985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.023701695439748e-05,
+      "loss": 2.9065,
+      "step": 16990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.02314318045303e-05,
+      "loss": 3.0076,
+      "step": 16995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.022584523050856e-05,
+      "loss": 2.9334,
+      "step": 17000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.022025723253001e-05,
+      "loss": 2.9308,
+      "step": 17005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.021466781079246e-05,
+      "loss": 2.8266,
+      "step": 17010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.020907696549377e-05,
+      "loss": 2.9744,
+      "step": 17015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.020348469683187e-05,
+      "loss": 2.8871,
+      "step": 17020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.019789100500469e-05,
+      "loss": 3.0008,
+      "step": 17025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.019229589021025e-05,
+      "loss": 2.8644,
+      "step": 17030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.018669935264661e-05,
+      "loss": 2.8927,
+      "step": 17035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.018110139251191e-05,
+      "loss": 2.9084,
+      "step": 17040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.017550201000428e-05,
+      "loss": 2.9402,
+      "step": 17045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.016990120532197e-05,
+      "loss": 2.969,
+      "step": 17050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.016429897866318e-05,
+      "loss": 2.8672,
+      "step": 17055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.015869533022627e-05,
+      "loss": 2.9954,
+      "step": 17060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.015309026020961e-05,
+      "loss": 2.9244,
+      "step": 17065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.014748376881159e-05,
+      "loss": 2.869,
+      "step": 17070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.014187585623068e-05,
+      "loss": 2.9661,
+      "step": 17075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.013626652266541e-05,
+      "loss": 2.964,
+      "step": 17080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.013065576831433e-05,
+      "loss": 3.0034,
+      "step": 17085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.012504359337606e-05,
+      "loss": 2.8184,
+      "step": 17090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.011942999804925e-05,
+      "loss": 2.8439,
+      "step": 17095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.011381498253264e-05,
+      "loss": 2.9251,
+      "step": 17100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.0108198547025e-05,
+      "loss": 2.998,
+      "step": 17105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.010258069172512e-05,
+      "loss": 2.884,
+      "step": 17110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.009696141683189e-05,
+      "loss": 2.8652,
+      "step": 17115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.009134072254419e-05,
+      "loss": 2.9891,
+      "step": 17120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.008571860906104e-05,
+      "loss": 2.9192,
+      "step": 17125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.008009507658143e-05,
+      "loss": 2.8874,
+      "step": 17130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.007447012530443e-05,
+      "loss": 2.9624,
+      "step": 17135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.006884375542914e-05,
+      "loss": 2.9242,
+      "step": 17140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.006321596715477e-05,
+      "loss": 2.8839,
+      "step": 17145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.00575867606805e-05,
+      "loss": 2.9235,
+      "step": 17150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.00519561362056e-05,
+      "loss": 2.9428,
+      "step": 17155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.004632409392943e-05,
+      "loss": 2.999,
+      "step": 17160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.004069063405132e-05,
+      "loss": 2.9555,
+      "step": 17165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.003505575677069e-05,
+      "loss": 2.8335,
+      "step": 17170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.002941946228702e-05,
+      "loss": 2.9681,
+      "step": 17175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.002378175079984e-05,
+      "loss": 2.9875,
+      "step": 17180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.00181426225087e-05,
+      "loss": 3.0219,
+      "step": 17185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.001250207761323e-05,
+      "loss": 2.9433,
+      "step": 17190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.000686011631308e-05,
+      "loss": 2.9187,
+      "step": 17195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.000121673880802e-05,
+      "loss": 2.9272,
+      "step": 17200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.999557194529776e-05,
+      "loss": 2.9177,
+      "step": 17205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.998992573598216e-05,
+      "loss": 2.9119,
+      "step": 17210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.998427811106109e-05,
+      "loss": 2.9818,
+      "step": 17215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.997862907073444e-05,
+      "loss": 2.9889,
+      "step": 17220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.99729786152022e-05,
+      "loss": 2.8544,
+      "step": 17225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.996732674466442e-05,
+      "loss": 2.928,
+      "step": 17230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.996167345932111e-05,
+      "loss": 2.9704,
+      "step": 17235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.995601875937242e-05,
+      "loss": 2.8861,
+      "step": 17240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.995036264501855e-05,
+      "loss": 2.8689,
+      "step": 17245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.994470511645968e-05,
+      "loss": 2.9655,
+      "step": 17250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.993904617389611e-05,
+      "loss": 2.8891,
+      "step": 17255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.993338581752812e-05,
+      "loss": 2.9339,
+      "step": 17260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.992772404755612e-05,
+      "loss": 2.9051,
+      "step": 17265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.992206086418053e-05,
+      "loss": 2.9434,
+      "step": 17270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.99163962676018e-05,
+      "loss": 2.8728,
+      "step": 17275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.991073025802047e-05,
+      "loss": 2.8934,
+      "step": 17280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.990506283563708e-05,
+      "loss": 2.9063,
+      "step": 17285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.989939400065231e-05,
+      "loss": 2.992,
+      "step": 17290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.989372375326678e-05,
+      "loss": 2.8813,
+      "step": 17295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.988805209368123e-05,
+      "loss": 2.9187,
+      "step": 17300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.988237902209641e-05,
+      "loss": 2.9569,
+      "step": 17305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.98767045387132e-05,
+      "loss": 3.0181,
+      "step": 17310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.98710286437324e-05,
+      "loss": 3.004,
+      "step": 17315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.986535133735498e-05,
+      "loss": 2.9359,
+      "step": 17320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.985967261978187e-05,
+      "loss": 2.8999,
+      "step": 17325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.985399249121415e-05,
+      "loss": 2.9838,
+      "step": 17330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.984831095185285e-05,
+      "loss": 2.8732,
+      "step": 17335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.984262800189907e-05,
+      "loss": 2.9158,
+      "step": 17340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.983694364155404e-05,
+      "loss": 3.0546,
+      "step": 17345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.983125787101893e-05,
+      "loss": 2.9296,
+      "step": 17350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.982557069049504e-05,
+      "loss": 2.8621,
+      "step": 17355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.981988210018368e-05,
+      "loss": 2.9178,
+      "step": 17360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.981419210028622e-05,
+      "loss": 2.8911,
+      "step": 17365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.980850069100408e-05,
+      "loss": 2.9281,
+      "step": 17370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.980280787253874e-05,
+      "loss": 2.951,
+      "step": 17375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.979711364509172e-05,
+      "loss": 2.9404,
+      "step": 17380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.979141800886456e-05,
+      "loss": 2.9395,
+      "step": 17385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.978572096405893e-05,
+      "loss": 2.9461,
+      "step": 17390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.978002251087647e-05,
+      "loss": 2.986,
+      "step": 17395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.97743226495189e-05,
+      "loss": 2.9225,
+      "step": 17400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.976862138018798e-05,
+      "loss": 2.8853,
+      "step": 17405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.976291870308557e-05,
+      "loss": 2.935,
+      "step": 17410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.975721461841348e-05,
+      "loss": 2.9129,
+      "step": 17415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.975150912637368e-05,
+      "loss": 2.8798,
+      "step": 17420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.974580222716815e-05,
+      "loss": 2.995,
+      "step": 17425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.974009392099883e-05,
+      "loss": 2.8753,
+      "step": 17430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.973438420806786e-05,
+      "loss": 2.8194,
+      "step": 17435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.972867308857734e-05,
+      "loss": 2.9659,
+      "step": 17440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.972296056272945e-05,
+      "loss": 2.8957,
+      "step": 17445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.971724663072636e-05,
+      "loss": 2.9641,
+      "step": 17450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.97115312927704e-05,
+      "loss": 3.0686,
+      "step": 17455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.970581454906383e-05,
+      "loss": 2.9783,
+      "step": 17460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.970009639980906e-05,
+      "loss": 3.0603,
+      "step": 17465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.969437684520848e-05,
+      "loss": 2.9466,
+      "step": 17470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.968865588546458e-05,
+      "loss": 2.9257,
+      "step": 17475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.968293352077988e-05,
+      "loss": 2.9064,
+      "step": 17480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.967720975135691e-05,
+      "loss": 2.8713,
+      "step": 17485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.967148457739831e-05,
+      "loss": 3.0052,
+      "step": 17490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.966575799910676e-05,
+      "loss": 3.0594,
+      "step": 17495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.966003001668494e-05,
+      "loss": 3.0044,
+      "step": 17500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.965430063033567e-05,
+      "loss": 2.9831,
+      "step": 17505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.964856984026168e-05,
+      "loss": 2.8995,
+      "step": 17510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.964283764666592e-05,
+      "loss": 2.9366,
+      "step": 17515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.963710404975125e-05,
+      "loss": 2.9465,
+      "step": 17520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.963136904972067e-05,
+      "loss": 2.9748,
+      "step": 17525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.962563264677717e-05,
+      "loss": 2.9881,
+      "step": 17530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.96198948411238e-05,
+      "loss": 2.8259,
+      "step": 17535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.961415563296371e-05,
+      "loss": 2.91,
+      "step": 17540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.960841502250004e-05,
+      "loss": 2.8804,
+      "step": 17545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.960267300993602e-05,
+      "loss": 2.9661,
+      "step": 17550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.959692959547487e-05,
+      "loss": 2.962,
+      "step": 17555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.959118477931995e-05,
+      "loss": 2.9488,
+      "step": 17560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.958543856167461e-05,
+      "loss": 2.9807,
+      "step": 17565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.957969094274222e-05,
+      "loss": 2.8938,
+      "step": 17570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.95739419227263e-05,
+      "loss": 2.8142,
+      "step": 17575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.956819150183031e-05,
+      "loss": 2.9747,
+      "step": 17580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.956243968025785e-05,
+      "loss": 2.9934,
+      "step": 17585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.95566864582125e-05,
+      "loss": 2.934,
+      "step": 17590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.955093183589794e-05,
+      "loss": 2.9148,
+      "step": 17595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.954517581351785e-05,
+      "loss": 2.8713,
+      "step": 17600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.953941839127601e-05,
+      "loss": 2.8933,
+      "step": 17605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.953365956937623e-05,
+      "loss": 3.002,
+      "step": 17610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.952789934802235e-05,
+      "loss": 2.8796,
+      "step": 17615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.952213772741828e-05,
+      "loss": 2.8192,
+      "step": 17620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.951637470776799e-05,
+      "loss": 2.9459,
+      "step": 17625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.951061028927547e-05,
+      "loss": 2.8357,
+      "step": 17630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.95048444721448e-05,
+      "loss": 2.8578,
+      "step": 17635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.949907725658004e-05,
+      "loss": 2.9556,
+      "step": 17640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.949330864278541e-05,
+      "loss": 2.9031,
+      "step": 17645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.948753863096503e-05,
+      "loss": 2.8751,
+      "step": 17650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.948176722132321e-05,
+      "loss": 2.9322,
+      "step": 17655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.947599441406424e-05,
+      "loss": 2.9466,
+      "step": 17660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.947022020939248e-05,
+      "loss": 2.9793,
+      "step": 17665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.946444460751232e-05,
+      "loss": 2.9148,
+      "step": 17670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.94586676086282e-05,
+      "loss": 3.0654,
+      "step": 17675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.945288921294463e-05,
+      "loss": 2.8676,
+      "step": 17680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.944710942066617e-05,
+      "loss": 3.0045,
+      "step": 17685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.944132823199742e-05,
+      "loss": 2.8934,
+      "step": 17690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.943554564714303e-05,
+      "loss": 2.8704,
+      "step": 17695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.942976166630766e-05,
+      "loss": 2.8731,
+      "step": 17700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.94239762896961e-05,
+      "loss": 2.991,
+      "step": 17705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.941818951751312e-05,
+      "loss": 3.0344,
+      "step": 17710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.94124013499636e-05,
+      "loss": 2.9015,
+      "step": 17715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.940661178725241e-05,
+      "loss": 2.8729,
+      "step": 17720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.94008208295845e-05,
+      "loss": 2.8724,
+      "step": 17725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.939502847716486e-05,
+      "loss": 2.9766,
+      "step": 17730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.938923473019853e-05,
+      "loss": 2.8735,
+      "step": 17735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.938343958889063e-05,
+      "loss": 2.8156,
+      "step": 17740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.937764305344628e-05,
+      "loss": 2.9609,
+      "step": 17745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.937184512407067e-05,
+      "loss": 2.9422,
+      "step": 17750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.936604580096904e-05,
+      "loss": 2.9644,
+      "step": 17755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.93602450843467e-05,
+      "loss": 2.959,
+      "step": 17760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.935444297440897e-05,
+      "loss": 2.9214,
+      "step": 17765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.934863947136125e-05,
+      "loss": 2.9036,
+      "step": 17770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.934283457540897e-05,
+      "loss": 3.0533,
+      "step": 17775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.933702828675762e-05,
+      "loss": 2.8953,
+      "step": 17780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.933122060561275e-05,
+      "loss": 3.0263,
+      "step": 17785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.932541153217993e-05,
+      "loss": 2.8081,
+      "step": 17790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.93196010666648e-05,
+      "loss": 2.8624,
+      "step": 17795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.931378920927306e-05,
+      "loss": 2.9223,
+      "step": 17800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.930797596021041e-05,
+      "loss": 2.9227,
+      "step": 17805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.930216131968268e-05,
+      "loss": 2.9034,
+      "step": 17810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.929634528789565e-05,
+      "loss": 2.8767,
+      "step": 17815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.929052786505525e-05,
+      "loss": 2.9933,
+      "step": 17820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.928470905136739e-05,
+      "loss": 2.9846,
+      "step": 17825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.927888884703804e-05,
+      "loss": 2.9015,
+      "step": 17830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.927306725227325e-05,
+      "loss": 2.9873,
+      "step": 17835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.926724426727908e-05,
+      "loss": 2.89,
+      "step": 17840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.926141989226166e-05,
+      "loss": 2.9009,
+      "step": 17845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.925559412742721e-05,
+      "loss": 2.9359,
+      "step": 17850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.924976697298188e-05,
+      "loss": 2.8599,
+      "step": 17855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.924393842913202e-05,
+      "loss": 2.9548,
+      "step": 17860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.923810849608392e-05,
+      "loss": 2.9568,
+      "step": 17865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.923227717404395e-05,
+      "loss": 3.0456,
+      "step": 17870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.922644446321852e-05,
+      "loss": 2.9826,
+      "step": 17875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.922061036381415e-05,
+      "loss": 2.9045,
+      "step": 17880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.921477487603733e-05,
+      "loss": 2.9559,
+      "step": 17885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.920893800009462e-05,
+      "loss": 3.0029,
+      "step": 17890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.920309973619267e-05,
+      "loss": 2.9695,
+      "step": 17895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.919726008453813e-05,
+      "loss": 2.9331,
+      "step": 17900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.919141904533772e-05,
+      "loss": 2.8441,
+      "step": 17905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.918557661879821e-05,
+      "loss": 2.8397,
+      "step": 17910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.917973280512642e-05,
+      "loss": 2.9676,
+      "step": 17915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.91738876045292e-05,
+      "loss": 2.952,
+      "step": 17920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.916804101721348e-05,
+      "loss": 2.9478,
+      "step": 17925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.916219304338622e-05,
+      "loss": 2.9338,
+      "step": 17930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.915634368325443e-05,
+      "loss": 2.9934,
+      "step": 17935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.915049293702518e-05,
+      "loss": 2.8861,
+      "step": 17940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.914464080490557e-05,
+      "loss": 2.9836,
+      "step": 17945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.913878728710276e-05,
+      "loss": 2.863,
+      "step": 17950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.913293238382397e-05,
+      "loss": 2.9942,
+      "step": 17955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.912707609527644e-05,
+      "loss": 2.9748,
+      "step": 17960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.91212184216675e-05,
+      "loss": 3.0169,
+      "step": 17965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.91153593632045e-05,
+      "loss": 2.9436,
+      "step": 17970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.910949892009483e-05,
+      "loss": 2.901,
+      "step": 17975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.910363709254595e-05,
+      "loss": 2.9797,
+      "step": 17980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.909777388076537e-05,
+      "loss": 2.868,
+      "step": 17985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.909190928496064e-05,
+      "loss": 2.9286,
+      "step": 17990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.908604330533936e-05,
+      "loss": 3.0117,
+      "step": 17995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.908017594210918e-05,
+      "loss": 3.0604,
+      "step": 18000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.90743071954778e-05,
+      "loss": 2.8165,
+      "step": 18005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.906843706565298e-05,
+      "loss": 2.8588,
+      "step": 18010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.90625655528425e-05,
+      "loss": 3.0358,
+      "step": 18015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.905669265725421e-05,
+      "loss": 2.9691,
+      "step": 18020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.905081837909599e-05,
+      "loss": 2.9015,
+      "step": 18025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.90449427185758e-05,
+      "loss": 2.9424,
+      "step": 18030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.903906567590166e-05,
+      "loss": 3.0974,
+      "step": 18035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.903318725128156e-05,
+      "loss": 2.8456,
+      "step": 18040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.902730744492362e-05,
+      "loss": 2.9598,
+      "step": 18045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.902142625703598e-05,
+      "loss": 2.9004,
+      "step": 18050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.90155436878268e-05,
+      "loss": 2.9088,
+      "step": 18055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.900965973750437e-05,
+      "loss": 2.9059,
+      "step": 18060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.900377440627693e-05,
+      "loss": 2.9617,
+      "step": 18065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.899788769435284e-05,
+      "loss": 3.0062,
+      "step": 18070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.899199960194046e-05,
+      "loss": 2.8525,
+      "step": 18075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.898611012924825e-05,
+      "loss": 3.0194,
+      "step": 18080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.898021927648468e-05,
+      "loss": 3.0489,
+      "step": 18085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.897432704385828e-05,
+      "loss": 2.8801,
+      "step": 18090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.896843343157764e-05,
+      "loss": 2.8929,
+      "step": 18095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.896253843985136e-05,
+      "loss": 2.9559,
+      "step": 18100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.895664206888814e-05,
+      "loss": 2.8672,
+      "step": 18105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.895074431889671e-05,
+      "loss": 2.9945,
+      "step": 18110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.894484519008584e-05,
+      "loss": 3.0413,
+      "step": 18115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.893894468266434e-05,
+      "loss": 2.9708,
+      "step": 18120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.893304279684109e-05,
+      "loss": 2.9438,
+      "step": 18125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.892713953282502e-05,
+      "loss": 2.9344,
+      "step": 18130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.892123489082509e-05,
+      "loss": 2.9213,
+      "step": 18135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.891532887105032e-05,
+      "loss": 2.9372,
+      "step": 18140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.890942147370979e-05,
+      "loss": 2.957,
+      "step": 18145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.89035126990126e-05,
+      "loss": 2.9449,
+      "step": 18150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.889760254716792e-05,
+      "loss": 2.8977,
+      "step": 18155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.889169101838496e-05,
+      "loss": 2.8763,
+      "step": 18160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8885778112873e-05,
+      "loss": 3.0188,
+      "step": 18165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.887986383084133e-05,
+      "loss": 2.9393,
+      "step": 18170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.887394817249931e-05,
+      "loss": 3.039,
+      "step": 18175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.886803113805637e-05,
+      "loss": 3.0118,
+      "step": 18180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.886211272772195e-05,
+      "loss": 2.9517,
+      "step": 18185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.885619294170554e-05,
+      "loss": 2.9458,
+      "step": 18190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.885027178021676e-05,
+      "loss": 3.029,
+      "step": 18195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.884434924346514e-05,
+      "loss": 2.9576,
+      "step": 18200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.883842533166036e-05,
+      "loss": 2.909,
+      "step": 18205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.883250004501213e-05,
+      "loss": 2.9644,
+      "step": 18210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.882657338373019e-05,
+      "loss": 2.9147,
+      "step": 18215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.882064534802435e-05,
+      "loss": 2.9586,
+      "step": 18220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.881471593810444e-05,
+      "loss": 2.9422,
+      "step": 18225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.880878515418034e-05,
+      "loss": 2.8712,
+      "step": 18230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.880285299646205e-05,
+      "loss": 3.0229,
+      "step": 18235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.879691946515951e-05,
+      "loss": 2.9505,
+      "step": 18240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.879098456048278e-05,
+      "loss": 2.9298,
+      "step": 18245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.878504828264196e-05,
+      "loss": 2.85,
+      "step": 18250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.877911063184717e-05,
+      "loss": 2.9394,
+      "step": 18255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.877317160830862e-05,
+      "loss": 2.9506,
+      "step": 18260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.876723121223652e-05,
+      "loss": 2.9139,
+      "step": 18265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.876128944384117e-05,
+      "loss": 2.9197,
+      "step": 18270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.875534630333289e-05,
+      "loss": 2.8629,
+      "step": 18275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.874940179092208e-05,
+      "loss": 2.9497,
+      "step": 18280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.874345590681916e-05,
+      "loss": 2.9141,
+      "step": 18285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.873750865123462e-05,
+      "loss": 2.7893,
+      "step": 18290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.873156002437896e-05,
+      "loss": 2.9476,
+      "step": 18295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.872561002646278e-05,
+      "loss": 2.954,
+      "step": 18300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.871965865769669e-05,
+      "loss": 2.8378,
+      "step": 18305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.871370591829138e-05,
+      "loss": 2.9507,
+      "step": 18310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.870775180845756e-05,
+      "loss": 2.9893,
+      "step": 18315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8701796328406e-05,
+      "loss": 2.9623,
+      "step": 18320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.86958394783475e-05,
+      "loss": 2.8692,
+      "step": 18325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.868988125849298e-05,
+      "loss": 2.8921,
+      "step": 18330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.868392166905331e-05,
+      "loss": 2.9415,
+      "step": 18335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.867796071023945e-05,
+      "loss": 2.9051,
+      "step": 18340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.867199838226243e-05,
+      "loss": 3.0112,
+      "step": 18345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.866603468533332e-05,
+      "loss": 3.027,
+      "step": 18350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.866006961966322e-05,
+      "loss": 2.8754,
+      "step": 18355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.865410318546327e-05,
+      "loss": 2.9202,
+      "step": 18360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.86481353829447e-05,
+      "loss": 2.9593,
+      "step": 18365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.864216621231877e-05,
+      "loss": 2.922,
+      "step": 18370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.863619567379676e-05,
+      "loss": 2.8919,
+      "step": 18375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.863022376759001e-05,
+      "loss": 2.9581,
+      "step": 18380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.862425049390997e-05,
+      "loss": 2.9659,
+      "step": 18385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.861827585296804e-05,
+      "loss": 2.9713,
+      "step": 18390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.861229984497574e-05,
+      "loss": 2.9463,
+      "step": 18395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.86063224701446e-05,
+      "loss": 3.0162,
+      "step": 18400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.860034372868625e-05,
+      "loss": 2.9606,
+      "step": 18405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.859436362081228e-05,
+      "loss": 3.0154,
+      "step": 18410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.858838214673442e-05,
+      "loss": 2.9573,
+      "step": 18415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.858239930666438e-05,
+      "loss": 2.9232,
+      "step": 18420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.857641510081397e-05,
+      "loss": 2.8914,
+      "step": 18425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8570429529395e-05,
+      "loss": 2.9642,
+      "step": 18430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.85644425926194e-05,
+      "loss": 2.9612,
+      "step": 18435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.855845429069904e-05,
+      "loss": 2.8696,
+      "step": 18440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.855246462384595e-05,
+      "loss": 2.8933,
+      "step": 18445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.854647359227214e-05,
+      "loss": 2.9973,
+      "step": 18450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.85404811961897e-05,
+      "loss": 2.9998,
+      "step": 18455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.853448743581073e-05,
+      "loss": 2.9627,
+      "step": 18460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.85284923113474e-05,
+      "loss": 2.9052,
+      "step": 18465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.852249582301195e-05,
+      "loss": 2.9239,
+      "step": 18470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.851649797101667e-05,
+      "loss": 2.8541,
+      "step": 18475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.851049875557382e-05,
+      "loss": 2.9436,
+      "step": 18480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.850449817689583e-05,
+      "loss": 2.9681,
+      "step": 18485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.849849623519509e-05,
+      "loss": 2.9137,
+      "step": 18490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.849249293068404e-05,
+      "loss": 2.8946,
+      "step": 18495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.848648826357523e-05,
+      "loss": 3.007,
+      "step": 18500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.848048223408119e-05,
+      "loss": 2.9323,
+      "step": 18505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.847447484241454e-05,
+      "loss": 2.9286,
+      "step": 18510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.846846608878793e-05,
+      "loss": 2.977,
+      "step": 18515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.846245597341408e-05,
+      "loss": 2.9182,
+      "step": 18520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.845644449650573e-05,
+      "loss": 2.9616,
+      "step": 18525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.845043165827567e-05,
+      "loss": 2.924,
+      "step": 18530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.844441745893679e-05,
+      "loss": 2.9434,
+      "step": 18535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.843840189870192e-05,
+      "loss": 2.9314,
+      "step": 18540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.843238497778405e-05,
+      "loss": 3.0338,
+      "step": 18545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.842636669639619e-05,
+      "loss": 2.8929,
+      "step": 18550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.842034705475134e-05,
+      "loss": 2.9214,
+      "step": 18555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.841432605306261e-05,
+      "loss": 2.9732,
+      "step": 18560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.840830369154314e-05,
+      "loss": 2.915,
+      "step": 18565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.840227997040608e-05,
+      "loss": 2.9383,
+      "step": 18570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.839625488986473e-05,
+      "loss": 2.9282,
+      "step": 18575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.839022845013232e-05,
+      "loss": 3.0339,
+      "step": 18580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.838420065142221e-05,
+      "loss": 2.9829,
+      "step": 18585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.837817149394776e-05,
+      "loss": 2.9516,
+      "step": 18590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.83721409779224e-05,
+      "loss": 2.9053,
+      "step": 18595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.83661091035596e-05,
+      "loss": 2.9537,
+      "step": 18600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.83600758710729e-05,
+      "loss": 2.8848,
+      "step": 18605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.835404128067585e-05,
+      "loss": 2.9306,
+      "step": 18610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.834800533258207e-05,
+      "loss": 2.9324,
+      "step": 18615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.834196802700525e-05,
+      "loss": 2.917,
+      "step": 18620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.833592936415909e-05,
+      "loss": 2.8363,
+      "step": 18625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.832988934425735e-05,
+      "loss": 2.9456,
+      "step": 18630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.832384796751385e-05,
+      "loss": 2.7852,
+      "step": 18635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.831780523414244e-05,
+      "loss": 2.8796,
+      "step": 18640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.831176114435703e-05,
+      "loss": 2.9078,
+      "step": 18645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.830571569837157e-05,
+      "loss": 2.9451,
+      "step": 18650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.829966889640008e-05,
+      "loss": 2.9539,
+      "step": 18655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.829362073865659e-05,
+      "loss": 2.8654,
+      "step": 18660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.828757122535522e-05,
+      "loss": 2.9318,
+      "step": 18665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.828152035671009e-05,
+      "loss": 2.9788,
+      "step": 18670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.827546813293542e-05,
+      "loss": 2.9101,
+      "step": 18675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.826941455424543e-05,
+      "loss": 2.9268,
+      "step": 18680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.826335962085444e-05,
+      "loss": 2.935,
+      "step": 18685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.825730333297677e-05,
+      "loss": 2.9746,
+      "step": 18690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.82512456908268e-05,
+      "loss": 2.929,
+      "step": 18695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.824518669461899e-05,
+      "loss": 2.9206,
+      "step": 18700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.823912634456779e-05,
+      "loss": 2.9221,
+      "step": 18705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.823306464088777e-05,
+      "loss": 2.9241,
+      "step": 18710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.822700158379346e-05,
+      "loss": 3.0079,
+      "step": 18715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.822093717349954e-05,
+      "loss": 2.8981,
+      "step": 18720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.821487141022064e-05,
+      "loss": 2.8909,
+      "step": 18725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.820880429417151e-05,
+      "loss": 2.8983,
+      "step": 18730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.820273582556692e-05,
+      "loss": 2.8879,
+      "step": 18735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.819666600462166e-05,
+      "loss": 2.9668,
+      "step": 18740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.819059483155063e-05,
+      "loss": 3.0127,
+      "step": 18745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.81845223065687e-05,
+      "loss": 3.0104,
+      "step": 18750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.81784484298909e-05,
+      "loss": 2.9251,
+      "step": 18755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.817237320173217e-05,
+      "loss": 2.9345,
+      "step": 18760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.81662966223076e-05,
+      "loss": 2.8277,
+      "step": 18765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.816021869183229e-05,
+      "loss": 2.9005,
+      "step": 18770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.81541394105214e-05,
+      "loss": 3.0337,
+      "step": 18775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.814805877859012e-05,
+      "loss": 2.9321,
+      "step": 18780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.814197679625371e-05,
+      "loss": 2.988,
+      "step": 18785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.813589346372746e-05,
+      "loss": 2.9603,
+      "step": 18790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.812980878122672e-05,
+      "loss": 2.943,
+      "step": 18795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.812372274896686e-05,
+      "loss": 2.9483,
+      "step": 18800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.811763536716335e-05,
+      "loss": 2.8215,
+      "step": 18805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.811154663603166e-05,
+      "loss": 2.8075,
+      "step": 18810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.810545655578733e-05,
+      "loss": 2.9081,
+      "step": 18815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.809936512664594e-05,
+      "loss": 3.0322,
+      "step": 18820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.809327234882313e-05,
+      "loss": 2.9882,
+      "step": 18825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.808717822253456e-05,
+      "loss": 2.9707,
+      "step": 18830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.808108274799598e-05,
+      "loss": 2.8835,
+      "step": 18835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.807498592542314e-05,
+      "loss": 2.9678,
+      "step": 18840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.806888775503188e-05,
+      "loss": 2.9375,
+      "step": 18845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.806278823703806e-05,
+      "loss": 2.8984,
+      "step": 18850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.80566873716576e-05,
+      "loss": 2.8885,
+      "step": 18855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.805058515910647e-05,
+      "loss": 2.8496,
+      "step": 18860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.804448159960068e-05,
+      "loss": 2.9576,
+      "step": 18865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.803837669335629e-05,
+      "loss": 2.9177,
+      "step": 18870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.80322704405894e-05,
+      "loss": 2.9048,
+      "step": 18875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.802616284151616e-05,
+      "loss": 2.9516,
+      "step": 18880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.80200538963528e-05,
+      "loss": 2.9457,
+      "step": 18885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.801394360531555e-05,
+      "loss": 3.0459,
+      "step": 18890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.800783196862072e-05,
+      "loss": 2.8926,
+      "step": 18895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.800171898648463e-05,
+      "loss": 2.936,
+      "step": 18900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.799560465912372e-05,
+      "loss": 2.8965,
+      "step": 18905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.79894889867544e-05,
+      "loss": 2.9141,
+      "step": 18910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.798337196959315e-05,
+      "loss": 2.8968,
+      "step": 18915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.797725360785653e-05,
+      "loss": 2.8724,
+      "step": 18920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.797113390176111e-05,
+      "loss": 3.0457,
+      "step": 18925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.796501285152355e-05,
+      "loss": 2.9872,
+      "step": 18930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.795889045736048e-05,
+      "loss": 2.9192,
+      "step": 18935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.795276671948865e-05,
+      "loss": 2.9261,
+      "step": 18940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.794664163812485e-05,
+      "loss": 2.8726,
+      "step": 18945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.79405152134859e-05,
+      "loss": 2.9214,
+      "step": 18950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.793438744578864e-05,
+      "loss": 2.9925,
+      "step": 18955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.792825833525002e-05,
+      "loss": 2.9734,
+      "step": 18960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.792212788208698e-05,
+      "loss": 2.988,
+      "step": 18965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.791599608651655e-05,
+      "loss": 2.9335,
+      "step": 18970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.790986294875579e-05,
+      "loss": 2.8938,
+      "step": 18975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.79037284690218e-05,
+      "loss": 2.9806,
+      "step": 18980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.789759264753172e-05,
+      "loss": 2.8691,
+      "step": 18985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.789145548450277e-05,
+      "loss": 2.8905,
+      "step": 18990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.788531698015221e-05,
+      "loss": 3.0312,
+      "step": 18995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.787917713469732e-05,
+      "loss": 2.876,
+      "step": 19000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.787303594835545e-05,
+      "loss": 2.9366,
+      "step": 19005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.7866893421344e-05,
+      "loss": 2.9716,
+      "step": 19010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.786074955388038e-05,
+      "loss": 2.8257,
+      "step": 19015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.785460434618211e-05,
+      "loss": 2.9897,
+      "step": 19020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.784845779846671e-05,
+      "loss": 2.8844,
+      "step": 19025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.784230991095176e-05,
+      "loss": 3.0083,
+      "step": 19030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.783616068385491e-05,
+      "loss": 2.8073,
+      "step": 19035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.783001011739381e-05,
+      "loss": 2.9476,
+      "step": 19040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.78238582117862e-05,
+      "loss": 2.8413,
+      "step": 19045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.781770496724984e-05,
+      "loss": 2.9792,
+      "step": 19050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.781155038400258e-05,
+      "loss": 2.9162,
+      "step": 19055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.780539446226223e-05,
+      "loss": 2.9717,
+      "step": 19060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.779923720224674e-05,
+      "loss": 2.8999,
+      "step": 19065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.779307860417408e-05,
+      "loss": 2.8876,
+      "step": 19070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.778691866826225e-05,
+      "loss": 2.9565,
+      "step": 19075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.778075739472928e-05,
+      "loss": 2.8183,
+      "step": 19080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.77745947837933e-05,
+      "loss": 2.9337,
+      "step": 19085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.776843083567248e-05,
+      "loss": 2.882,
+      "step": 19090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.776226555058497e-05,
+      "loss": 3.005,
+      "step": 19095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.775609892874904e-05,
+      "loss": 2.9876,
+      "step": 19100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.774993097038298e-05,
+      "loss": 2.9509,
+      "step": 19105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.774376167570515e-05,
+      "loss": 2.9444,
+      "step": 19110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.77375910449339e-05,
+      "loss": 2.8673,
+      "step": 19115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.773141907828769e-05,
+      "loss": 2.9776,
+      "step": 19120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.7725245775985e-05,
+      "loss": 2.9576,
+      "step": 19125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.771907113824433e-05,
+      "loss": 2.8297,
+      "step": 19130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.771289516528431e-05,
+      "loss": 2.987,
+      "step": 19135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.770671785732351e-05,
+      "loss": 2.9628,
+      "step": 19140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.770053921458064e-05,
+      "loss": 2.9764,
+      "step": 19145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.76943592372744e-05,
+      "loss": 2.9077,
+      "step": 19150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.768817792562357e-05,
+      "loss": 2.8612,
+      "step": 19155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.768199527984693e-05,
+      "loss": 2.8329,
+      "step": 19160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.767581130016335e-05,
+      "loss": 2.9359,
+      "step": 19165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.766962598679176e-05,
+      "loss": 2.966,
+      "step": 19170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.766343933995112e-05,
+      "loss": 2.9945,
+      "step": 19175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.765725135986038e-05,
+      "loss": 2.9582,
+      "step": 19180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.765106204673866e-05,
+      "loss": 2.952,
+      "step": 19185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.764487140080498e-05,
+      "loss": 2.9198,
+      "step": 19190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.763867942227855e-05,
+      "loss": 2.9301,
+      "step": 19195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.76324861113785e-05,
+      "loss": 2.8737,
+      "step": 19200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.762629146832412e-05,
+      "loss": 2.8953,
+      "step": 19205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.762009549333468e-05,
+      "loss": 2.9425,
+      "step": 19210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.761389818662949e-05,
+      "loss": 2.9697,
+      "step": 19215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.760769954842794e-05,
+      "loss": 2.9685,
+      "step": 19220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.760149957894945e-05,
+      "loss": 2.967,
+      "step": 19225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.759529827841352e-05,
+      "loss": 2.9995,
+      "step": 19230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.758909564703964e-05,
+      "loss": 3.0211,
+      "step": 19235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.75828916850474e-05,
+      "loss": 2.9483,
+      "step": 19240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.757668639265639e-05,
+      "loss": 2.9791,
+      "step": 19245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.75704797700863e-05,
+      "loss": 2.9122,
+      "step": 19250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.756427181755682e-05,
+      "loss": 2.9029,
+      "step": 19255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.755806253528772e-05,
+      "loss": 2.9244,
+      "step": 19260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.755185192349876e-05,
+      "loss": 3.0072,
+      "step": 19265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.754563998240983e-05,
+      "loss": 2.9827,
+      "step": 19270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.753942671224085e-05,
+      "loss": 2.9395,
+      "step": 19275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.753321211321171e-05,
+      "loss": 2.861,
+      "step": 19280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.752699618554241e-05,
+      "loss": 2.9948,
+      "step": 19285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.752077892945303e-05,
+      "loss": 2.9447,
+      "step": 19290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.751456034516361e-05,
+      "loss": 2.9484,
+      "step": 19295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.75083404328943e-05,
+      "loss": 2.9627,
+      "step": 19300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.750211919286527e-05,
+      "loss": 2.8991,
+      "step": 19305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.749589662529676e-05,
+      "loss": 2.878,
+      "step": 19310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.748967273040905e-05,
+      "loss": 2.9735,
+      "step": 19315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.748344750842242e-05,
+      "loss": 2.9549,
+      "step": 19320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.747722095955728e-05,
+      "loss": 2.9901,
+      "step": 19325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.747099308403402e-05,
+      "loss": 2.9827,
+      "step": 19330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.74647638820731e-05,
+      "loss": 2.9591,
+      "step": 19335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.745853335389505e-05,
+      "loss": 2.9035,
+      "step": 19340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.745230149972042e-05,
+      "loss": 3.047,
+      "step": 19345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.744606831976978e-05,
+      "loss": 2.8725,
+      "step": 19350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.743983381426381e-05,
+      "loss": 3.0138,
+      "step": 19355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.743359798342321e-05,
+      "loss": 2.9667,
+      "step": 19360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.74273608274687e-05,
+      "loss": 2.9012,
+      "step": 19365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.742112234662108e-05,
+      "loss": 2.9128,
+      "step": 19370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.741488254110117e-05,
+      "loss": 2.9901,
+      "step": 19375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.740864141112987e-05,
+      "loss": 2.9546,
+      "step": 19380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.740239895692812e-05,
+      "loss": 2.9518,
+      "step": 19385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.739615517871688e-05,
+      "loss": 2.9306,
+      "step": 19390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.738991007671718e-05,
+      "loss": 2.9599,
+      "step": 19395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.73836636511501e-05,
+      "loss": 3.0153,
+      "step": 19400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.737741590223674e-05,
+      "loss": 2.9809,
+      "step": 19405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.737116683019826e-05,
+      "loss": 2.823,
+      "step": 19410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.736491643525589e-05,
+      "loss": 2.9554,
+      "step": 19415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.735866471763087e-05,
+      "loss": 2.793,
+      "step": 19420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.735241167754454e-05,
+      "loss": 2.9772,
+      "step": 19425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.734615731521823e-05,
+      "loss": 2.8899,
+      "step": 19430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.733990163087331e-05,
+      "loss": 2.9512,
+      "step": 19435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.733364462473126e-05,
+      "loss": 3.0442,
+      "step": 19440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.732738629701357e-05,
+      "loss": 2.9325,
+      "step": 19445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.732112664794177e-05,
+      "loss": 2.9497,
+      "step": 19450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.731486567773745e-05,
+      "loss": 2.9445,
+      "step": 19455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.730860338662224e-05,
+      "loss": 2.9714,
+      "step": 19460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.73023397748178e-05,
+      "loss": 2.9228,
+      "step": 19465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.72960748425459e-05,
+      "loss": 2.9804,
+      "step": 19470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.728980859002829e-05,
+      "loss": 2.9498,
+      "step": 19475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.728354101748678e-05,
+      "loss": 2.9155,
+      "step": 19480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.727727212514323e-05,
+      "loss": 2.9301,
+      "step": 19485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.727100191321958e-05,
+      "loss": 2.8975,
+      "step": 19490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.726473038193776e-05,
+      "loss": 2.9667,
+      "step": 19495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.725845753151982e-05,
+      "loss": 2.8816,
+      "step": 19500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.725218336218776e-05,
+      "loss": 2.927,
+      "step": 19505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.724590787416371e-05,
+      "loss": 2.9157,
+      "step": 19510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.723963106766981e-05,
+      "loss": 2.9581,
+      "step": 19515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.723335294292827e-05,
+      "loss": 2.9217,
+      "step": 19520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.72270735001613e-05,
+      "loss": 2.8521,
+      "step": 19525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.722079273959121e-05,
+      "loss": 2.9865,
+      "step": 19530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.721451066144031e-05,
+      "loss": 2.9384,
+      "step": 19535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.7208227265931e-05,
+      "loss": 2.9554,
+      "step": 19540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.720194255328571e-05,
+      "loss": 2.8881,
+      "step": 19545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.719565652372688e-05,
+      "loss": 2.881,
+      "step": 19550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.718936917747706e-05,
+      "loss": 2.9213,
+      "step": 19555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.718308051475881e-05,
+      "loss": 2.9091,
+      "step": 19560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.717679053579475e-05,
+      "loss": 3.0146,
+      "step": 19565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.717049924080752e-05,
+      "loss": 2.946,
+      "step": 19570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.716420663001982e-05,
+      "loss": 2.9223,
+      "step": 19575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.715791270365444e-05,
+      "loss": 2.8883,
+      "step": 19580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.715161746193414e-05,
+      "loss": 2.8271,
+      "step": 19585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.714532090508181e-05,
+      "loss": 2.9929,
+      "step": 19590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.71390230333203e-05,
+      "loss": 2.9997,
+      "step": 19595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.713272384687254e-05,
+      "loss": 2.9036,
+      "step": 19600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.712642334596158e-05,
+      "loss": 2.9937,
+      "step": 19605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.712012153081038e-05,
+      "loss": 2.9834,
+      "step": 19610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.711381840164206e-05,
+      "loss": 2.9767,
+      "step": 19615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.710751395867972e-05,
+      "loss": 2.8445,
+      "step": 19620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.710120820214655e-05,
+      "loss": 2.9824,
+      "step": 19625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.709490113226577e-05,
+      "loss": 3.068,
+      "step": 19630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.708859274926065e-05,
+      "loss": 2.8936,
+      "step": 19635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.708228305335447e-05,
+      "loss": 2.8849,
+      "step": 19640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.70759720447706e-05,
+      "loss": 2.9365,
+      "step": 19645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.706965972373245e-05,
+      "loss": 2.8774,
+      "step": 19650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.706334609046347e-05,
+      "loss": 3.017,
+      "step": 19655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.705703114518717e-05,
+      "loss": 2.9534,
+      "step": 19660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.705071488812706e-05,
+      "loss": 3.006,
+      "step": 19665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.704439731950675e-05,
+      "loss": 2.9253,
+      "step": 19670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.70380784395499e-05,
+      "loss": 2.8847,
+      "step": 19675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.703175824848013e-05,
+      "loss": 3.0086,
+      "step": 19680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.702543674652122e-05,
+      "loss": 2.8917,
+      "step": 19685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.701911393389693e-05,
+      "loss": 2.958,
+      "step": 19690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.701278981083109e-05,
+      "loss": 2.8728,
+      "step": 19695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.700646437754755e-05,
+      "loss": 2.8712,
+      "step": 19700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.700013763427025e-05,
+      "loss": 2.8859,
+      "step": 19705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.69938095812231e-05,
+      "loss": 2.9251,
+      "step": 19710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.698748021863017e-05,
+      "loss": 2.7846,
+      "step": 19715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.698114954671548e-05,
+      "loss": 2.8394,
+      "step": 19720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.697481756570314e-05,
+      "loss": 2.9428,
+      "step": 19725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.696848427581728e-05,
+      "loss": 2.8741,
+      "step": 19730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.69621496772821e-05,
+      "loss": 2.8682,
+      "step": 19735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.695581377032185e-05,
+      "loss": 3.0342,
+      "step": 19740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.694947655516081e-05,
+      "loss": 2.8888,
+      "step": 19745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.694313803202331e-05,
+      "loss": 2.9544,
+      "step": 19750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.693679820113371e-05,
+      "loss": 2.9523,
+      "step": 19755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.693045706271647e-05,
+      "loss": 3.0345,
+      "step": 19760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.692411461699604e-05,
+      "loss": 2.9011,
+      "step": 19765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.691777086419693e-05,
+      "loss": 2.9449,
+      "step": 19770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.691142580454372e-05,
+      "loss": 2.9541,
+      "step": 19775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.6905079438261e-05,
+      "loss": 2.9404,
+      "step": 19780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.689873176557344e-05,
+      "loss": 2.9665,
+      "step": 19785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.689238278670573e-05,
+      "loss": 2.956,
+      "step": 19790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.688603250188264e-05,
+      "loss": 2.913,
+      "step": 19795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.687968091132893e-05,
+      "loss": 2.932,
+      "step": 19800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.687332801526947e-05,
+      "loss": 2.9941,
+      "step": 19805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.686697381392912e-05,
+      "loss": 2.9891,
+      "step": 19810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.686061830753285e-05,
+      "loss": 2.8634,
+      "step": 19815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.68542614963056e-05,
+      "loss": 2.9411,
+      "step": 19820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.684790338047241e-05,
+      "loss": 3.0111,
+      "step": 19825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.684154396025837e-05,
+      "loss": 2.9408,
+      "step": 19830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.683518323588857e-05,
+      "loss": 2.9509,
+      "step": 19835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.682882120758818e-05,
+      "loss": 2.9246,
+      "step": 19840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.682245787558244e-05,
+      "loss": 3.0011,
+      "step": 19845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.681609324009654e-05,
+      "loss": 2.917,
+      "step": 19850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.680972730135585e-05,
+      "loss": 3.0111,
+      "step": 19855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.680336005958566e-05,
+      "loss": 2.9907,
+      "step": 19860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.679699151501144e-05,
+      "loss": 2.9361,
+      "step": 19865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.679062166785855e-05,
+      "loss": 2.9179,
+      "step": 19870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.678425051835252e-05,
+      "loss": 2.9237,
+      "step": 19875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.677787806671889e-05,
+      "loss": 2.9559,
+      "step": 19880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.67715043131832e-05,
+      "loss": 2.9653,
+      "step": 19885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.676512925797111e-05,
+      "loss": 2.8752,
+      "step": 19890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.675875290130827e-05,
+      "loss": 2.9515,
+      "step": 19895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.675237524342041e-05,
+      "loss": 2.9092,
+      "step": 19900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.674599628453329e-05,
+      "loss": 3.0518,
+      "step": 19905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.673961602487273e-05,
+      "loss": 2.8082,
+      "step": 19910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.673323446466455e-05,
+      "loss": 2.962,
+      "step": 19915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.67268516041347e-05,
+      "loss": 2.9933,
+      "step": 19920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.672046744350908e-05,
+      "loss": 2.9424,
+      "step": 19925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.67140819830137e-05,
+      "loss": 2.9194,
+      "step": 19930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.670769522287461e-05,
+      "loss": 2.9067,
+      "step": 19935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.67013071633179e-05,
+      "loss": 3.0398,
+      "step": 19940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.669491780456968e-05,
+      "loss": 3.0258,
+      "step": 19945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.668852714685616e-05,
+      "loss": 2.993,
+      "step": 19950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.668213519040352e-05,
+      "loss": 3.0468,
+      "step": 19955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.667574193543803e-05,
+      "loss": 2.8485,
+      "step": 19960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.666934738218607e-05,
+      "loss": 3.0153,
+      "step": 19965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.666295153087393e-05,
+      "loss": 2.9388,
+      "step": 19970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.665655438172804e-05,
+      "loss": 3.0225,
+      "step": 19975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.665015593497486e-05,
+      "loss": 2.9177,
+      "step": 19980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.664375619084089e-05,
+      "loss": 2.8946,
+      "step": 19985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.663735514955267e-05,
+      "loss": 2.8445,
+      "step": 19990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.663095281133679e-05,
+      "loss": 2.9133,
+      "step": 19995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.662454917641989e-05,
+      "loss": 2.8042,
+      "step": 20000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.661814424502866e-05,
+      "loss": 2.8886,
+      "step": 20005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.66117380173898e-05,
+      "loss": 2.8817,
+      "step": 20010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.660533049373011e-05,
+      "loss": 2.9272,
+      "step": 20015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.65989216742764e-05,
+      "loss": 2.9308,
+      "step": 20020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.659251155925554e-05,
+      "loss": 2.9087,
+      "step": 20025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.658610014889445e-05,
+      "loss": 2.9043,
+      "step": 20030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.657968744342006e-05,
+      "loss": 2.9226,
+      "step": 20035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.657327344305943e-05,
+      "loss": 2.9314,
+      "step": 20040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.656685814803954e-05,
+      "loss": 2.9583,
+      "step": 20045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.656044155858753e-05,
+      "loss": 2.8817,
+      "step": 20050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.655402367493053e-05,
+      "loss": 2.9902,
+      "step": 20055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.654760449729574e-05,
+      "loss": 2.8899,
+      "step": 20060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.654118402591035e-05,
+      "loss": 2.8477,
+      "step": 20065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.653476226100168e-05,
+      "loss": 2.9644,
+      "step": 20070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.652833920279703e-05,
+      "loss": 2.9569,
+      "step": 20075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.652191485152381e-05,
+      "loss": 2.9686,
+      "step": 20080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.65154892074094e-05,
+      "loss": 2.9459,
+      "step": 20085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.650906227068125e-05,
+      "loss": 2.8423,
+      "step": 20090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.650263404156691e-05,
+      "loss": 2.8509,
+      "step": 20095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.64962045202939e-05,
+      "loss": 2.9396,
+      "step": 20100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.648977370708982e-05,
+      "loss": 2.8929,
+      "step": 20105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.648334160218234e-05,
+      "loss": 2.8353,
+      "step": 20110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.647690820579912e-05,
+      "loss": 3.0134,
+      "step": 20115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.647047351816793e-05,
+      "loss": 3.0023,
+      "step": 20120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.646403753951652e-05,
+      "loss": 2.8681,
+      "step": 20125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.645760027007273e-05,
+      "loss": 2.8736,
+      "step": 20130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.645116171006445e-05,
+      "loss": 2.872,
+      "step": 20135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.644472185971956e-05,
+      "loss": 2.8711,
+      "step": 20140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.643828071926606e-05,
+      "loss": 2.934,
+      "step": 20145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.643183828893194e-05,
+      "loss": 2.9772,
+      "step": 20150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.642539456894528e-05,
+      "loss": 2.8106,
+      "step": 20155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.641894955953414e-05,
+      "loss": 2.9197,
+      "step": 20160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.64125032609267e-05,
+      "loss": 2.881,
+      "step": 20165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.640605567335116e-05,
+      "loss": 2.8952,
+      "step": 20170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.639960679703571e-05,
+      "loss": 2.927,
+      "step": 20175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.63931566322087e-05,
+      "loss": 2.9062,
+      "step": 20180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.638670517909842e-05,
+      "loss": 2.7553,
+      "step": 20185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.638025243793324e-05,
+      "loss": 2.7981,
+      "step": 20190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.63737984089416e-05,
+      "loss": 2.9179,
+      "step": 20195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.636734309235197e-05,
+      "loss": 2.9604,
+      "step": 20200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.636088648839284e-05,
+      "loss": 2.9564,
+      "step": 20205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.63544285972928e-05,
+      "loss": 2.9688,
+      "step": 20210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.634796941928041e-05,
+      "loss": 2.8964,
+      "step": 20215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.634150895458433e-05,
+      "loss": 2.9234,
+      "step": 20220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.63350472034333e-05,
+      "loss": 2.9444,
+      "step": 20225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.6328584166056e-05,
+      "loss": 2.9227,
+      "step": 20230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.632211984268126e-05,
+      "loss": 2.8267,
+      "step": 20235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.631565423353789e-05,
+      "loss": 2.8537,
+      "step": 20240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.630918733885476e-05,
+      "loss": 2.8795,
+      "step": 20245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.630271915886079e-05,
+      "loss": 2.9477,
+      "step": 20250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.629624969378498e-05,
+      "loss": 3.0396,
+      "step": 20255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.62897789438563e-05,
+      "loss": 3.0556,
+      "step": 20260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.628330690930384e-05,
+      "loss": 2.8773,
+      "step": 20265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.627683359035669e-05,
+      "loss": 2.9152,
+      "step": 20270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.627035898724402e-05,
+      "loss": 2.9092,
+      "step": 20275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.626388310019499e-05,
+      "loss": 2.9558,
+      "step": 20280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.625740592943886e-05,
+      "loss": 2.9248,
+      "step": 20285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.625092747520491e-05,
+      "loss": 2.9571,
+      "step": 20290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.624444773772249e-05,
+      "loss": 2.9365,
+      "step": 20295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.623796671722096e-05,
+      "loss": 2.8448,
+      "step": 20300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.623148441392971e-05,
+      "loss": 2.9087,
+      "step": 20305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.622500082807827e-05,
+      "loss": 2.8196,
+      "step": 20310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.621851595989613e-05,
+      "loss": 2.8878,
+      "step": 20315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.621202980961282e-05,
+      "loss": 2.8703,
+      "step": 20320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.620554237745799e-05,
+      "loss": 2.9684,
+      "step": 20325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.619905366366123e-05,
+      "loss": 2.9326,
+      "step": 20330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.61925636684523e-05,
+      "loss": 2.9842,
+      "step": 20335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.61860723920609e-05,
+      "loss": 2.81,
+      "step": 20340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.617957983471681e-05,
+      "loss": 2.9876,
+      "step": 20345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.617308599664988e-05,
+      "loss": 2.9532,
+      "step": 20350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.616659087808997e-05,
+      "loss": 3.031,
+      "step": 20355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.616009447926702e-05,
+      "loss": 2.9145,
+      "step": 20360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.6153596800411e-05,
+      "loss": 2.8756,
+      "step": 20365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.61470978417519e-05,
+      "loss": 2.9448,
+      "step": 20370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.614059760351978e-05,
+      "loss": 2.9379,
+      "step": 20375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.613409608594474e-05,
+      "loss": 2.862,
+      "step": 20380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.612759328925695e-05,
+      "loss": 2.8415,
+      "step": 20385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.612108921368658e-05,
+      "loss": 2.9571,
+      "step": 20390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.611458385946388e-05,
+      "loss": 2.8467,
+      "step": 20395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.610807722681913e-05,
+      "loss": 2.8467,
+      "step": 20400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.610156931598266e-05,
+      "loss": 2.9124,
+      "step": 20405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.609506012718483e-05,
+      "loss": 2.9392,
+      "step": 20410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.608854966065607e-05,
+      "loss": 2.9303,
+      "step": 20415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.608203791662686e-05,
+      "loss": 3.0301,
+      "step": 20420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.607552489532769e-05,
+      "loss": 2.8429,
+      "step": 20425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.606901059698911e-05,
+      "loss": 2.929,
+      "step": 20430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.606249502184173e-05,
+      "loss": 2.9565,
+      "step": 20435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.605597817011621e-05,
+      "loss": 2.9948,
+      "step": 20440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.604946004204321e-05,
+      "loss": 2.896,
+      "step": 20445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.60429406378535e-05,
+      "loss": 2.932,
+      "step": 20450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.603641995777782e-05,
+      "loss": 3.0582,
+      "step": 20455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.602989800204704e-05,
+      "loss": 2.864,
+      "step": 20460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.6023374770892e-05,
+      "loss": 2.9442,
+      "step": 20465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.601685026454361e-05,
+      "loss": 2.9861,
+      "step": 20470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.601032448323286e-05,
+      "loss": 2.945,
+      "step": 20475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.600379742719074e-05,
+      "loss": 2.9361,
+      "step": 20480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.59972690966483e-05,
+      "loss": 2.9094,
+      "step": 20485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.599073949183664e-05,
+      "loss": 2.8568,
+      "step": 20490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.59842086129869e-05,
+      "loss": 2.9069,
+      "step": 20495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.597767646033028e-05,
+      "loss": 2.9163,
+      "step": 20500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.5971143034098e-05,
+      "loss": 2.9085,
+      "step": 20505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.596460833452134e-05,
+      "loss": 2.8592,
+      "step": 20510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.595807236183162e-05,
+      "loss": 2.9136,
+      "step": 20515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.59515351162602e-05,
+      "loss": 2.8695,
+      "step": 20520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.59449965980385e-05,
+      "loss": 2.8654,
+      "step": 20525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.593845680739798e-05,
+      "loss": 2.9083,
+      "step": 20530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.593191574457015e-05,
+      "loss": 3.0008,
+      "step": 20535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.592537340978655e-05,
+      "loss": 2.8814,
+      "step": 20540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.591882980327876e-05,
+      "loss": 2.9738,
+      "step": 20545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.591228492527843e-05,
+      "loss": 2.945,
+      "step": 20550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.590573877601723e-05,
+      "loss": 2.9097,
+      "step": 20555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.58991913557269e-05,
+      "loss": 2.9359,
+      "step": 20560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.589264266463922e-05,
+      "loss": 2.9342,
+      "step": 20565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.5886092702986e-05,
+      "loss": 2.9764,
+      "step": 20570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.58795414709991e-05,
+      "loss": 3.039,
+      "step": 20575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.587298896891041e-05,
+      "loss": 2.8857,
+      "step": 20580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.586643519695192e-05,
+      "loss": 2.8877,
+      "step": 20585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.585988015535559e-05,
+      "loss": 2.8356,
+      "step": 20590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.585332384435347e-05,
+      "loss": 2.8745,
+      "step": 20595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.58467662641777e-05,
+      "loss": 2.9011,
+      "step": 20600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.584020741506032e-05,
+      "loss": 3.0077,
+      "step": 20605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.583364729723357e-05,
+      "loss": 2.9608,
+      "step": 20610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.582708591092967e-05,
+      "loss": 2.8803,
+      "step": 20615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.582052325638088e-05,
+      "loss": 2.9203,
+      "step": 20620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.581395933381948e-05,
+      "loss": 2.9313,
+      "step": 20625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.580739414347787e-05,
+      "loss": 2.9625,
+      "step": 20630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.580082768558844e-05,
+      "loss": 2.7778,
+      "step": 20635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.579425996038362e-05,
+      "loss": 2.8467,
+      "step": 20640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.578769096809592e-05,
+      "loss": 2.981,
+      "step": 20645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.578112070895785e-05,
+      "loss": 2.8849,
+      "step": 20650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.577454918320203e-05,
+      "loss": 3.0072,
+      "step": 20655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.576797639106104e-05,
+      "loss": 2.9739,
+      "step": 20660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.57614023327676e-05,
+      "loss": 2.8561,
+      "step": 20665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.575482700855439e-05,
+      "loss": 2.8881,
+      "step": 20670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.574825041865419e-05,
+      "loss": 2.8837,
+      "step": 20675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.574167256329979e-05,
+      "loss": 2.9071,
+      "step": 20680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.573509344272404e-05,
+      "loss": 2.9397,
+      "step": 20685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.572851305715986e-05,
+      "loss": 3.0269,
+      "step": 20690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.572193140684014e-05,
+      "loss": 2.7939,
+      "step": 20695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.571534849199792e-05,
+      "loss": 2.9903,
+      "step": 20700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.570876431286618e-05,
+      "loss": 2.8783,
+      "step": 20705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.570217886967803e-05,
+      "loss": 2.9578,
+      "step": 20710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.569559216266658e-05,
+      "loss": 2.8458,
+      "step": 20715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.568900419206497e-05,
+      "loss": 3.0031,
+      "step": 20720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.568241495810645e-05,
+      "loss": 2.9399,
+      "step": 20725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.567582446102424e-05,
+      "loss": 3.0367,
+      "step": 20730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.566923270105165e-05,
+      "loss": 2.9271,
+      "step": 20735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.5662639678422e-05,
+      "loss": 2.9238,
+      "step": 20740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.56560453933687e-05,
+      "loss": 2.853,
+      "step": 20745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.56494498461252e-05,
+      "loss": 2.9123,
+      "step": 20750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.564285303692494e-05,
+      "loss": 3.013,
+      "step": 20755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.563625496600146e-05,
+      "loss": 2.8892,
+      "step": 20760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.56296556335883e-05,
+      "loss": 2.8446,
+      "step": 20765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.562305503991909e-05,
+      "loss": 2.8725,
+      "step": 20770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.56164531852275e-05,
+      "loss": 2.879,
+      "step": 20775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.56098500697472e-05,
+      "loss": 3.0246,
+      "step": 20780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.560324569371196e-05,
+      "loss": 2.8587,
+      "step": 20785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.559664005735554e-05,
+      "loss": 2.914,
+      "step": 20790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.559003316091181e-05,
+      "loss": 2.9196,
+      "step": 20795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.558342500461461e-05,
+      "loss": 2.9785,
+      "step": 20800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.55768155886979e-05,
+      "loss": 2.9996,
+      "step": 20805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.55702049133956e-05,
+      "loss": 2.8344,
+      "step": 20810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.556359297894177e-05,
+      "loss": 2.8817,
+      "step": 20815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.555697978557042e-05,
+      "loss": 2.8977,
+      "step": 20820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.55503653335157e-05,
+      "loss": 2.8696,
+      "step": 20825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.554374962301173e-05,
+      "loss": 3.008,
+      "step": 20830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.553713265429268e-05,
+      "loss": 2.9707,
+      "step": 20835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.553051442759283e-05,
+      "loss": 2.9975,
+      "step": 20840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.55238949431464e-05,
+      "loss": 2.8179,
+      "step": 20845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.551727420118778e-05,
+      "loss": 2.9295,
+      "step": 20850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.551065220195129e-05,
+      "loss": 2.9375,
+      "step": 20855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.550402894567137e-05,
+      "loss": 2.8513,
+      "step": 20860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.549740443258245e-05,
+      "loss": 2.9646,
+      "step": 20865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.549077866291905e-05,
+      "loss": 2.952,
+      "step": 20870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.548415163691571e-05,
+      "loss": 2.8564,
+      "step": 20875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.547752335480704e-05,
+      "loss": 2.9121,
+      "step": 20880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.547089381682763e-05,
+      "loss": 2.921,
+      "step": 20885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.546426302321221e-05,
+      "loss": 2.8993,
+      "step": 20890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.545763097419546e-05,
+      "loss": 2.9094,
+      "step": 20895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.545099767001219e-05,
+      "loss": 2.9676,
+      "step": 20900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.544436311089719e-05,
+      "loss": 2.9525,
+      "step": 20905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.54377272970853e-05,
+      "loss": 2.902,
+      "step": 20910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.543109022881146e-05,
+      "loss": 2.9519,
+      "step": 20915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.542445190631059e-05,
+      "loss": 2.9633,
+      "step": 20920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.541781232981768e-05,
+      "loss": 2.9626,
+      "step": 20925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.541117149956779e-05,
+      "loss": 2.9743,
+      "step": 20930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.540452941579595e-05,
+      "loss": 2.8701,
+      "step": 20935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.539788607873733e-05,
+      "loss": 2.9739,
+      "step": 20940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.539124148862709e-05,
+      "loss": 2.9421,
+      "step": 20945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.538459564570043e-05,
+      "loss": 2.9325,
+      "step": 20950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.537794855019261e-05,
+      "loss": 2.9031,
+      "step": 20955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.537130020233892e-05,
+      "loss": 2.8654,
+      "step": 20960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.536465060237473e-05,
+      "loss": 2.9423,
+      "step": 20965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.535799975053541e-05,
+      "loss": 2.9132,
+      "step": 20970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.53513476470564e-05,
+      "loss": 3.0108,
+      "step": 20975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.534469429217318e-05,
+      "loss": 2.9069,
+      "step": 20980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.533803968612127e-05,
+      "loss": 2.8673,
+      "step": 20985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.533138382913624e-05,
+      "loss": 2.9721,
+      "step": 20990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.532472672145368e-05,
+      "loss": 2.9538,
+      "step": 20995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.531806836330929e-05,
+      "loss": 2.9586,
+      "step": 21000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.531140875493872e-05,
+      "loss": 2.8899,
+      "step": 21005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.530474789657777e-05,
+      "loss": 2.9745,
+      "step": 21010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.529808578846217e-05,
+      "loss": 2.9714,
+      "step": 21015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.529142243082778e-05,
+      "loss": 2.9142,
+      "step": 21020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.528475782391048e-05,
+      "loss": 2.9825,
+      "step": 21025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.527809196794618e-05,
+      "loss": 2.9068,
+      "step": 21030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.527142486317084e-05,
+      "loss": 3.0155,
+      "step": 21035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.52647565098205e-05,
+      "loss": 2.9353,
+      "step": 21040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.525808690813118e-05,
+      "loss": 3.0605,
+      "step": 21045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.5251416058339e-05,
+      "loss": 2.9195,
+      "step": 21050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.524474396068007e-05,
+      "loss": 2.9916,
+      "step": 21055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.523807061539062e-05,
+      "loss": 2.9442,
+      "step": 21060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.523139602270684e-05,
+      "loss": 3.0373,
+      "step": 21065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.522472018286504e-05,
+      "loss": 2.9298,
+      "step": 21070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.521804309610151e-05,
+      "loss": 3.0145,
+      "step": 21075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.521136476265261e-05,
+      "loss": 2.9899,
+      "step": 21080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.520468518275478e-05,
+      "loss": 2.9462,
+      "step": 21085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.519800435664445e-05,
+      "loss": 2.9466,
+      "step": 21090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.519132228455811e-05,
+      "loss": 2.9532,
+      "step": 21095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.518463896673231e-05,
+      "loss": 2.9082,
+      "step": 21100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.517795440340361e-05,
+      "loss": 3.0222,
+      "step": 21105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.517126859480867e-05,
+      "loss": 2.9359,
+      "step": 21110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.516458154118415e-05,
+      "loss": 2.9329,
+      "step": 21115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.515789324276675e-05,
+      "loss": 2.9656,
+      "step": 21120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.515120369979325e-05,
+      "loss": 2.9003,
+      "step": 21125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.514451291250043e-05,
+      "loss": 2.9261,
+      "step": 21130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.513782088112517e-05,
+      "loss": 2.871,
+      "step": 21135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.513112760590433e-05,
+      "loss": 2.9203,
+      "step": 21140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.512443308707485e-05,
+      "loss": 2.949,
+      "step": 21145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.511773732487374e-05,
+      "loss": 2.9485,
+      "step": 21150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.511104031953801e-05,
+      "loss": 2.908,
+      "step": 21155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.510434207130469e-05,
+      "loss": 2.9341,
+      "step": 21160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.509764258041093e-05,
+      "loss": 3.0507,
+      "step": 21165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.50909418470939e-05,
+      "loss": 2.8011,
+      "step": 21170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.508423987159074e-05,
+      "loss": 2.9641,
+      "step": 21175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.507753665413876e-05,
+      "loss": 2.9855,
+      "step": 21180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.50708321949752e-05,
+      "loss": 3.0022,
+      "step": 21185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.506412649433741e-05,
+      "loss": 2.9071,
+      "step": 21190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.505741955246278e-05,
+      "loss": 2.9202,
+      "step": 21195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.50507113695887e-05,
+      "loss": 2.7944,
+      "step": 21200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.504400194595266e-05,
+      "loss": 3.0029,
+      "step": 21205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.503729128179214e-05,
+      "loss": 2.9794,
+      "step": 21210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.503057937734472e-05,
+      "loss": 3.0099,
+      "step": 21215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.502386623284797e-05,
+      "loss": 2.8865,
+      "step": 21220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.501715184853954e-05,
+      "loss": 3.0222,
+      "step": 21225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.501043622465712e-05,
+      "loss": 2.8715,
+      "step": 21230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.500371936143843e-05,
+      "loss": 2.8183,
+      "step": 21235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.499700125912124e-05,
+      "loss": 2.8736,
+      "step": 21240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.499028191794336e-05,
+      "loss": 2.9576,
+      "step": 21245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.498356133814264e-05,
+      "loss": 2.9407,
+      "step": 21250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.497683951995702e-05,
+      "loss": 2.9008,
+      "step": 21255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.497011646362441e-05,
+      "loss": 2.9225,
+      "step": 21260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.49633921693828e-05,
+      "loss": 3.0408,
+      "step": 21265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.495666663747026e-05,
+      "loss": 2.9857,
+      "step": 21270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.494993986812482e-05,
+      "loss": 2.9771,
+      "step": 21275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.494321186158462e-05,
+      "loss": 2.973,
+      "step": 21280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.493648261808784e-05,
+      "loss": 2.9322,
+      "step": 21285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.492975213787268e-05,
+      "loss": 2.9021,
+      "step": 21290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.492302042117739e-05,
+      "loss": 2.9379,
+      "step": 21295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.491628746824026e-05,
+      "loss": 2.9297,
+      "step": 21300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.490955327929964e-05,
+      "loss": 2.9995,
+      "step": 21305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.490281785459391e-05,
+      "loss": 3.0204,
+      "step": 21310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.48960811943615e-05,
+      "loss": 2.9379,
+      "step": 21315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.488934329884087e-05,
+      "loss": 2.9455,
+      "step": 21320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.488260416827054e-05,
+      "loss": 2.8929,
+      "step": 21325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.487586380288908e-05,
+      "loss": 3.0172,
+      "step": 21330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.48691222029351e-05,
+      "loss": 2.9741,
+      "step": 21335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.486237936864722e-05,
+      "loss": 2.9472,
+      "step": 21340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.485563530026415e-05,
+      "loss": 3.0135,
+      "step": 21345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.484888999802459e-05,
+      "loss": 2.9634,
+      "step": 21350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.484214346216737e-05,
+      "loss": 2.9443,
+      "step": 21355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.483539569293127e-05,
+      "loss": 2.8862,
+      "step": 21360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.482864669055519e-05,
+      "loss": 2.9701,
+      "step": 21365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.482189645527798e-05,
+      "loss": 2.9786,
+      "step": 21370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.481514498733864e-05,
+      "loss": 2.9527,
+      "step": 21375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.480839228697617e-05,
+      "loss": 2.9709,
+      "step": 21380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.48016383544296e-05,
+      "loss": 2.9384,
+      "step": 21385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.479488318993797e-05,
+      "loss": 2.9336,
+      "step": 21390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.478812679374048e-05,
+      "loss": 2.94,
+      "step": 21395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.478136916607624e-05,
+      "loss": 2.9334,
+      "step": 21400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.47746103071845e-05,
+      "loss": 2.9493,
+      "step": 21405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.47678502173045e-05,
+      "loss": 2.9934,
+      "step": 21410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.476108889667553e-05,
+      "loss": 2.8602,
+      "step": 21415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.475432634553697e-05,
+      "loss": 2.8981,
+      "step": 21420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.474756256412817e-05,
+      "loss": 2.9019,
+      "step": 21425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.474079755268858e-05,
+      "loss": 2.8746,
+      "step": 21430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.473403131145769e-05,
+      "loss": 2.9493,
+      "step": 21435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.472726384067499e-05,
+      "loss": 2.8965,
+      "step": 21440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.472049514058004e-05,
+      "loss": 2.9581,
+      "step": 21445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.471372521141247e-05,
+      "loss": 2.9454,
+      "step": 21450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.470695405341194e-05,
+      "loss": 2.8438,
+      "step": 21455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.470018166681809e-05,
+      "loss": 2.8654,
+      "step": 21460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.469340805187069e-05,
+      "loss": 2.9338,
+      "step": 21465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.468663320880955e-05,
+      "loss": 2.9667,
+      "step": 21470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.467985713787441e-05,
+      "loss": 2.8925,
+      "step": 21475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.467307983930522e-05,
+      "loss": 3.0252,
+      "step": 21480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.466630131334184e-05,
+      "loss": 2.9597,
+      "step": 21485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.465952156022425e-05,
+      "loss": 2.9672,
+      "step": 21490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.465274058019243e-05,
+      "loss": 3.0036,
+      "step": 21495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.464595837348642e-05,
+      "loss": 3.0003,
+      "step": 21500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.463917494034632e-05,
+      "loss": 2.8314,
+      "step": 21505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.463239028101225e-05,
+      "loss": 2.8743,
+      "step": 21510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.462560439572438e-05,
+      "loss": 2.935,
+      "step": 21515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.461881728472291e-05,
+      "loss": 2.8676,
+      "step": 21520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.461202894824812e-05,
+      "loss": 2.8907,
+      "step": 21525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.460523938654027e-05,
+      "loss": 3.0162,
+      "step": 21530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.459844859983976e-05,
+      "loss": 2.9449,
+      "step": 21535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.459165658838695e-05,
+      "loss": 2.8787,
+      "step": 21540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.458486335242227e-05,
+      "loss": 2.9671,
+      "step": 21545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.45780688921862e-05,
+      "loss": 2.9349,
+      "step": 21550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.457127320791925e-05,
+      "loss": 2.9449,
+      "step": 21555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.456447629986198e-05,
+      "loss": 3.0151,
+      "step": 21560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.455767816825499e-05,
+      "loss": 2.9937,
+      "step": 21565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.455087881333894e-05,
+      "loss": 2.9848,
+      "step": 21570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.454407823535452e-05,
+      "loss": 2.9459,
+      "step": 21575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.453727643454247e-05,
+      "loss": 2.8649,
+      "step": 21580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.453047341114356e-05,
+      "loss": 2.888,
+      "step": 21585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.45236691653986e-05,
+      "loss": 3.005,
+      "step": 21590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.451686369754846e-05,
+      "loss": 3.0201,
+      "step": 21595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.451005700783406e-05,
+      "loss": 2.9842,
+      "step": 21600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.450324909649632e-05,
+      "loss": 2.9587,
+      "step": 21605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.449643996377627e-05,
+      "loss": 2.8708,
+      "step": 21610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.448962960991493e-05,
+      "loss": 2.9804,
+      "step": 21615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.448281803515336e-05,
+      "loss": 2.9196,
+      "step": 21620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.447600523973272e-05,
+      "loss": 2.905,
+      "step": 21625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.446919122389416e-05,
+      "loss": 2.9438,
+      "step": 21630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.446237598787887e-05,
+      "loss": 2.8546,
+      "step": 21635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.445555953192816e-05,
+      "loss": 2.8865,
+      "step": 21640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.444874185628326e-05,
+      "loss": 2.9119,
+      "step": 21645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.444192296118555e-05,
+      "loss": 2.9885,
+      "step": 21650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.443510284687639e-05,
+      "loss": 2.8092,
+      "step": 21655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.442828151359724e-05,
+      "loss": 2.8771,
+      "step": 21660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.442145896158951e-05,
+      "loss": 2.8917,
+      "step": 21665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.441463519109476e-05,
+      "loss": 2.8934,
+      "step": 21670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.440781020235453e-05,
+      "loss": 2.9582,
+      "step": 21675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.440098399561042e-05,
+      "loss": 3.0535,
+      "step": 21680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.43941565711041e-05,
+      "loss": 2.995,
+      "step": 21685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.43873279290772e-05,
+      "loss": 2.9195,
+      "step": 21690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.438049806977146e-05,
+      "loss": 2.9026,
+      "step": 21695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.437366699342868e-05,
+      "loss": 2.8503,
+      "step": 21700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.436683470029066e-05,
+      "loss": 2.9382,
+      "step": 21705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.436000119059925e-05,
+      "loss": 2.8424,
+      "step": 21710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.435316646459635e-05,
+      "loss": 2.9366,
+      "step": 21715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.43463305225239e-05,
+      "loss": 2.8982,
+      "step": 21720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.43394933646239e-05,
+      "loss": 2.9033,
+      "step": 21725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.433265499113839e-05,
+      "loss": 2.9608,
+      "step": 21730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.43258154023094e-05,
+      "loss": 2.806,
+      "step": 21735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.431897459837907e-05,
+      "loss": 2.8149,
+      "step": 21740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.431213257958955e-05,
+      "loss": 2.9231,
+      "step": 21745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.430528934618305e-05,
+      "loss": 2.9575,
+      "step": 21750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.429844489840182e-05,
+      "loss": 2.9759,
+      "step": 21755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.429159923648814e-05,
+      "loss": 2.9535,
+      "step": 21760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.428475236068433e-05,
+      "loss": 2.9506,
+      "step": 21765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.427790427123277e-05,
+      "loss": 2.92,
+      "step": 21770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.427105496837586e-05,
+      "loss": 2.8897,
+      "step": 21775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.42642044523561e-05,
+      "loss": 2.8134,
+      "step": 21780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.425735272341598e-05,
+      "loss": 2.9998,
+      "step": 21785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.425049978179799e-05,
+      "loss": 2.9158,
+      "step": 21790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.424364562774479e-05,
+      "loss": 2.8932,
+      "step": 21795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.423679026149898e-05,
+      "loss": 2.9793,
+      "step": 21800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.422993368330322e-05,
+      "loss": 2.9576,
+      "step": 21805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.422307589340024e-05,
+      "loss": 2.9208,
+      "step": 21810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.421621689203281e-05,
+      "loss": 2.8631,
+      "step": 21815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.420935667944372e-05,
+      "loss": 2.9984,
+      "step": 21820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.42024952558758e-05,
+      "loss": 2.9073,
+      "step": 21825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.419563262157198e-05,
+      "loss": 2.8472,
+      "step": 21830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.418876877677515e-05,
+      "loss": 2.9958,
+      "step": 21835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.41819037217283e-05,
+      "loss": 2.9794,
+      "step": 21840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.417503745667444e-05,
+      "loss": 2.9002,
+      "step": 21845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.416816998185665e-05,
+      "loss": 2.9419,
+      "step": 21850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.4161301297518e-05,
+      "loss": 2.9339,
+      "step": 21855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.415443140390166e-05,
+      "loss": 2.8999,
+      "step": 21860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.414756030125081e-05,
+      "loss": 2.943,
+      "step": 21865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.414068798980868e-05,
+      "loss": 3.0049,
+      "step": 21870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.413381446981854e-05,
+      "loss": 2.8995,
+      "step": 21875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.412693974152371e-05,
+      "loss": 3.0457,
+      "step": 21880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.412006380516753e-05,
+      "loss": 2.8622,
+      "step": 21885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.411318666099344e-05,
+      "loss": 2.8955,
+      "step": 21890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.410630830924486e-05,
+      "loss": 2.9251,
+      "step": 21895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.409942875016529e-05,
+      "loss": 3.0051,
+      "step": 21900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.409254798399823e-05,
+      "loss": 2.8117,
+      "step": 21905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.408566601098729e-05,
+      "loss": 2.9269,
+      "step": 21910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.407878283137607e-05,
+      "loss": 3.0248,
+      "step": 21915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.407189844540821e-05,
+      "loss": 2.9227,
+      "step": 21920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.406501285332744e-05,
+      "loss": 2.8468,
+      "step": 21925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.405812605537748e-05,
+      "loss": 2.8658,
+      "step": 21930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.405123805180215e-05,
+      "loss": 2.8347,
+      "step": 21935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.404434884284525e-05,
+      "loss": 2.9008,
+      "step": 21940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.403745842875063e-05,
+      "loss": 2.8924,
+      "step": 21945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.403056680976227e-05,
+      "loss": 2.8411,
+      "step": 21950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.402367398612408e-05,
+      "loss": 2.9389,
+      "step": 21955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.401677995808004e-05,
+      "loss": 2.8957,
+      "step": 21960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.400988472587424e-05,
+      "loss": 2.8115,
+      "step": 21965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.400298828975074e-05,
+      "loss": 2.9301,
+      "step": 21970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.399609064995368e-05,
+      "loss": 2.895,
+      "step": 21975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.398919180672721e-05,
+      "loss": 2.9492,
+      "step": 21980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.398229176031556e-05,
+      "loss": 2.9892,
+      "step": 21985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.397539051096298e-05,
+      "loss": 2.9772,
+      "step": 21990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.396848805891378e-05,
+      "loss": 2.9653,
+      "step": 21995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.396158440441226e-05,
+      "loss": 2.9044,
+      "step": 22000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.395467954770285e-05,
+      "loss": 2.8875,
+      "step": 22005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.394777348902995e-05,
+      "loss": 2.9636,
+      "step": 22010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.394086622863803e-05,
+      "loss": 2.8992,
+      "step": 22015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.39339577667716e-05,
+      "loss": 2.9721,
+      "step": 22020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.392704810367522e-05,
+      "loss": 2.9087,
+      "step": 22025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.392013723959349e-05,
+      "loss": 3.0172,
+      "step": 22030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.391322517477105e-05,
+      "loss": 2.8434,
+      "step": 22035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.390631190945255e-05,
+      "loss": 2.8596,
+      "step": 22040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.389939744388274e-05,
+      "loss": 2.9236,
+      "step": 22045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.389248177830639e-05,
+      "loss": 2.9784,
+      "step": 22050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.388556491296828e-05,
+      "loss": 3.03,
+      "step": 22055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.387864684811329e-05,
+      "loss": 2.997,
+      "step": 22060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.387172758398629e-05,
+      "loss": 2.9658,
+      "step": 22065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.386480712083224e-05,
+      "loss": 2.9622,
+      "step": 22070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.38578854588961e-05,
+      "loss": 2.8691,
+      "step": 22075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.38509625984229e-05,
+      "loss": 2.9138,
+      "step": 22080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.384403853965769e-05,
+      "loss": 2.9245,
+      "step": 22085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.383711328284558e-05,
+      "loss": 2.8784,
+      "step": 22090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.383018682823173e-05,
+      "loss": 2.8894,
+      "step": 22095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.382325917606131e-05,
+      "loss": 2.9106,
+      "step": 22100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.381633032657956e-05,
+      "loss": 2.882,
+      "step": 22105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.380940028003177e-05,
+      "loss": 2.9391,
+      "step": 22110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.380246903666324e-05,
+      "loss": 2.953,
+      "step": 22115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.379553659671932e-05,
+      "loss": 2.8896,
+      "step": 22120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.378860296044544e-05,
+      "loss": 2.8851,
+      "step": 22125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.378166812808702e-05,
+      "loss": 2.974,
+      "step": 22130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.377473209988957e-05,
+      "loss": 2.9061,
+      "step": 22135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.376779487609858e-05,
+      "loss": 2.9248,
+      "step": 22140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.376085645695967e-05,
+      "loss": 2.9683,
+      "step": 22145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.375391684271841e-05,
+      "loss": 2.8808,
+      "step": 22150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.37469760336205e-05,
+      "loss": 2.9562,
+      "step": 22155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.37400340299116e-05,
+      "loss": 2.9334,
+      "step": 22160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.373309083183745e-05,
+      "loss": 2.9819,
+      "step": 22165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.372614643964386e-05,
+      "loss": 2.8081,
+      "step": 22170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.371920085357664e-05,
+      "loss": 2.9153,
+      "step": 22175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.371225407388166e-05,
+      "loss": 2.9221,
+      "step": 22180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.370530610080483e-05,
+      "loss": 2.9212,
+      "step": 22185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.36983569345921e-05,
+      "loss": 3.0687,
+      "step": 22190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.369140657548944e-05,
+      "loss": 2.9826,
+      "step": 22195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.368445502374293e-05,
+      "loss": 2.9581,
+      "step": 22200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.367750227959863e-05,
+      "loss": 2.9708,
+      "step": 22205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.367054834330265e-05,
+      "loss": 2.847,
+      "step": 22210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.366359321510116e-05,
+      "loss": 2.9387,
+      "step": 22215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.365663689524036e-05,
+      "loss": 2.9142,
+      "step": 22220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.364967938396652e-05,
+      "loss": 3.0582,
+      "step": 22225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.36427206815259e-05,
+      "loss": 3.0274,
+      "step": 22230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.363576078816486e-05,
+      "loss": 2.9794,
+      "step": 22235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.362879970412973e-05,
+      "loss": 2.9673,
+      "step": 22240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.362183742966698e-05,
+      "loss": 2.942,
+      "step": 22245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.361487396502304e-05,
+      "loss": 2.9116,
+      "step": 22250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.360790931044439e-05,
+      "loss": 2.9151,
+      "step": 22255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.36009434661776e-05,
+      "loss": 2.9166,
+      "step": 22260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.359397643246925e-05,
+      "loss": 3.0126,
+      "step": 22265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.358700820956597e-05,
+      "loss": 2.8999,
+      "step": 22270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.358003879771443e-05,
+      "loss": 2.8506,
+      "step": 22275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.35730681971613e-05,
+      "loss": 2.9577,
+      "step": 22280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.35660964081534e-05,
+      "loss": 2.9131,
+      "step": 22285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.355912343093749e-05,
+      "loss": 2.9649,
+      "step": 22290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.355214926576039e-05,
+      "loss": 2.9088,
+      "step": 22295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.3545173912869e-05,
+      "loss": 2.9373,
+      "step": 22300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.353819737251023e-05,
+      "loss": 2.9125,
+      "step": 22305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.353121964493107e-05,
+      "loss": 2.9501,
+      "step": 22310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.352424073037848e-05,
+      "loss": 2.9109,
+      "step": 22315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.351726062909955e-05,
+      "loss": 2.9404,
+      "step": 22320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.351027934134135e-05,
+      "loss": 2.9417,
+      "step": 22325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.3503296867351e-05,
+      "loss": 2.8757,
+      "step": 22330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.34963132073757e-05,
+      "loss": 2.9294,
+      "step": 22335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.348932836166265e-05,
+      "loss": 2.9248,
+      "step": 22340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.348234233045909e-05,
+      "loss": 2.9213,
+      "step": 22345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.347535511401234e-05,
+      "loss": 2.9954,
+      "step": 22350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.346836671256974e-05,
+      "loss": 2.9719,
+      "step": 22355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.346137712637867e-05,
+      "loss": 2.9968,
+      "step": 22360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.345438635568656e-05,
+      "loss": 2.9001,
+      "step": 22365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.344739440074087e-05,
+      "loss": 2.9671,
+      "step": 22370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.34404012617891e-05,
+      "loss": 2.9861,
+      "step": 22375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.343340693907881e-05,
+      "loss": 2.8866,
+      "step": 22380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.34264114328576e-05,
+      "loss": 3.0275,
+      "step": 22385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.341941474337309e-05,
+      "loss": 2.9697,
+      "step": 22390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.341241687087295e-05,
+      "loss": 2.9316,
+      "step": 22395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.340541781560494e-05,
+      "loss": 2.9276,
+      "step": 22400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.339841757781677e-05,
+      "loss": 2.9393,
+      "step": 22405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.339141615775626e-05,
+      "loss": 2.9138,
+      "step": 22410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.338441355567126e-05,
+      "loss": 2.8876,
+      "step": 22415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.337740977180966e-05,
+      "loss": 2.9385,
+      "step": 22420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.337040480641938e-05,
+      "loss": 2.9361,
+      "step": 22425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.336339865974838e-05,
+      "loss": 2.9561,
+      "step": 22430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.335639133204467e-05,
+      "loss": 2.9274,
+      "step": 22435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.334938282355633e-05,
+      "loss": 2.9495,
+      "step": 22440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.334237313453143e-05,
+      "loss": 2.9942,
+      "step": 22445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.333536226521811e-05,
+      "loss": 2.9238,
+      "step": 22450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.332835021586454e-05,
+      "loss": 2.8494,
+      "step": 22455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.332133698671896e-05,
+      "loss": 3.0198,
+      "step": 22460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.331432257802962e-05,
+      "loss": 2.9849,
+      "step": 22465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.330730699004483e-05,
+      "loss": 2.8586,
+      "step": 22470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.330029022301293e-05,
+      "loss": 2.9196,
+      "step": 22475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.32932722771823e-05,
+      "loss": 2.866,
+      "step": 22480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.328625315280139e-05,
+      "loss": 2.9303,
+      "step": 22485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.327923285011865e-05,
+      "loss": 2.8961,
+      "step": 22490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.32722113693826e-05,
+      "loss": 2.9259,
+      "step": 22495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.326518871084179e-05,
+      "loss": 2.9072,
+      "step": 22500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.325816487474482e-05,
+      "loss": 2.9976,
+      "step": 22505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.325113986134032e-05,
+      "loss": 2.9869,
+      "step": 22510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.324411367087697e-05,
+      "loss": 2.9386,
+      "step": 22515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.323708630360351e-05,
+      "loss": 2.9852,
+      "step": 22520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.323005775976866e-05,
+      "loss": 2.927,
+      "step": 22525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.322302803962127e-05,
+      "loss": 2.8798,
+      "step": 22530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.321599714341017e-05,
+      "loss": 2.9275,
+      "step": 22535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.320896507138425e-05,
+      "loss": 2.9314,
+      "step": 22540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.320193182379241e-05,
+      "loss": 2.842,
+      "step": 22545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.319489740088364e-05,
+      "loss": 3.058,
+      "step": 22550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.318786180290698e-05,
+      "loss": 3.0275,
+      "step": 22555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.318082503011144e-05,
+      "loss": 2.9275,
+      "step": 22560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.317378708274612e-05,
+      "loss": 2.9147,
+      "step": 22565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.316674796106019e-05,
+      "loss": 2.9079,
+      "step": 22570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.315970766530279e-05,
+      "loss": 2.9753,
+      "step": 22575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.315266619572317e-05,
+      "loss": 2.8166,
+      "step": 22580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.314562355257057e-05,
+      "loss": 2.857,
+      "step": 22585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.313857973609431e-05,
+      "loss": 2.9716,
+      "step": 22590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.313153474654372e-05,
+      "loss": 2.8236,
+      "step": 22595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.312448858416819e-05,
+      "loss": 2.939,
+      "step": 22600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.311744124921715e-05,
+      "loss": 2.8895,
+      "step": 22605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.311039274194007e-05,
+      "loss": 2.829,
+      "step": 22610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.310334306258646e-05,
+      "loss": 2.8511,
+      "step": 22615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.309629221140586e-05,
+      "loss": 2.9024,
+      "step": 22620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.308924018864789e-05,
+      "loss": 3.0195,
+      "step": 22625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.308218699456216e-05,
+      "loss": 2.9607,
+      "step": 22630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.307513262939835e-05,
+      "loss": 2.9209,
+      "step": 22635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.306807709340619e-05,
+      "loss": 2.9016,
+      "step": 22640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.306102038683545e-05,
+      "loss": 2.9004,
+      "step": 22645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.30539625099359e-05,
+      "loss": 2.8553,
+      "step": 22650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.304690346295738e-05,
+      "loss": 2.8152,
+      "step": 22655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.30398432461498e-05,
+      "loss": 2.9564,
+      "step": 22660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.303278185976306e-05,
+      "loss": 2.9362,
+      "step": 22665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.302571930404715e-05,
+      "loss": 2.8388,
+      "step": 22670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.301865557925208e-05,
+      "loss": 2.8646,
+      "step": 22675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.301159068562787e-05,
+      "loss": 2.8258,
+      "step": 22680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.300452462342465e-05,
+      "loss": 2.9606,
+      "step": 22685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.299745739289251e-05,
+      "loss": 2.9137,
+      "step": 22690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.299038899428164e-05,
+      "loss": 2.9008,
+      "step": 22695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.298331942784228e-05,
+      "loss": 2.9216,
+      "step": 22700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.297624869382465e-05,
+      "loss": 2.9145,
+      "step": 22705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.296917679247904e-05,
+      "loss": 2.9176,
+      "step": 22710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.296210372405584e-05,
+      "loss": 2.8253,
+      "step": 22715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.295502948880538e-05,
+      "loss": 3.0409,
+      "step": 22720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.294795408697811e-05,
+      "loss": 2.9609,
+      "step": 22725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.29408775188245e-05,
+      "loss": 2.8961,
+      "step": 22730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.293379978459501e-05,
+      "loss": 2.8484,
+      "step": 22735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.292672088454022e-05,
+      "loss": 2.9264,
+      "step": 22740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.291964081891071e-05,
+      "loss": 2.974,
+      "step": 22745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.29125595879571e-05,
+      "loss": 2.9742,
+      "step": 22750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.290547719193007e-05,
+      "loss": 2.881,
+      "step": 22755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.289839363108032e-05,
+      "loss": 2.9364,
+      "step": 22760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.289130890565861e-05,
+      "loss": 2.9236,
+      "step": 22765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.288422301591574e-05,
+      "loss": 2.8572,
+      "step": 22770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.287713596210251e-05,
+      "loss": 2.9482,
+      "step": 22775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.287004774446983e-05,
+      "loss": 2.975,
+      "step": 22780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.286295836326861e-05,
+      "loss": 2.773,
+      "step": 22785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.285586781874978e-05,
+      "loss": 2.9275,
+      "step": 22790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.28487761111644e-05,
+      "loss": 2.8703,
+      "step": 22795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.284168324076343e-05,
+      "loss": 2.9294,
+      "step": 22800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.2834589207798e-05,
+      "loss": 2.8993,
+      "step": 22805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.282749401251922e-05,
+      "loss": 2.9944,
+      "step": 22810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.282039765517826e-05,
+      "loss": 2.9082,
+      "step": 22815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.281330013602632e-05,
+      "loss": 2.8797,
+      "step": 22820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.280620145531466e-05,
+      "loss": 2.9535,
+      "step": 22825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.279910161329454e-05,
+      "loss": 2.9453,
+      "step": 22830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.27920006102173e-05,
+      "loss": 2.9693,
+      "step": 22835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.27848984463343e-05,
+      "loss": 2.8391,
+      "step": 22840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.277779512189696e-05,
+      "loss": 3.0055,
+      "step": 22845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.277069063715674e-05,
+      "loss": 2.954,
+      "step": 22850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.276358499236509e-05,
+      "loss": 2.8717,
+      "step": 22855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.275647818777362e-05,
+      "loss": 2.8157,
+      "step": 22860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.274937022363383e-05,
+      "loss": 2.9301,
+      "step": 22865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.274226110019737e-05,
+      "loss": 2.9032,
+      "step": 22870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.27351508177159e-05,
+      "loss": 3.0262,
+      "step": 22875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.27280393764411e-05,
+      "loss": 3.0561,
+      "step": 22880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.27209267766247e-05,
+      "loss": 2.8915,
+      "step": 22885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.271381301851851e-05,
+      "loss": 2.9334,
+      "step": 22890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.270669810237433e-05,
+      "loss": 2.9469,
+      "step": 22895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.269958202844404e-05,
+      "loss": 2.9544,
+      "step": 22900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.269246479697951e-05,
+      "loss": 2.9335,
+      "step": 22905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.26853464082327e-05,
+      "loss": 2.918,
+      "step": 22910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.267822686245562e-05,
+      "loss": 3.0217,
+      "step": 22915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.267110615990024e-05,
+      "loss": 2.8609,
+      "step": 22920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.266398430081867e-05,
+      "loss": 2.9333,
+      "step": 22925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.265686128546301e-05,
+      "loss": 2.8879,
+      "step": 22930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.264973711408541e-05,
+      "loss": 2.9462,
+      "step": 22935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.264261178693804e-05,
+      "loss": 2.926,
+      "step": 22940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.263548530427314e-05,
+      "loss": 2.9539,
+      "step": 22945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.262835766634299e-05,
+      "loss": 2.967,
+      "step": 22950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.262122887339988e-05,
+      "loss": 2.9046,
+      "step": 22955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.261409892569619e-05,
+      "loss": 2.8939,
+      "step": 22960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.260696782348429e-05,
+      "loss": 2.9131,
+      "step": 22965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.259983556701663e-05,
+      "loss": 2.9494,
+      "step": 22970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.259270215654567e-05,
+      "loss": 2.8937,
+      "step": 22975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.258556759232394e-05,
+      "loss": 2.8955,
+      "step": 22980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.257843187460399e-05,
+      "loss": 2.9372,
+      "step": 22985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.257129500363843e-05,
+      "loss": 2.8525,
+      "step": 22990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.256415697967987e-05,
+      "loss": 2.9376,
+      "step": 22995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.255701780298102e-05,
+      "loss": 2.9394,
+      "step": 23000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.254987747379459e-05,
+      "loss": 2.9859,
+      "step": 23005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.254273599237333e-05,
+      "loss": 2.9537,
+      "step": 23010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.253559335897005e-05,
+      "loss": 2.8285,
+      "step": 23015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.252844957383758e-05,
+      "loss": 2.8929,
+      "step": 23020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.252130463722883e-05,
+      "loss": 2.9167,
+      "step": 23025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.251415854939672e-05,
+      "loss": 2.9121,
+      "step": 23030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.250701131059419e-05,
+      "loss": 2.8631,
+      "step": 23035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.249986292107426e-05,
+      "loss": 2.9499,
+      "step": 23040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.249271338108997e-05,
+      "loss": 2.9858,
+      "step": 23045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.248556269089443e-05,
+      "loss": 2.9424,
+      "step": 23050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.247841085074072e-05,
+      "loss": 3.0115,
+      "step": 23055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.247125786088205e-05,
+      "loss": 2.9824,
+      "step": 23060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.246410372157163e-05,
+      "loss": 2.9092,
+      "step": 23065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.24569484330627e-05,
+      "loss": 2.923,
+      "step": 23070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.244979199560854e-05,
+      "loss": 2.8796,
+      "step": 23075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.24426344094625e-05,
+      "loss": 2.913,
+      "step": 23080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.243547567487792e-05,
+      "loss": 2.9469,
+      "step": 23085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.242831579210825e-05,
+      "loss": 2.9745,
+      "step": 23090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.242115476140691e-05,
+      "loss": 2.777,
+      "step": 23095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.241399258302742e-05,
+      "loss": 2.9846,
+      "step": 23100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.240682925722331e-05,
+      "loss": 2.8233,
+      "step": 23105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.239966478424814e-05,
+      "loss": 2.8868,
+      "step": 23110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.239249916435554e-05,
+      "loss": 2.9086,
+      "step": 23115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.238533239779916e-05,
+      "loss": 2.9575,
+      "step": 23120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.23781644848327e-05,
+      "loss": 3.0603,
+      "step": 23125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.237099542570989e-05,
+      "loss": 2.9094,
+      "step": 23130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.236382522068451e-05,
+      "loss": 2.9538,
+      "step": 23135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.235665387001039e-05,
+      "loss": 2.9098,
+      "step": 23140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.234948137394138e-05,
+      "loss": 2.9904,
+      "step": 23145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.234230773273137e-05,
+      "loss": 2.9999,
+      "step": 23150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.233513294663432e-05,
+      "loss": 2.907,
+      "step": 23155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.232795701590419e-05,
+      "loss": 2.8917,
+      "step": 23160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.232077994079504e-05,
+      "loss": 2.9923,
+      "step": 23165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.231360172156088e-05,
+      "loss": 2.9151,
+      "step": 23170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.230642235845582e-05,
+      "loss": 2.9302,
+      "step": 23175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.229924185173405e-05,
+      "loss": 2.8873,
+      "step": 23180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.22920602016497e-05,
+      "loss": 2.923,
+      "step": 23185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.228487740845703e-05,
+      "loss": 2.9148,
+      "step": 23190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.227769347241027e-05,
+      "loss": 2.9292,
+      "step": 23195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.227050839376378e-05,
+      "loss": 2.8367,
+      "step": 23200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.226332217277183e-05,
+      "loss": 2.9024,
+      "step": 23205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.225613480968885e-05,
+      "loss": 2.9685,
+      "step": 23210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.224894630476928e-05,
+      "loss": 2.9452,
+      "step": 23215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.224175665826756e-05,
+      "loss": 2.8653,
+      "step": 23220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.223456587043818e-05,
+      "loss": 2.9276,
+      "step": 23225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.222737394153575e-05,
+      "loss": 2.9469,
+      "step": 23230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.22201808718148e-05,
+      "loss": 2.9399,
+      "step": 23235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.221298666152996e-05,
+      "loss": 2.8237,
+      "step": 23240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.220579131093593e-05,
+      "loss": 2.911,
+      "step": 23245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.219859482028737e-05,
+      "loss": 2.801,
+      "step": 23250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.21913971898391e-05,
+      "loss": 2.9768,
+      "step": 23255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.218419841984586e-05,
+      "loss": 2.7592,
+      "step": 23260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.217699851056249e-05,
+      "loss": 2.8341,
+      "step": 23265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.216979746224386e-05,
+      "loss": 3.013,
+      "step": 23270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.216259527514487e-05,
+      "loss": 2.7669,
+      "step": 23275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.215539194952049e-05,
+      "loss": 2.968,
+      "step": 23280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.214818748562571e-05,
+      "loss": 2.9461,
+      "step": 23285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.214098188371554e-05,
+      "loss": 2.8985,
+      "step": 23290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.213377514404507e-05,
+      "loss": 2.9063,
+      "step": 23295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.21265672668694e-05,
+      "loss": 2.9301,
+      "step": 23300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.211935825244368e-05,
+      "loss": 2.9158,
+      "step": 23305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.211214810102312e-05,
+      "loss": 2.8187,
+      "step": 23310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.210493681286296e-05,
+      "loss": 2.9764,
+      "step": 23315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.209772438821843e-05,
+      "loss": 2.9512,
+      "step": 23320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.20905108273449e-05,
+      "loss": 2.9873,
+      "step": 23325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.208329613049765e-05,
+      "loss": 2.8969,
+      "step": 23330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.207608029793214e-05,
+      "loss": 2.9707,
+      "step": 23335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.206886332990378e-05,
+      "loss": 2.9765,
+      "step": 23340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.206164522666803e-05,
+      "loss": 2.8103,
+      "step": 23345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.205442598848044e-05,
+      "loss": 2.8727,
+      "step": 23350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.204720561559653e-05,
+      "loss": 2.9229,
+      "step": 23355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.203998410827189e-05,
+      "loss": 2.939,
+      "step": 23360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.203276146676219e-05,
+      "loss": 2.985,
+      "step": 23365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.202553769132307e-05,
+      "loss": 2.9288,
+      "step": 23370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.201831278221027e-05,
+      "loss": 2.9377,
+      "step": 23375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.201108673967954e-05,
+      "loss": 2.922,
+      "step": 23380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.200385956398666e-05,
+      "loss": 2.927,
+      "step": 23385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.199663125538749e-05,
+      "loss": 2.8783,
+      "step": 23390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.198940181413786e-05,
+      "loss": 2.8294,
+      "step": 23395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.198217124049372e-05,
+      "loss": 2.9168,
+      "step": 23400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.197493953471103e-05,
+      "loss": 2.9248,
+      "step": 23405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.196770669704578e-05,
+      "loss": 3.0189,
+      "step": 23410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.1960472727754e-05,
+      "loss": 3.0387,
+      "step": 23415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.195323762709176e-05,
+      "loss": 2.9743,
+      "step": 23420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.194600139531518e-05,
+      "loss": 2.9831,
+      "step": 23425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.193876403268044e-05,
+      "loss": 2.8479,
+      "step": 23430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.19315255394437e-05,
+      "loss": 2.9309,
+      "step": 23435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.192428591586121e-05,
+      "loss": 2.9584,
+      "step": 23440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.191704516218925e-05,
+      "loss": 2.9091,
+      "step": 23445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.190980327868413e-05,
+      "loss": 3.0544,
+      "step": 23450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.190256026560221e-05,
+      "loss": 3.0201,
+      "step": 23455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.18953161231999e-05,
+      "loss": 3.05,
+      "step": 23460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.18880708517336e-05,
+      "loss": 2.8645,
+      "step": 23465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.188082445145979e-05,
+      "loss": 2.9355,
+      "step": 23470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.187357692263502e-05,
+      "loss": 2.7581,
+      "step": 23475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.186632826551584e-05,
+      "loss": 2.941,
+      "step": 23480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.185907848035883e-05,
+      "loss": 2.8673,
+      "step": 23485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.18518275674206e-05,
+      "loss": 2.9256,
+      "step": 23490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.184457552695788e-05,
+      "loss": 3.0015,
+      "step": 23495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.183732235922735e-05,
+      "loss": 2.9537,
+      "step": 23500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.183006806448576e-05,
+      "loss": 3.0393,
+      "step": 23505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.182281264298992e-05,
+      "loss": 2.9023,
+      "step": 23510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.181555609499669e-05,
+      "loss": 2.8688,
+      "step": 23515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.18082984207629e-05,
+      "loss": 2.8826,
+      "step": 23520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.180103962054548e-05,
+      "loss": 2.9872,
+      "step": 23525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.17937796946014e-05,
+      "loss": 2.9175,
+      "step": 23530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.178651864318765e-05,
+      "loss": 2.8606,
+      "step": 23535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.177925646656124e-05,
+      "loss": 3.0182,
+      "step": 23540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.177199316497928e-05,
+      "loss": 2.9601,
+      "step": 23545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.176472873869883e-05,
+      "loss": 2.8336,
+      "step": 23550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.17574631879771e-05,
+      "loss": 2.9918,
+      "step": 23555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.175019651307127e-05,
+      "loss": 2.8521,
+      "step": 23560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.174292871423856e-05,
+      "loss": 2.9403,
+      "step": 23565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.173565979173625e-05,
+      "loss": 2.9562,
+      "step": 23570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.172838974582166e-05,
+      "loss": 2.9152,
+      "step": 23575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.172111857675214e-05,
+      "loss": 2.9646,
+      "step": 23580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.171384628478506e-05,
+      "loss": 2.9063,
+      "step": 23585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.170657287017787e-05,
+      "loss": 2.9304,
+      "step": 23590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.169929833318806e-05,
+      "loss": 2.8621,
+      "step": 23595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.16920226740731e-05,
+      "loss": 2.9154,
+      "step": 23600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.168474589309058e-05,
+      "loss": 3.0251,
+      "step": 23605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.167746799049806e-05,
+      "loss": 2.9868,
+      "step": 23610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.16701889665532e-05,
+      "loss": 2.9387,
+      "step": 23615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.166290882151366e-05,
+      "loss": 2.9543,
+      "step": 23620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.165562755563715e-05,
+      "loss": 2.9831,
+      "step": 23625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.164834516918142e-05,
+      "loss": 2.9778,
+      "step": 23630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.164106166240424e-05,
+      "loss": 2.8558,
+      "step": 23635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.163377703556348e-05,
+      "loss": 2.9052,
+      "step": 23640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.162649128891696e-05,
+      "loss": 3.0245,
+      "step": 23645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.161920442272261e-05,
+      "loss": 3.0206,
+      "step": 23650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.161191643723841e-05,
+      "loss": 3.0225,
+      "step": 23655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.160462733272232e-05,
+      "loss": 2.7732,
+      "step": 23660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.159733710943235e-05,
+      "loss": 2.9883,
+      "step": 23665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.159004576762658e-05,
+      "loss": 2.9307,
+      "step": 23670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.158275330756313e-05,
+      "loss": 2.9296,
+      "step": 23675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.157545972950015e-05,
+      "loss": 2.8496,
+      "step": 23680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.15681650336958e-05,
+      "loss": 2.9739,
+      "step": 23685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.156086922040831e-05,
+      "loss": 2.9188,
+      "step": 23690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.155357228989596e-05,
+      "loss": 2.9623,
+      "step": 23695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.154627424241706e-05,
+      "loss": 2.9336,
+      "step": 23700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.153897507822991e-05,
+      "loss": 2.9029,
+      "step": 23705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.153167479759296e-05,
+      "loss": 2.9233,
+      "step": 23710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.152437340076457e-05,
+      "loss": 2.931,
+      "step": 23715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.151707088800326e-05,
+      "loss": 2.9134,
+      "step": 23720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.150976725956748e-05,
+      "loss": 2.8597,
+      "step": 23725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.150246251571581e-05,
+      "loss": 2.8569,
+      "step": 23730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.149515665670679e-05,
+      "loss": 2.9234,
+      "step": 23735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.148784968279908e-05,
+      "loss": 2.8551,
+      "step": 23740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.148054159425134e-05,
+      "loss": 2.9434,
+      "step": 23745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.147323239132223e-05,
+      "loss": 2.973,
+      "step": 23750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.146592207427053e-05,
+      "loss": 2.8753,
+      "step": 23755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.1458610643355e-05,
+      "loss": 2.9539,
+      "step": 23760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.145129809883445e-05,
+      "loss": 2.9709,
+      "step": 23765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.144398444096776e-05,
+      "loss": 3.0174,
+      "step": 23770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.143666967001382e-05,
+      "loss": 2.9306,
+      "step": 23775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.142935378623153e-05,
+      "loss": 2.9211,
+      "step": 23780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.142203678987992e-05,
+      "loss": 2.9484,
+      "step": 23785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.141471868121797e-05,
+      "loss": 2.9448,
+      "step": 23790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.140739946050473e-05,
+      "loss": 2.9458,
+      "step": 23795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.140007912799933e-05,
+      "loss": 2.9434,
+      "step": 23800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.139275768396086e-05,
+      "loss": 2.9456,
+      "step": 23805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.138543512864853e-05,
+      "loss": 2.825,
+      "step": 23810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.137811146232153e-05,
+      "loss": 2.9551,
+      "step": 23815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.137078668523912e-05,
+      "loss": 2.9649,
+      "step": 23820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.136346079766057e-05,
+      "loss": 2.9118,
+      "step": 23825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.135613379984524e-05,
+      "loss": 2.9606,
+      "step": 23830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.134880569205247e-05,
+      "loss": 2.8356,
+      "step": 23835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.134147647454169e-05,
+      "loss": 2.9796,
+      "step": 23840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.133414614757232e-05,
+      "loss": 3.0245,
+      "step": 23845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.132681471140388e-05,
+      "loss": 2.9552,
+      "step": 23850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.131948216629587e-05,
+      "loss": 2.8916,
+      "step": 23855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.131214851250788e-05,
+      "loss": 2.9145,
+      "step": 23860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.13048137502995e-05,
+      "loss": 2.9891,
+      "step": 23865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.129747787993036e-05,
+      "loss": 2.8668,
+      "step": 23870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.129014090166019e-05,
+      "loss": 2.9856,
+      "step": 23875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.128280281574864e-05,
+      "loss": 3.0173,
+      "step": 23880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.127546362245553e-05,
+      "loss": 2.9843,
+      "step": 23885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.126812332204063e-05,
+      "loss": 2.8798,
+      "step": 23890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.126078191476379e-05,
+      "loss": 2.9842,
+      "step": 23895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.12534394008849e-05,
+      "loss": 2.8936,
+      "step": 23900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.124609578066386e-05,
+      "loss": 2.8749,
+      "step": 23905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.123875105436065e-05,
+      "loss": 2.8669,
+      "step": 23910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.123140522223524e-05,
+      "loss": 2.9723,
+      "step": 23915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.122405828454766e-05,
+      "loss": 2.9447,
+      "step": 23920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.121671024155802e-05,
+      "loss": 2.9178,
+      "step": 23925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.120936109352642e-05,
+      "loss": 2.9183,
+      "step": 23930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.1202010840713e-05,
+      "loss": 2.9031,
+      "step": 23935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.119465948337795e-05,
+      "loss": 2.8674,
+      "step": 23940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.118730702178153e-05,
+      "loss": 2.8832,
+      "step": 23945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.117995345618396e-05,
+      "loss": 2.9071,
+      "step": 23950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.117259878684562e-05,
+      "loss": 3.1178,
+      "step": 23955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.116524301402679e-05,
+      "loss": 2.9106,
+      "step": 23960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.11578861379879e-05,
+      "loss": 2.9313,
+      "step": 23965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.115052815898936e-05,
+      "loss": 2.9775,
+      "step": 23970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.114316907729163e-05,
+      "loss": 2.8858,
+      "step": 23975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.113580889315522e-05,
+      "loss": 2.9672,
+      "step": 23980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.112844760684068e-05,
+      "loss": 2.8815,
+      "step": 23985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.112108521860859e-05,
+      "loss": 2.9536,
+      "step": 23990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.111372172871957e-05,
+      "loss": 2.8685,
+      "step": 23995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.110635713743428e-05,
+      "loss": 2.9637,
+      "step": 24000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.109899144501342e-05,
+      "loss": 3.0228,
+      "step": 24005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.109162465171773e-05,
+      "loss": 2.9099,
+      "step": 24010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.108425675780799e-05,
+      "loss": 2.9153,
+      "step": 24015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.107688776354502e-05,
+      "loss": 2.8713,
+      "step": 24020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.106951766918965e-05,
+      "loss": 2.8394,
+      "step": 24025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.106214647500281e-05,
+      "loss": 3.0112,
+      "step": 24030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.105477418124541e-05,
+      "loss": 2.9449,
+      "step": 24035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.104740078817841e-05,
+      "loss": 2.9122,
+      "step": 24040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.104002629606286e-05,
+      "loss": 3.0233,
+      "step": 24045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.103265070515978e-05,
+      "loss": 2.8312,
+      "step": 24050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.102527401573026e-05,
+      "loss": 2.8994,
+      "step": 24055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.101789622803546e-05,
+      "loss": 2.8916,
+      "step": 24060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.101051734233649e-05,
+      "loss": 3.002,
+      "step": 24065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.100313735889463e-05,
+      "loss": 2.8536,
+      "step": 24070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.099575627797104e-05,
+      "loss": 2.9055,
+      "step": 24075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.098837409982708e-05,
+      "loss": 2.9742,
+      "step": 24080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.098099082472401e-05,
+      "loss": 2.9957,
+      "step": 24085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.097360645292322e-05,
+      "loss": 2.8198,
+      "step": 24090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.096622098468613e-05,
+      "loss": 2.9048,
+      "step": 24095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.095883442027415e-05,
+      "loss": 2.8933,
+      "step": 24100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.095144675994874e-05,
+      "loss": 2.9636,
+      "step": 24105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.094405800397146e-05,
+      "loss": 2.9987,
+      "step": 24110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.093666815260385e-05,
+      "loss": 2.8622,
+      "step": 24115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.092927720610751e-05,
+      "loss": 2.8989,
+      "step": 24120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.092188516474405e-05,
+      "loss": 3.0403,
+      "step": 24125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.091449202877515e-05,
+      "loss": 2.9439,
+      "step": 24130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.090709779846253e-05,
+      "loss": 2.9706,
+      "step": 24135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.089970247406791e-05,
+      "loss": 2.8967,
+      "step": 24140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.089230605585312e-05,
+      "loss": 2.8879,
+      "step": 24145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.088490854407997e-05,
+      "loss": 2.9031,
+      "step": 24150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.087750993901031e-05,
+      "loss": 2.8867,
+      "step": 24155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.087011024090606e-05,
+      "loss": 2.9582,
+      "step": 24160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.086270945002915e-05,
+      "loss": 2.8463,
+      "step": 24165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.085530756664157e-05,
+      "loss": 2.8714,
+      "step": 24170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.084790459100534e-05,
+      "loss": 2.8245,
+      "step": 24175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.084050052338251e-05,
+      "loss": 2.9705,
+      "step": 24180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.083309536403518e-05,
+      "loss": 2.924,
+      "step": 24185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.08256891132255e-05,
+      "loss": 2.9786,
+      "step": 24190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.081828177121562e-05,
+      "loss": 2.932,
+      "step": 24195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.081087333826777e-05,
+      "loss": 2.9357,
+      "step": 24200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.08034638146442e-05,
+      "loss": 2.9222,
+      "step": 24205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.079605320060719e-05,
+      "loss": 2.8936,
+      "step": 24210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.07886414964191e-05,
+      "loss": 2.865,
+      "step": 24215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.078122870234223e-05,
+      "loss": 2.8573,
+      "step": 24220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.077381481863906e-05,
+      "loss": 2.9613,
+      "step": 24225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.0766399845572e-05,
+      "loss": 2.9192,
+      "step": 24230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.075898378340352e-05,
+      "loss": 2.8702,
+      "step": 24235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.075156663239616e-05,
+      "loss": 2.9304,
+      "step": 24240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.07441483928125e-05,
+      "loss": 3.0352,
+      "step": 24245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.07367290649151e-05,
+      "loss": 2.9127,
+      "step": 24250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.072930864896662e-05,
+      "loss": 2.9577,
+      "step": 24255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.072188714522973e-05,
+      "loss": 2.9241,
+      "step": 24260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.071446455396711e-05,
+      "loss": 2.9039,
+      "step": 24265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.070704087544158e-05,
+      "loss": 2.9913,
+      "step": 24270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.06996161099159e-05,
+      "loss": 3.05,
+      "step": 24275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.069219025765289e-05,
+      "loss": 2.9753,
+      "step": 24280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.068476331891543e-05,
+      "loss": 2.8724,
+      "step": 24285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.067733529396642e-05,
+      "loss": 2.9617,
+      "step": 24290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.06699061830688e-05,
+      "loss": 2.9083,
+      "step": 24295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.066247598648556e-05,
+      "loss": 2.9792,
+      "step": 24300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.065504470447973e-05,
+      "loss": 2.8763,
+      "step": 24305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.064761233731435e-05,
+      "loss": 2.9026,
+      "step": 24310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.064017888525254e-05,
+      "loss": 2.886,
+      "step": 24315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.063274434855744e-05,
+      "loss": 2.9692,
+      "step": 24320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.06253087274922e-05,
+      "loss": 2.9912,
+      "step": 24325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.061787202232005e-05,
+      "loss": 2.8324,
+      "step": 24330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.061043423330424e-05,
+      "loss": 2.9089,
+      "step": 24335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.060299536070805e-05,
+      "loss": 2.8413,
+      "step": 24340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.059555540479484e-05,
+      "loss": 2.9865,
+      "step": 24345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.058811436582795e-05,
+      "loss": 2.8751,
+      "step": 24350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.05806722440708e-05,
+      "loss": 2.9521,
+      "step": 24355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.057322903978681e-05,
+      "loss": 2.9591,
+      "step": 24360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.056578475323951e-05,
+      "loss": 2.9235,
+      "step": 24365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.055833938469234e-05,
+      "loss": 3.0113,
+      "step": 24370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.055089293440896e-05,
+      "loss": 2.878,
+      "step": 24375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.054344540265289e-05,
+      "loss": 2.9928,
+      "step": 24380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.05359967896878e-05,
+      "loss": 2.8827,
+      "step": 24385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.052854709577736e-05,
+      "loss": 2.8974,
+      "step": 24390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.052109632118526e-05,
+      "loss": 2.9661,
+      "step": 24395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.051364446617528e-05,
+      "loss": 2.9829,
+      "step": 24400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.05061915310112e-05,
+      "loss": 2.9476,
+      "step": 24405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.049873751595685e-05,
+      "loss": 2.9014,
+      "step": 24410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.049128242127608e-05,
+      "loss": 2.9094,
+      "step": 24415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.04838262472328e-05,
+      "loss": 2.9687,
+      "step": 24420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.047636899409097e-05,
+      "loss": 2.9535,
+      "step": 24425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.046891066211453e-05,
+      "loss": 2.9508,
+      "step": 24430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.046145125156753e-05,
+      "loss": 2.9597,
+      "step": 24435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.045399076271403e-05,
+      "loss": 2.8475,
+      "step": 24440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.044652919581809e-05,
+      "loss": 2.8875,
+      "step": 24445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.043906655114389e-05,
+      "loss": 2.8803,
+      "step": 24450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.043160282895554e-05,
+      "loss": 2.9539,
+      "step": 24455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.042413802951731e-05,
+      "loss": 2.9281,
+      "step": 24460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.041667215309341e-05,
+      "loss": 2.9765,
+      "step": 24465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.040920519994815e-05,
+      "loss": 2.9147,
+      "step": 24470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.040173717034582e-05,
+      "loss": 2.8718,
+      "step": 24475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.03942680645508e-05,
+      "loss": 2.8336,
+      "step": 24480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.03867978828275e-05,
+      "loss": 2.9261,
+      "step": 24485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.037932662544033e-05,
+      "loss": 2.9433,
+      "step": 24490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.03718542926538e-05,
+      "loss": 2.869,
+      "step": 24495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.03643808847324e-05,
+      "loss": 2.9176,
+      "step": 24500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.035690640194067e-05,
+      "loss": 2.9766,
+      "step": 24505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.034943084454324e-05,
+      "loss": 2.8865,
+      "step": 24510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.03419542128047e-05,
+      "loss": 2.834,
+      "step": 24515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.033447650698973e-05,
+      "loss": 2.8532,
+      "step": 24520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.032699772736304e-05,
+      "loss": 2.8811,
+      "step": 24525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.031951787418936e-05,
+      "loss": 2.8825,
+      "step": 24530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.031203694773347e-05,
+      "loss": 2.9444,
+      "step": 24535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.030455494826019e-05,
+      "loss": 2.9043,
+      "step": 24540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.029707187603439e-05,
+      "loss": 2.9304,
+      "step": 24545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.028958773132094e-05,
+      "loss": 2.8959,
+      "step": 24550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.02821025143848e-05,
+      "loss": 2.9594,
+      "step": 24555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.027461622549091e-05,
+      "loss": 2.9813,
+      "step": 24560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.026712886490431e-05,
+      "loss": 2.8498,
+      "step": 24565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.025964043289001e-05,
+      "loss": 2.8662,
+      "step": 24570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.02521509297131e-05,
+      "loss": 2.8651,
+      "step": 24575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.024466035563871e-05,
+      "loss": 2.9109,
+      "step": 24580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.023716871093201e-05,
+      "loss": 2.9069,
+      "step": 24585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.022967599585817e-05,
+      "loss": 2.9973,
+      "step": 24590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.022218221068247e-05,
+      "loss": 2.9161,
+      "step": 24595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.021468735567014e-05,
+      "loss": 2.8729,
+      "step": 24600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.02071914310865e-05,
+      "loss": 2.9179,
+      "step": 24605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.019969443719692e-05,
+      "loss": 2.9242,
+      "step": 24610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.019219637426676e-05,
+      "loss": 2.8843,
+      "step": 24615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.018469724256146e-05,
+      "loss": 2.9112,
+      "step": 24620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.017719704234647e-05,
+      "loss": 2.944,
+      "step": 24625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.016969577388731e-05,
+      "loss": 2.9298,
+      "step": 24630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.016219343744949e-05,
+      "loss": 2.969,
+      "step": 24635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.015469003329861e-05,
+      "loss": 2.9332,
+      "step": 24640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.014718556170027e-05,
+      "loss": 2.9057,
+      "step": 24645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.013968002292012e-05,
+      "loss": 2.9609,
+      "step": 24650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.013217341722386e-05,
+      "loss": 2.9141,
+      "step": 24655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.012466574487722e-05,
+      "loss": 2.9166,
+      "step": 24660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.011715700614595e-05,
+      "loss": 2.957,
+      "step": 24665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.010964720129584e-05,
+      "loss": 3.0458,
+      "step": 24670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.010213633059277e-05,
+      "loss": 2.9246,
+      "step": 24675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.009462439430257e-05,
+      "loss": 2.9492,
+      "step": 24680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.008711139269121e-05,
+      "loss": 2.8467,
+      "step": 24685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.00795973260246e-05,
+      "loss": 2.9223,
+      "step": 24690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.007208219456874e-05,
+      "loss": 2.9691,
+      "step": 24695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.006456599858967e-05,
+      "loss": 2.8867,
+      "step": 24700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.005704873835345e-05,
+      "loss": 2.8811,
+      "step": 24705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.004953041412617e-05,
+      "loss": 2.9332,
+      "step": 24710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.0042011026174e-05,
+      "loss": 2.9569,
+      "step": 24715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.00344905747631e-05,
+      "loss": 2.91,
+      "step": 24720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.00269690601597e-05,
+      "loss": 2.9542,
+      "step": 24725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.001944648263004e-05,
+      "loss": 2.8729,
+      "step": 24730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.00119228424404e-05,
+      "loss": 2.8982,
+      "step": 24735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.000439813985714e-05,
+      "loss": 2.9075,
+      "step": 24740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.999687237514662e-05,
+      "loss": 2.9977,
+      "step": 24745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.998934554857524e-05,
+      "loss": 2.8227,
+      "step": 24750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.998181766040942e-05,
+      "loss": 2.9213,
+      "step": 24755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.997428871091568e-05,
+      "loss": 2.8604,
+      "step": 24760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.996675870036053e-05,
+      "loss": 2.9698,
+      "step": 24765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.99592276290105e-05,
+      "loss": 2.9768,
+      "step": 24770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.995169549713221e-05,
+      "loss": 2.8934,
+      "step": 24775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.994416230499226e-05,
+      "loss": 2.8856,
+      "step": 24780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.993662805285735e-05,
+      "loss": 2.8668,
+      "step": 24785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.992909274099416e-05,
+      "loss": 2.9673,
+      "step": 24790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.992155636966945e-05,
+      "loss": 2.9946,
+      "step": 24795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.991401893915e-05,
+      "loss": 2.9263,
+      "step": 24800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.990648044970261e-05,
+      "loss": 2.9271,
+      "step": 24805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.989894090159417e-05,
+      "loss": 2.9792,
+      "step": 24810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.989140029509153e-05,
+      "loss": 3.0796,
+      "step": 24815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.988385863046165e-05,
+      "loss": 2.8542,
+      "step": 24820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.987631590797149e-05,
+      "loss": 2.8875,
+      "step": 24825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.986877212788806e-05,
+      "loss": 2.9471,
+      "step": 24830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.986122729047838e-05,
+      "loss": 3.0302,
+      "step": 24835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.985368139600955e-05,
+      "loss": 2.9275,
+      "step": 24840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.984613444474869e-05,
+      "loss": 2.9727,
+      "step": 24845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.983858643696296e-05,
+      "loss": 2.9956,
+      "step": 24850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.983103737291953e-05,
+      "loss": 2.9684,
+      "step": 24855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.982348725288563e-05,
+      "loss": 2.9867,
+      "step": 24860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.981593607712856e-05,
+      "loss": 2.9187,
+      "step": 24865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.980838384591559e-05,
+      "loss": 2.7559,
+      "step": 24870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.980083055951408e-05,
+      "loss": 2.9107,
+      "step": 24875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.97932762181914e-05,
+      "loss": 2.9436,
+      "step": 24880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.978572082221497e-05,
+      "loss": 2.9233,
+      "step": 24885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.977816437185226e-05,
+      "loss": 2.9518,
+      "step": 24890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.977060686737072e-05,
+      "loss": 2.8858,
+      "step": 24895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.976304830903791e-05,
+      "loss": 2.8405,
+      "step": 24900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.975548869712141e-05,
+      "loss": 2.9129,
+      "step": 24905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.97479280318888e-05,
+      "loss": 2.8463,
+      "step": 24910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.974036631360771e-05,
+      "loss": 2.9582,
+      "step": 24915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.973280354254583e-05,
+      "loss": 2.9976,
+      "step": 24920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.972523971897088e-05,
+      "loss": 2.8918,
+      "step": 24925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.971767484315061e-05,
+      "loss": 2.9819,
+      "step": 24930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.97101089153528e-05,
+      "loss": 2.9059,
+      "step": 24935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.97025419358453e-05,
+      "loss": 2.9692,
+      "step": 24940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.969497390489594e-05,
+      "loss": 2.9752,
+      "step": 24945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.968740482277266e-05,
+      "loss": 2.8973,
+      "step": 24950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.967983468974336e-05,
+      "loss": 2.85,
+      "step": 24955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.967226350607604e-05,
+      "loss": 2.8957,
+      "step": 24960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.96646912720387e-05,
+      "loss": 2.8015,
+      "step": 24965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.965711798789941e-05,
+      "loss": 2.9012,
+      "step": 24970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.964954365392623e-05,
+      "loss": 3.1074,
+      "step": 24975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.96419682703873e-05,
+      "loss": 2.945,
+      "step": 24980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.963439183755077e-05,
+      "loss": 2.8779,
+      "step": 24985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.962681435568487e-05,
+      "loss": 3.034,
+      "step": 24990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.961923582505779e-05,
+      "loss": 2.9537,
+      "step": 24995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.961165624593782e-05,
+      "loss": 2.8138,
+      "step": 25000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.960407561859332e-05,
+      "loss": 2.9195,
+      "step": 25005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.959649394329255e-05,
+      "loss": 2.9931,
+      "step": 25010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.958891122030393e-05,
+      "loss": 3.0728,
+      "step": 25015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.958132744989593e-05,
+      "loss": 2.9356,
+      "step": 25020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.957374263233695e-05,
+      "loss": 3.0326,
+      "step": 25025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.956615676789548e-05,
+      "loss": 3.055,
+      "step": 25030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.955856985684009e-05,
+      "loss": 2.8615,
+      "step": 25035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.955098189943933e-05,
+      "loss": 2.8917,
+      "step": 25040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.954339289596181e-05,
+      "loss": 3.0198,
+      "step": 25045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.953580284667617e-05,
+      "loss": 2.9472,
+      "step": 25050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.95282117518511e-05,
+      "loss": 3.0154,
+      "step": 25055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.952061961175532e-05,
+      "loss": 2.8749,
+      "step": 25060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.951302642665757e-05,
+      "loss": 2.9944,
+      "step": 25065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.950543219682665e-05,
+      "loss": 2.9742,
+      "step": 25070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.949783692253136e-05,
+      "loss": 2.9932,
+      "step": 25075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.949024060404061e-05,
+      "loss": 3.0734,
+      "step": 25080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.948264324162329e-05,
+      "loss": 2.8439,
+      "step": 25085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.947504483554833e-05,
+      "loss": 2.9034,
+      "step": 25090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.946744538608472e-05,
+      "loss": 2.9793,
+      "step": 25095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.945984489350145e-05,
+      "loss": 2.9271,
+      "step": 25100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.945224335806758e-05,
+      "loss": 2.9696,
+      "step": 25105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.944464078005221e-05,
+      "loss": 2.9143,
+      "step": 25110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.943703715972445e-05,
+      "loss": 2.9482,
+      "step": 25115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.942943249735347e-05,
+      "loss": 2.8785,
+      "step": 25120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.942182679320847e-05,
+      "loss": 2.9471,
+      "step": 25125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.941422004755865e-05,
+      "loss": 2.9631,
+      "step": 25130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.940661226067335e-05,
+      "loss": 2.8324,
+      "step": 25135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.93990034328218e-05,
+      "loss": 2.8992,
+      "step": 25140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.939139356427341e-05,
+      "loss": 2.9312,
+      "step": 25145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.938378265529753e-05,
+      "loss": 2.9356,
+      "step": 25150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.937617070616357e-05,
+      "loss": 2.9227,
+      "step": 25155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.936855771714103e-05,
+      "loss": 2.9372,
+      "step": 25160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.936094368849933e-05,
+      "loss": 2.8212,
+      "step": 25165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.935332862050808e-05,
+      "loss": 2.8579,
+      "step": 25170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.934571251343678e-05,
+      "loss": 2.8307,
+      "step": 25175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.933809536755509e-05,
+      "loss": 2.9737,
+      "step": 25180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.93304771831326e-05,
+      "loss": 2.912,
+      "step": 25185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.932285796043902e-05,
+      "loss": 3.0044,
+      "step": 25190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.931523769974405e-05,
+      "loss": 2.9361,
+      "step": 25195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.930761640131744e-05,
+      "loss": 3.0114,
+      "step": 25200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.929999406542899e-05,
+      "loss": 2.9626,
+      "step": 25205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.929237069234849e-05,
+      "loss": 2.9245,
+      "step": 25210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.928474628234584e-05,
+      "loss": 2.9387,
+      "step": 25215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.927712083569089e-05,
+      "loss": 2.8331,
+      "step": 25220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.926949435265362e-05,
+      "loss": 2.8427,
+      "step": 25225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.926186683350398e-05,
+      "loss": 2.9132,
+      "step": 25230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.925423827851199e-05,
+      "loss": 2.9906,
+      "step": 25235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.924660868794767e-05,
+      "loss": 2.8272,
+      "step": 25240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.923897806208112e-05,
+      "loss": 2.9143,
+      "step": 25245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.923134640118245e-05,
+      "loss": 2.9117,
+      "step": 25250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.922371370552182e-05,
+      "loss": 2.9901,
+      "step": 25255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.92160799753694e-05,
+      "loss": 2.802,
+      "step": 25260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.920844521099543e-05,
+      "loss": 2.9798,
+      "step": 25265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.920080941267019e-05,
+      "loss": 2.9763,
+      "step": 25270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.919317258066394e-05,
+      "loss": 2.9032,
+      "step": 25275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.918553471524705e-05,
+      "loss": 2.9996,
+      "step": 25280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.91778958166899e-05,
+      "loss": 2.8187,
+      "step": 25285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.917025588526288e-05,
+      "loss": 2.9528,
+      "step": 25290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.916261492123642e-05,
+      "loss": 2.9551,
+      "step": 25295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.915497292488105e-05,
+      "loss": 2.9237,
+      "step": 25300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.914732989646723e-05,
+      "loss": 2.9448,
+      "step": 25305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.913968583626558e-05,
+      "loss": 2.9145,
+      "step": 25310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.913204074454664e-05,
+      "loss": 2.8183,
+      "step": 25315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.912439462158108e-05,
+      "loss": 2.8856,
+      "step": 25320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.911674746763953e-05,
+      "loss": 2.9644,
+      "step": 25325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.91090992829927e-05,
+      "loss": 2.9475,
+      "step": 25330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.910145006791134e-05,
+      "loss": 2.9717,
+      "step": 25335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.909379982266622e-05,
+      "loss": 2.875,
+      "step": 25340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.908614854752815e-05,
+      "loss": 2.8947,
+      "step": 25345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.9078496242768e-05,
+      "loss": 3.0313,
+      "step": 25350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.907084290865662e-05,
+      "loss": 2.8918,
+      "step": 25355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.906318854546497e-05,
+      "loss": 2.9779,
+      "step": 25360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.905553315346394e-05,
+      "loss": 2.8807,
+      "step": 25365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.90478767329246e-05,
+      "loss": 3.0079,
+      "step": 25370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.904021928411792e-05,
+      "loss": 2.8721,
+      "step": 25375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.9032560807315e-05,
+      "loss": 2.9453,
+      "step": 25380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.902490130278694e-05,
+      "loss": 3.0337,
+      "step": 25385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.901724077080488e-05,
+      "loss": 2.9717,
+      "step": 25390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.900957921163998e-05,
+      "loss": 2.9851,
+      "step": 25395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.900191662556348e-05,
+      "loss": 2.9224,
+      "step": 25400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.899425301284659e-05,
+      "loss": 2.8983,
+      "step": 25405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.898658837376063e-05,
+      "loss": 2.975,
+      "step": 25410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.897892270857689e-05,
+      "loss": 2.8782,
+      "step": 25415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.897125601756677e-05,
+      "loss": 2.9472,
+      "step": 25420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.896358830100159e-05,
+      "loss": 2.8812,
+      "step": 25425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.895591955915287e-05,
+      "loss": 3.0369,
+      "step": 25430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.894824979229201e-05,
+      "loss": 2.9872,
+      "step": 25435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.894057900069055e-05,
+      "loss": 2.9041,
+      "step": 25440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.893290718462001e-05,
+      "loss": 2.9084,
+      "step": 25445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.892523434435196e-05,
+      "loss": 2.9811,
+      "step": 25450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.891756048015803e-05,
+      "loss": 2.9214,
+      "step": 25455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.890988559230985e-05,
+      "loss": 2.9369,
+      "step": 25460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.890220968107912e-05,
+      "loss": 2.9858,
+      "step": 25465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.889453274673755e-05,
+      "loss": 2.9839,
+      "step": 25470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.888685478955689e-05,
+      "loss": 2.9781,
+      "step": 25475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.887917580980895e-05,
+      "loss": 2.8605,
+      "step": 25480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.887149580776554e-05,
+      "loss": 2.9163,
+      "step": 25485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.886381478369853e-05,
+      "loss": 2.9158,
+      "step": 25490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.885613273787984e-05,
+      "loss": 2.8949,
+      "step": 25495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.884844967058138e-05,
+      "loss": 3.0116,
+      "step": 25500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.884076558207515e-05,
+      "loss": 2.984,
+      "step": 25505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.883308047263314e-05,
+      "loss": 2.8722,
+      "step": 25510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.882539434252739e-05,
+      "loss": 2.9274,
+      "step": 25515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.881770719203e-05,
+      "loss": 2.9456,
+      "step": 25520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.881001902141307e-05,
+      "loss": 3.0253,
+      "step": 25525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.880232983094876e-05,
+      "loss": 2.8232,
+      "step": 25530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.879463962090928e-05,
+      "loss": 3.013,
+      "step": 25535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.878694839156685e-05,
+      "loss": 2.9656,
+      "step": 25540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.877925614319369e-05,
+      "loss": 2.9573,
+      "step": 25545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.877156287606215e-05,
+      "loss": 2.9571,
+      "step": 25550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.876386859044453e-05,
+      "loss": 2.9585,
+      "step": 25555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.875617328661324e-05,
+      "loss": 2.9399,
+      "step": 25560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.874847696484064e-05,
+      "loss": 2.9183,
+      "step": 25565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.874077962539921e-05,
+      "loss": 2.9145,
+      "step": 25570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.87330812685614e-05,
+      "loss": 2.9341,
+      "step": 25575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.872538189459974e-05,
+      "loss": 2.9521,
+      "step": 25580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.871768150378678e-05,
+      "loss": 2.9978,
+      "step": 25585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.87099800963951e-05,
+      "loss": 2.7685,
+      "step": 25590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.870227767269733e-05,
+      "loss": 2.8518,
+      "step": 25595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.869457423296614e-05,
+      "loss": 2.9795,
+      "step": 25600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.868686977747419e-05,
+      "loss": 2.9771,
+      "step": 25605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.867916430649423e-05,
+      "loss": 2.8759,
+      "step": 25610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.867145782029904e-05,
+      "loss": 2.8994,
+      "step": 25615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.86637503191614e-05,
+      "loss": 2.8687,
+      "step": 25620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.865604180335415e-05,
+      "loss": 2.9744,
+      "step": 25625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.864833227315018e-05,
+      "loss": 2.8887,
+      "step": 25630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.86406217288224e-05,
+      "loss": 2.8831,
+      "step": 25635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.863291017064373e-05,
+      "loss": 2.9671,
+      "step": 25640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.862519759888719e-05,
+      "loss": 3.0172,
+      "step": 25645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.861748401382576e-05,
+      "loss": 2.9858,
+      "step": 25650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.860976941573252e-05,
+      "loss": 3.0226,
+      "step": 25655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.860205380488054e-05,
+      "loss": 2.8799,
+      "step": 25660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.859433718154297e-05,
+      "loss": 3.0297,
+      "step": 25665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.858661954599294e-05,
+      "loss": 2.9018,
+      "step": 25670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.857890089850367e-05,
+      "loss": 2.8706,
+      "step": 25675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.857118123934839e-05,
+      "loss": 2.8535,
+      "step": 25680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.856346056880036e-05,
+      "loss": 3.0648,
+      "step": 25685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.855573888713288e-05,
+      "loss": 2.9137,
+      "step": 25690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.854801619461933e-05,
+      "loss": 2.914,
+      "step": 25695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.854029249153303e-05,
+      "loss": 2.8708,
+      "step": 25700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.853256777814743e-05,
+      "loss": 2.821,
+      "step": 25705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.852484205473595e-05,
+      "loss": 2.9716,
+      "step": 25710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.851711532157211e-05,
+      "loss": 2.9579,
+      "step": 25715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.850938757892939e-05,
+      "loss": 2.9283,
+      "step": 25720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.850165882708136e-05,
+      "loss": 2.9492,
+      "step": 25725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.84939290663016e-05,
+      "loss": 3.025,
+      "step": 25730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.848619829686377e-05,
+      "loss": 2.9007,
+      "step": 25735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.847846651904151e-05,
+      "loss": 2.9814,
+      "step": 25740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.847073373310853e-05,
+      "loss": 2.9213,
+      "step": 25745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.846299993933853e-05,
+      "loss": 2.9549,
+      "step": 25750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.845526513800531e-05,
+      "loss": 2.8828,
+      "step": 25755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.844752932938266e-05,
+      "loss": 2.8809,
+      "step": 25760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.843979251374442e-05,
+      "loss": 2.9174,
+      "step": 25765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.843205469136447e-05,
+      "loss": 2.9125,
+      "step": 25770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.842431586251675e-05,
+      "loss": 2.986,
+      "step": 25775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.841657602747515e-05,
+      "loss": 2.9903,
+      "step": 25780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.84088351865137e-05,
+      "loss": 2.9582,
+      "step": 25785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.84010933399064e-05,
+      "loss": 2.9737,
+      "step": 25790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.83933504879273e-05,
+      "loss": 2.925,
+      "step": 25795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.83856066308505e-05,
+      "loss": 2.8979,
+      "step": 25800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.837786176895012e-05,
+      "loss": 2.9128,
+      "step": 25805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.837011590250031e-05,
+      "loss": 2.9633,
+      "step": 25810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.836236903177528e-05,
+      "loss": 3.0919,
+      "step": 25815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.835462115704926e-05,
+      "loss": 2.8266,
+      "step": 25820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.834687227859652e-05,
+      "loss": 2.8477,
+      "step": 25825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.833912239669137e-05,
+      "loss": 2.925,
+      "step": 25830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.833137151160812e-05,
+      "loss": 2.9465,
+      "step": 25835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.832361962362117e-05,
+      "loss": 2.9292,
+      "step": 25840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.831586673300493e-05,
+      "loss": 2.9531,
+      "step": 25845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.830811284003382e-05,
+      "loss": 2.9286,
+      "step": 25850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.830035794498235e-05,
+      "loss": 2.8631,
+      "step": 25855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.8292602048125e-05,
+      "loss": 2.9135,
+      "step": 25860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.828484514973638e-05,
+      "loss": 2.9714,
+      "step": 25865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.827708725009102e-05,
+      "loss": 2.9197,
+      "step": 25870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.826932834946356e-05,
+      "loss": 2.9578,
+      "step": 25875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.826156844812866e-05,
+      "loss": 2.9218,
+      "step": 25880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.825380754636104e-05,
+      "loss": 2.9167,
+      "step": 25885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.824604564443538e-05,
+      "loss": 2.9085,
+      "step": 25890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.823828274262646e-05,
+      "loss": 2.85,
+      "step": 25895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.82305188412091e-05,
+      "loss": 2.9321,
+      "step": 25900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.822275394045813e-05,
+      "loss": 2.9745,
+      "step": 25905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.82149880406484e-05,
+      "loss": 2.9713,
+      "step": 25910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.820722114205483e-05,
+      "loss": 2.8103,
+      "step": 25915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.819945324495235e-05,
+      "loss": 2.8995,
+      "step": 25920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.819168434961594e-05,
+      "loss": 2.8912,
+      "step": 25925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.818391445632062e-05,
+      "loss": 2.9999,
+      "step": 25930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.817614356534144e-05,
+      "loss": 2.943,
+      "step": 25935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.816837167695348e-05,
+      "loss": 2.9032,
+      "step": 25940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.816059879143183e-05,
+      "loss": 2.9505,
+      "step": 25945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.815282490905168e-05,
+      "loss": 2.9433,
+      "step": 25950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.81450500300882e-05,
+      "loss": 2.8805,
+      "step": 25955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.813727415481662e-05,
+      "loss": 2.8879,
+      "step": 25960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.812949728351218e-05,
+      "loss": 3.0653,
+      "step": 25965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.812171941645019e-05,
+      "loss": 2.9285,
+      "step": 25970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.8113940553906e-05,
+      "loss": 3.0618,
+      "step": 25975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.810616069615492e-05,
+      "loss": 2.8433,
+      "step": 25980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.809837984347241e-05,
+      "loss": 2.8916,
+      "step": 25985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.809059799613386e-05,
+      "loss": 2.9179,
+      "step": 25990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.808281515441476e-05,
+      "loss": 2.98,
+      "step": 25995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.807503131859062e-05,
+      "loss": 2.9978,
+      "step": 26000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.806724648893695e-05,
+      "loss": 2.9242,
+      "step": 26005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.805946066572938e-05,
+      "loss": 2.9088,
+      "step": 26010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.805167384924346e-05,
+      "loss": 2.8733,
+      "step": 26015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.804388603975488e-05,
+      "loss": 2.9967,
+      "step": 26020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.803609723753928e-05,
+      "loss": 2.9061,
+      "step": 26025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.802830744287244e-05,
+      "loss": 2.963,
+      "step": 26030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.802051665603005e-05,
+      "loss": 2.8698,
+      "step": 26035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.801272487728793e-05,
+      "loss": 2.7655,
+      "step": 26040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.800493210692187e-05,
+      "loss": 2.8679,
+      "step": 26045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.799713834520776e-05,
+      "loss": 2.9261,
+      "step": 26050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.79893435924215e-05,
+      "loss": 2.9719,
+      "step": 26055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.798154784883898e-05,
+      "loss": 2.8972,
+      "step": 26060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.797375111473617e-05,
+      "loss": 2.9565,
+      "step": 26065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.796595339038907e-05,
+      "loss": 2.8736,
+      "step": 26070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.795815467607372e-05,
+      "loss": 2.9679,
+      "step": 26075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.79503549720662e-05,
+      "loss": 2.8831,
+      "step": 26080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.794255427864259e-05,
+      "loss": 2.9382,
+      "step": 26085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.793475259607905e-05,
+      "loss": 2.9934,
+      "step": 26090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.79269499246517e-05,
+      "loss": 2.9605,
+      "step": 26095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.79191462646368e-05,
+      "loss": 2.9453,
+      "step": 26100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.791134161631059e-05,
+      "loss": 2.9367,
+      "step": 26105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.790353597994932e-05,
+      "loss": 2.8881,
+      "step": 26110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.789572935582931e-05,
+      "loss": 2.967,
+      "step": 26115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.78879217442269e-05,
+      "loss": 2.9,
+      "step": 26120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.788011314541852e-05,
+      "loss": 2.9142,
+      "step": 26125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.787230355968053e-05,
+      "loss": 2.8894,
+      "step": 26130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.786449298728939e-05,
+      "loss": 2.9615,
+      "step": 26135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.785668142852162e-05,
+      "loss": 2.8871,
+      "step": 26140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.784886888365371e-05,
+      "loss": 2.8725,
+      "step": 26145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.784105535296223e-05,
+      "loss": 2.9491,
+      "step": 26150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.783324083672377e-05,
+      "loss": 2.9101,
+      "step": 26155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.782542533521496e-05,
+      "loss": 2.9221,
+      "step": 26160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.781760884871246e-05,
+      "loss": 2.9851,
+      "step": 26165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.780979137749294e-05,
+      "loss": 3.0004,
+      "step": 26170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.780197292183318e-05,
+      "loss": 2.9337,
+      "step": 26175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.779415348200992e-05,
+      "loss": 2.9682,
+      "step": 26180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.778633305829994e-05,
+      "loss": 3.0079,
+      "step": 26185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.777851165098012e-05,
+      "loss": 2.9641,
+      "step": 26190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.777068926032729e-05,
+      "loss": 2.9212,
+      "step": 26195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.776286588661838e-05,
+      "loss": 2.9826,
+      "step": 26200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.775504153013032e-05,
+      "loss": 2.9237,
+      "step": 26205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.774721619114009e-05,
+      "loss": 3.0023,
+      "step": 26210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.77393898699247e-05,
+      "loss": 2.915,
+      "step": 26215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.773156256676118e-05,
+      "loss": 2.8464,
+      "step": 26220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.772373428192663e-05,
+      "loss": 2.9516,
+      "step": 26225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.771590501569816e-05,
+      "loss": 2.9293,
+      "step": 26230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.770807476835291e-05,
+      "loss": 2.8751,
+      "step": 26235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.770024354016806e-05,
+      "loss": 2.9088,
+      "step": 26240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.769241133142083e-05,
+      "loss": 2.8849,
+      "step": 26245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.768457814238849e-05,
+      "loss": 3.02,
+      "step": 26250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.767674397334829e-05,
+      "loss": 2.896,
+      "step": 26255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.766890882457761e-05,
+      "loss": 2.9181,
+      "step": 26260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.766107269635374e-05,
+      "loss": 2.9172,
+      "step": 26265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.765323558895413e-05,
+      "loss": 2.8569,
+      "step": 26270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.764539750265618e-05,
+      "loss": 2.8627,
+      "step": 26275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.763755843773734e-05,
+      "loss": 3.0347,
+      "step": 26280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.762971839447513e-05,
+      "loss": 2.9147,
+      "step": 26285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.762187737314706e-05,
+      "loss": 2.9459,
+      "step": 26290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.76140353740307e-05,
+      "loss": 2.9416,
+      "step": 26295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.760619239740365e-05,
+      "loss": 2.8994,
+      "step": 26300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.759834844354356e-05,
+      "loss": 2.9007,
+      "step": 26305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.759050351272807e-05,
+      "loss": 2.8987,
+      "step": 26310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.758265760523489e-05,
+      "loss": 2.9092,
+      "step": 26315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.757481072134178e-05,
+      "loss": 2.9854,
+      "step": 26320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.75669628613265e-05,
+      "loss": 2.9127,
+      "step": 26325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.755911402546685e-05,
+      "loss": 2.9622,
+      "step": 26330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.755126421404066e-05,
+      "loss": 2.9286,
+      "step": 26335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.754341342732583e-05,
+      "loss": 2.9085,
+      "step": 26340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.753556166560028e-05,
+      "loss": 2.9601,
+      "step": 26345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.752770892914192e-05,
+      "loss": 2.8013,
+      "step": 26350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.751985521822876e-05,
+      "loss": 2.976,
+      "step": 26355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.751200053313877e-05,
+      "loss": 2.9617,
+      "step": 26360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.750414487415004e-05,
+      "loss": 2.9827,
+      "step": 26365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.749628824154065e-05,
+      "loss": 2.9978,
+      "step": 26370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.74884306355887e-05,
+      "loss": 2.8917,
+      "step": 26375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.748057205657237e-05,
+      "loss": 2.9146,
+      "step": 26380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.74727125047698e-05,
+      "loss": 2.9411,
+      "step": 26385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.746485198045925e-05,
+      "loss": 2.892,
+      "step": 26390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.745699048391896e-05,
+      "loss": 3.0241,
+      "step": 26395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.744912801542723e-05,
+      "loss": 2.8423,
+      "step": 26400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.744126457526237e-05,
+      "loss": 3.0443,
+      "step": 26405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.743340016370275e-05,
+      "loss": 2.8894,
+      "step": 26410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.742553478102674e-05,
+      "loss": 2.8031,
+      "step": 26415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.741766842751281e-05,
+      "loss": 2.9663,
+      "step": 26420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.740980110343937e-05,
+      "loss": 2.9661,
+      "step": 26425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.740193280908497e-05,
+      "loss": 2.9599,
+      "step": 26430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.739406354472808e-05,
+      "loss": 2.8672,
+      "step": 26435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.738619331064731e-05,
+      "loss": 2.9207,
+      "step": 26440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.737832210712125e-05,
+      "loss": 2.9555,
+      "step": 26445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.737044993442853e-05,
+      "loss": 2.9749,
+      "step": 26450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.73625767928478e-05,
+      "loss": 2.88,
+      "step": 26455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.735470268265779e-05,
+      "loss": 2.9812,
+      "step": 26460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.73468276041372e-05,
+      "loss": 2.9805,
+      "step": 26465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.733895155756485e-05,
+      "loss": 2.8849,
+      "step": 26470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.73310745432195e-05,
+      "loss": 2.8798,
+      "step": 26475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.732319656138002e-05,
+      "loss": 2.9786,
+      "step": 26480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.731531761232528e-05,
+      "loss": 2.9494,
+      "step": 26485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.730743769633416e-05,
+      "loss": 2.9305,
+      "step": 26490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.729955681368562e-05,
+      "loss": 2.9057,
+      "step": 26495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.729167496465863e-05,
+      "loss": 2.8565,
+      "step": 26500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.728379214953223e-05,
+      "loss": 2.8975,
+      "step": 26505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.727590836858541e-05,
+      "loss": 2.867,
+      "step": 26510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.726802362209729e-05,
+      "loss": 2.9101,
+      "step": 26515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.726013791034696e-05,
+      "loss": 2.8834,
+      "step": 26520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.72522512336136e-05,
+      "loss": 2.9493,
+      "step": 26525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.724436359217634e-05,
+      "loss": 2.8698,
+      "step": 26530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.723647498631443e-05,
+      "loss": 2.9436,
+      "step": 26535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.722858541630713e-05,
+      "loss": 2.9042,
+      "step": 26540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.722069488243368e-05,
+      "loss": 2.9163,
+      "step": 26545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.721280338497345e-05,
+      "loss": 2.921,
+      "step": 26550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.720491092420574e-05,
+      "loss": 2.9859,
+      "step": 26555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.719701750040996e-05,
+      "loss": 2.9333,
+      "step": 26560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.718912311386552e-05,
+      "loss": 2.8914,
+      "step": 26565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.718122776485192e-05,
+      "loss": 2.9839,
+      "step": 26570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.717333145364857e-05,
+      "loss": 2.9103,
+      "step": 26575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.716543418053507e-05,
+      "loss": 2.9822,
+      "step": 26580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.715753594579091e-05,
+      "loss": 2.9986,
+      "step": 26585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.714963674969573e-05,
+      "loss": 2.961,
+      "step": 26590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.714173659252912e-05,
+      "loss": 2.9673,
+      "step": 26595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.713383547457076e-05,
+      "loss": 2.8746,
+      "step": 26600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.712593339610032e-05,
+      "loss": 3.0275,
+      "step": 26605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.711803035739755e-05,
+      "loss": 2.8419,
+      "step": 26610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.71101263587422e-05,
+      "loss": 2.8986,
+      "step": 26615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.710222140041406e-05,
+      "loss": 2.8938,
+      "step": 26620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.709431548269295e-05,
+      "loss": 3.0136,
+      "step": 26625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.708640860585876e-05,
+      "loss": 2.9576,
+      "step": 26630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.707850077019137e-05,
+      "loss": 2.8982,
+      "step": 26635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.707059197597069e-05,
+      "loss": 2.9884,
+      "step": 26640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.706268222347673e-05,
+      "loss": 2.8998,
+      "step": 26645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.705477151298944e-05,
+      "loss": 2.8748,
+      "step": 26650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.704685984478888e-05,
+      "loss": 2.9658,
+      "step": 26655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.70389472191551e-05,
+      "loss": 2.9366,
+      "step": 26660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.703103363636821e-05,
+      "loss": 2.9848,
+      "step": 26665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.702311909670834e-05,
+      "loss": 2.9454,
+      "step": 26670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.701520360045566e-05,
+      "loss": 2.8806,
+      "step": 26675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.700728714789039e-05,
+      "loss": 3.0198,
+      "step": 26680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.699936973929272e-05,
+      "loss": 2.9343,
+      "step": 26685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.699145137494295e-05,
+      "loss": 2.8932,
+      "step": 26690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.698353205512136e-05,
+      "loss": 2.95,
+      "step": 26695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.697561178010833e-05,
+      "loss": 2.9171,
+      "step": 26700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.696769055018418e-05,
+      "loss": 2.8649,
+      "step": 26705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.695976836562935e-05,
+      "loss": 2.8336,
+      "step": 26710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.695184522672425e-05,
+      "loss": 2.9057,
+      "step": 26715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.694392113374938e-05,
+      "loss": 2.9244,
+      "step": 26720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.693599608698521e-05,
+      "loss": 2.9712,
+      "step": 26725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.692807008671232e-05,
+      "loss": 2.9303,
+      "step": 26730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.692014313321126e-05,
+      "loss": 2.9651,
+      "step": 26735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.691221522676264e-05,
+      "loss": 2.9721,
+      "step": 26740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.690428636764708e-05,
+      "loss": 2.9371,
+      "step": 26745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.689635655614532e-05,
+      "loss": 2.9686,
+      "step": 26750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.688842579253796e-05,
+      "loss": 2.9163,
+      "step": 26755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.688049407710584e-05,
+      "loss": 2.889,
+      "step": 26760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.687256141012968e-05,
+      "loss": 2.9345,
+      "step": 26765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.686462779189035e-05,
+      "loss": 2.8913,
+      "step": 26770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.685669322266861e-05,
+      "loss": 2.9987,
+      "step": 26775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.684875770274539e-05,
+      "loss": 2.9818,
+      "step": 26780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.684082123240157e-05,
+      "loss": 3.0017,
+      "step": 26785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.683288381191813e-05,
+      "loss": 2.9645,
+      "step": 26790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.682494544157603e-05,
+      "loss": 2.9254,
+      "step": 26795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.681700612165625e-05,
+      "loss": 2.9012,
+      "step": 26800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.68090658524399e-05,
+      "loss": 3.0296,
+      "step": 26805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.680112463420798e-05,
+      "loss": 2.9336,
+      "step": 26810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.679318246724167e-05,
+      "loss": 2.9242,
+      "step": 26815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.678523935182207e-05,
+      "loss": 2.9232,
+      "step": 26820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.677729528823038e-05,
+      "loss": 2.8062,
+      "step": 26825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.67693502767478e-05,
+      "loss": 2.8845,
+      "step": 26830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.67614043176556e-05,
+      "loss": 2.9503,
+      "step": 26835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.675345741123506e-05,
+      "loss": 2.8775,
+      "step": 26840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.674550955776745e-05,
+      "loss": 3.0456,
+      "step": 26845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.673756075753414e-05,
+      "loss": 2.9548,
+      "step": 26850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.672961101081652e-05,
+      "loss": 2.9622,
+      "step": 26855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.672166031789598e-05,
+      "loss": 2.997,
+      "step": 26860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.671370867905401e-05,
+      "loss": 2.9134,
+      "step": 26865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.670575609457205e-05,
+      "loss": 2.8792,
+      "step": 26870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.669780256473162e-05,
+      "loss": 2.9899,
+      "step": 26875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.668984808981429e-05,
+      "loss": 2.9456,
+      "step": 26880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.668189267010162e-05,
+      "loss": 2.8784,
+      "step": 26885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.667393630587523e-05,
+      "loss": 3.0027,
+      "step": 26890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.666597899741677e-05,
+      "loss": 2.927,
+      "step": 26895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.665802074500792e-05,
+      "loss": 2.9473,
+      "step": 26900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.665006154893039e-05,
+      "loss": 2.8907,
+      "step": 26905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.664210140946592e-05,
+      "loss": 2.8594,
+      "step": 26910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.663414032689631e-05,
+      "loss": 2.9425,
+      "step": 26915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.662617830150338e-05,
+      "loss": 2.9279,
+      "step": 26920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.661821533356896e-05,
+      "loss": 2.9417,
+      "step": 26925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.661025142337493e-05,
+      "loss": 2.9423,
+      "step": 26930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.660228657120323e-05,
+      "loss": 2.868,
+      "step": 26935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.659432077733577e-05,
+      "loss": 2.9443,
+      "step": 26940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.658635404205457e-05,
+      "loss": 2.9921,
+      "step": 26945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.657838636564163e-05,
+      "loss": 2.8132,
+      "step": 26950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.657041774837897e-05,
+      "loss": 2.9891,
+      "step": 26955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.656244819054872e-05,
+      "loss": 2.85,
+      "step": 26960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.655447769243296e-05,
+      "loss": 2.9466,
+      "step": 26965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.654650625431386e-05,
+      "loss": 2.9479,
+      "step": 26970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.653853387647358e-05,
+      "loss": 2.9641,
+      "step": 26975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.653056055919435e-05,
+      "loss": 2.8755,
+      "step": 26980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.65225863027584e-05,
+      "loss": 2.9265,
+      "step": 26985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.651461110744804e-05,
+      "loss": 2.8272,
+      "step": 26990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.650663497354557e-05,
+      "loss": 2.9264,
+      "step": 26995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.649865790133333e-05,
+      "loss": 2.9071,
+      "step": 27000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.649067989109371e-05,
+      "loss": 2.9356,
+      "step": 27005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.648270094310911e-05,
+      "loss": 2.7867,
+      "step": 27010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.647472105766201e-05,
+      "loss": 2.9874,
+      "step": 27015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.646674023503486e-05,
+      "loss": 2.9058,
+      "step": 27020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.645875847551017e-05,
+      "loss": 2.9769,
+      "step": 27025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.645077577937051e-05,
+      "loss": 2.9211,
+      "step": 27030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.644279214689846e-05,
+      "loss": 2.9231,
+      "step": 27035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.643480757837662e-05,
+      "loss": 2.9884,
+      "step": 27040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.642682207408763e-05,
+      "loss": 2.9125,
+      "step": 27045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.641883563431419e-05,
+      "loss": 3.0053,
+      "step": 27050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.641084825933896e-05,
+      "loss": 2.9688,
+      "step": 27055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.640285994944476e-05,
+      "loss": 2.9238,
+      "step": 27060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.639487070491432e-05,
+      "loss": 2.9061,
+      "step": 27065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.638688052603046e-05,
+      "loss": 2.9433,
+      "step": 27070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.637888941307606e-05,
+      "loss": 2.9739,
+      "step": 27075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.637089736633395e-05,
+      "loss": 2.9279,
+      "step": 27080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.636290438608707e-05,
+      "loss": 2.9081,
+      "step": 27085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.635491047261834e-05,
+      "loss": 2.8702,
+      "step": 27090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.634691562621075e-05,
+      "loss": 2.9552,
+      "step": 27095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.633891984714732e-05,
+      "loss": 2.8942,
+      "step": 27100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.633092313571106e-05,
+      "loss": 2.9011,
+      "step": 27105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.632292549218509e-05,
+      "loss": 2.9142,
+      "step": 27110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.63149269168525e-05,
+      "loss": 2.9295,
+      "step": 27115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.630692740999643e-05,
+      "loss": 2.8835,
+      "step": 27120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.629892697190004e-05,
+      "loss": 2.8974,
+      "step": 27125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.629092560284657e-05,
+      "loss": 2.8923,
+      "step": 27130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.628292330311923e-05,
+      "loss": 2.9352,
+      "step": 27135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.627492007300132e-05,
+      "loss": 2.8874,
+      "step": 27140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.626691591277614e-05,
+      "loss": 3.0226,
+      "step": 27145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.6258910822727e-05,
+      "loss": 2.8417,
+      "step": 27150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.625090480313732e-05,
+      "loss": 2.9399,
+      "step": 27155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.624289785429046e-05,
+      "loss": 2.9797,
+      "step": 27160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.623488997646991e-05,
+      "loss": 2.9855,
+      "step": 27165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.622688116995909e-05,
+      "loss": 2.8919,
+      "step": 27170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.621887143504151e-05,
+      "loss": 2.9389,
+      "step": 27175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.621086077200077e-05,
+      "loss": 2.8904,
+      "step": 27180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.620284918112035e-05,
+      "loss": 2.9063,
+      "step": 27185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.619483666268393e-05,
+      "loss": 2.9914,
+      "step": 27190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.618682321697509e-05,
+      "loss": 2.96,
+      "step": 27195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.617880884427753e-05,
+      "loss": 2.8958,
+      "step": 27200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.617079354487494e-05,
+      "loss": 2.9419,
+      "step": 27205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.616277731905103e-05,
+      "loss": 2.9866,
+      "step": 27210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.615476016708962e-05,
+      "loss": 2.8958,
+      "step": 27215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.614674208927446e-05,
+      "loss": 2.9298,
+      "step": 27220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.613872308588943e-05,
+      "loss": 3.0356,
+      "step": 27225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.613070315721836e-05,
+      "loss": 2.8554,
+      "step": 27230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.612268230354514e-05,
+      "loss": 2.9503,
+      "step": 27235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.611466052515373e-05,
+      "loss": 2.9556,
+      "step": 27240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.610663782232808e-05,
+      "loss": 2.8352,
+      "step": 27245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.60986141953522e-05,
+      "loss": 2.991,
+      "step": 27250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.609058964451008e-05,
+      "loss": 2.9801,
+      "step": 27255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.608256417008582e-05,
+      "loss": 2.8979,
+      "step": 27260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.60745377723635e-05,
+      "loss": 2.83,
+      "step": 27265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.606651045162724e-05,
+      "loss": 2.867,
+      "step": 27270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.605848220816123e-05,
+      "loss": 2.8357,
+      "step": 27275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.605045304224962e-05,
+      "loss": 2.8731,
+      "step": 27280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.604242295417667e-05,
+      "loss": 2.9073,
+      "step": 27285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.603439194422661e-05,
+      "loss": 2.9238,
+      "step": 27290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.602636001268373e-05,
+      "loss": 3.0026,
+      "step": 27295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.601832715983239e-05,
+      "loss": 3.0085,
+      "step": 27300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.601029338595691e-05,
+      "loss": 3.0089,
+      "step": 27305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.600225869134169e-05,
+      "loss": 3.0267,
+      "step": 27310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.599422307627114e-05,
+      "loss": 2.9414,
+      "step": 27315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.598618654102973e-05,
+      "loss": 3.0118,
+      "step": 27320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.597814908590195e-05,
+      "loss": 2.9754,
+      "step": 27325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.59701107111723e-05,
+      "loss": 2.8011,
+      "step": 27330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.596207141712532e-05,
+      "loss": 2.9684,
+      "step": 27335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.595403120404563e-05,
+      "loss": 2.9284,
+      "step": 27340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.594599007221782e-05,
+      "loss": 2.909,
+      "step": 27345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.593794802192655e-05,
+      "loss": 2.8854,
+      "step": 27350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.592990505345648e-05,
+      "loss": 2.8744,
+      "step": 27355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.592186116709234e-05,
+      "loss": 2.9523,
+      "step": 27360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.59138163631189e-05,
+      "loss": 2.8822,
+      "step": 27365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.590577064182089e-05,
+      "loss": 2.9212,
+      "step": 27370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.589772400348315e-05,
+      "loss": 2.9253,
+      "step": 27375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.588967644839053e-05,
+      "loss": 2.9122,
+      "step": 27380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.588162797682787e-05,
+      "loss": 2.9708,
+      "step": 27385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.587357858908013e-05,
+      "loss": 2.8438,
+      "step": 27390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.586552828543221e-05,
+      "loss": 2.9596,
+      "step": 27395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.585747706616908e-05,
+      "loss": 2.8793,
+      "step": 27400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.584942493157579e-05,
+      "loss": 2.9568,
+      "step": 27405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.584137188193732e-05,
+      "loss": 3.0055,
+      "step": 27410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.583331791753879e-05,
+      "loss": 2.9123,
+      "step": 27415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.582526303866528e-05,
+      "loss": 3.0124,
+      "step": 27420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.581720724560193e-05,
+      "loss": 2.8679,
+      "step": 27425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.580915053863389e-05,
+      "loss": 2.9208,
+      "step": 27430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.58010929180464e-05,
+      "loss": 2.923,
+      "step": 27435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.579303438412464e-05,
+      "loss": 2.8333,
+      "step": 27440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.578497493715391e-05,
+      "loss": 2.8906,
+      "step": 27445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.577691457741951e-05,
+      "loss": 2.9187,
+      "step": 27450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.576885330520674e-05,
+      "loss": 2.8749,
+      "step": 27455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.576079112080098e-05,
+      "loss": 2.9677,
+      "step": 27460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.575272802448763e-05,
+      "loss": 2.8248,
+      "step": 27465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.574466401655211e-05,
+      "loss": 3.04,
+      "step": 27470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.573659909727988e-05,
+      "loss": 3.0516,
+      "step": 27475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.572853326695641e-05,
+      "loss": 2.8254,
+      "step": 27480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.572046652586727e-05,
+      "loss": 2.942,
+      "step": 27485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.571239887429795e-05,
+      "loss": 2.9086,
+      "step": 27490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.57043303125341e-05,
+      "loss": 2.9401,
+      "step": 27495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.569626084086128e-05,
+      "loss": 2.8421,
+      "step": 27500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.56881904595652e-05,
+      "loss": 2.9259,
+      "step": 27505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.568011916893151e-05,
+      "loss": 2.9725,
+      "step": 27510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.567204696924592e-05,
+      "loss": 2.8217,
+      "step": 27515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.566397386079421e-05,
+      "loss": 2.9256,
+      "step": 27520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.565589984386214e-05,
+      "loss": 2.8855,
+      "step": 27525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.564782491873552e-05,
+      "loss": 2.9292,
+      "step": 27530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.563974908570019e-05,
+      "loss": 3.0303,
+      "step": 27535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.563167234504204e-05,
+      "loss": 2.8766,
+      "step": 27540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.562359469704698e-05,
+      "loss": 2.7988,
+      "step": 27545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.561551614200094e-05,
+      "loss": 2.8581,
+      "step": 27550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.56074366801899e-05,
+      "loss": 2.8408,
+      "step": 27555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.559935631189986e-05,
+      "loss": 2.909,
+      "step": 27560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.559127503741687e-05,
+      "loss": 2.9211,
+      "step": 27565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.558319285702699e-05,
+      "loss": 2.9013,
+      "step": 27570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.557510977101633e-05,
+      "loss": 2.9161,
+      "step": 27575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5567025779671e-05,
+      "loss": 2.8916,
+      "step": 27580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.555894088327718e-05,
+      "loss": 2.9408,
+      "step": 27585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.55508550821211e-05,
+      "loss": 3.021,
+      "step": 27590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.554276837648893e-05,
+      "loss": 2.9664,
+      "step": 27595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.553468076666697e-05,
+      "loss": 2.882,
+      "step": 27600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.55265922529415e-05,
+      "loss": 2.87,
+      "step": 27605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.551850283559885e-05,
+      "loss": 2.8782,
+      "step": 27610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.551041251492538e-05,
+      "loss": 2.7754,
+      "step": 27615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.550232129120746e-05,
+      "loss": 2.8997,
+      "step": 27620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.549422916473156e-05,
+      "loss": 2.9424,
+      "step": 27625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.548613613578408e-05,
+      "loss": 2.9394,
+      "step": 27630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.547804220465156e-05,
+      "loss": 2.9173,
+      "step": 27635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.546994737162046e-05,
+      "loss": 3.062,
+      "step": 27640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.546185163697737e-05,
+      "loss": 2.96,
+      "step": 27645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.545375500100884e-05,
+      "loss": 2.804,
+      "step": 27650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.54456574640015e-05,
+      "loss": 2.9691,
+      "step": 27655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.543755902624201e-05,
+      "loss": 2.8669,
+      "step": 27660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.542945968801702e-05,
+      "loss": 2.9166,
+      "step": 27665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.542135944961326e-05,
+      "loss": 2.9406,
+      "step": 27670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.541325831131745e-05,
+      "loss": 2.9691,
+      "step": 27675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.540515627341638e-05,
+      "loss": 2.8702,
+      "step": 27680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.539705333619687e-05,
+      "loss": 3.0168,
+      "step": 27685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.53889494999457e-05,
+      "loss": 3.0062,
+      "step": 27690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.538084476494979e-05,
+      "loss": 2.844,
+      "step": 27695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.537273913149602e-05,
+      "loss": 2.9245,
+      "step": 27700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.536463259987134e-05,
+      "loss": 2.8706,
+      "step": 27705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.535652517036268e-05,
+      "loss": 2.9321,
+      "step": 27710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.534841684325707e-05,
+      "loss": 2.9595,
+      "step": 27715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.53403076188415e-05,
+      "loss": 2.8001,
+      "step": 27720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.533219749740305e-05,
+      "loss": 2.7679,
+      "step": 27725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.532408647922883e-05,
+      "loss": 2.8885,
+      "step": 27730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.531597456460594e-05,
+      "loss": 2.9955,
+      "step": 27735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.530786175382154e-05,
+      "loss": 2.9305,
+      "step": 27740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.52997480471628e-05,
+      "loss": 2.8631,
+      "step": 27745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.529163344491694e-05,
+      "loss": 3.0329,
+      "step": 27750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.528351794737124e-05,
+      "loss": 2.944,
+      "step": 27755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.527540155481295e-05,
+      "loss": 2.8842,
+      "step": 27760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.52672842675294e-05,
+      "loss": 3.0461,
+      "step": 27765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.525916608580791e-05,
+      "loss": 2.894,
+      "step": 27770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.52510470099359e-05,
+      "loss": 2.9383,
+      "step": 27775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.524292704020074e-05,
+      "loss": 2.9055,
+      "step": 27780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.523480617688987e-05,
+      "loss": 3.0093,
+      "step": 27785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.522668442029079e-05,
+      "loss": 2.9301,
+      "step": 27790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.521856177069097e-05,
+      "loss": 3.0503,
+      "step": 27795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.521043822837796e-05,
+      "loss": 2.8971,
+      "step": 27800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.520231379363932e-05,
+      "loss": 2.8854,
+      "step": 27805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.519418846676266e-05,
+      "loss": 2.9436,
+      "step": 27810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.518606224803559e-05,
+      "loss": 2.9976,
+      "step": 27815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.517793513774579e-05,
+      "loss": 2.9418,
+      "step": 27820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.516980713618094e-05,
+      "loss": 2.9578,
+      "step": 27825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.516167824362875e-05,
+      "loss": 2.8257,
+      "step": 27830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.515354846037701e-05,
+      "loss": 2.8967,
+      "step": 27835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.514541778671347e-05,
+      "loss": 2.9269,
+      "step": 27840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.513728622292597e-05,
+      "loss": 2.8209,
+      "step": 27845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.512915376930236e-05,
+      "loss": 2.9585,
+      "step": 27850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.512102042613051e-05,
+      "loss": 2.9429,
+      "step": 27855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.511288619369833e-05,
+      "loss": 2.9129,
+      "step": 27860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.51047510722938e-05,
+      "loss": 2.9695,
+      "step": 27865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.509661506220486e-05,
+      "loss": 2.9822,
+      "step": 27870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.508847816371951e-05,
+      "loss": 3.0067,
+      "step": 27875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.508034037712582e-05,
+      "loss": 2.9127,
+      "step": 27880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.507220170271183e-05,
+      "loss": 2.9606,
+      "step": 27885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.506406214076566e-05,
+      "loss": 2.9213,
+      "step": 27890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.505592169157544e-05,
+      "loss": 2.9377,
+      "step": 27895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.504778035542934e-05,
+      "loss": 2.9031,
+      "step": 27900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.503963813261554e-05,
+      "loss": 2.8823,
+      "step": 27905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.503149502342227e-05,
+      "loss": 2.8413,
+      "step": 27910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.50233510281378e-05,
+      "loss": 2.9614,
+      "step": 27915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.501520614705042e-05,
+      "loss": 2.9309,
+      "step": 27920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.500706038044843e-05,
+      "loss": 2.8854,
+      "step": 27925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.499891372862021e-05,
+      "loss": 2.7949,
+      "step": 27930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.499076619185412e-05,
+      "loss": 2.9237,
+      "step": 27935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.498261777043859e-05,
+      "loss": 2.8996,
+      "step": 27940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.497446846466205e-05,
+      "loss": 2.9539,
+      "step": 27945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.4966318274813e-05,
+      "loss": 2.8898,
+      "step": 27950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.495816720117993e-05,
+      "loss": 2.943,
+      "step": 27955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.49500152440514e-05,
+      "loss": 3.0007,
+      "step": 27960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.494186240371597e-05,
+      "loss": 2.8902,
+      "step": 27965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.493370868046222e-05,
+      "loss": 2.9594,
+      "step": 27970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.492555407457884e-05,
+      "loss": 2.8327,
+      "step": 27975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.491739858635443e-05,
+      "loss": 2.9278,
+      "step": 27980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.490924221607776e-05,
+      "loss": 2.9091,
+      "step": 27985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.490108496403748e-05,
+      "loss": 2.9816,
+      "step": 27990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.489292683052239e-05,
+      "loss": 2.9441,
+      "step": 27995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.488476781582128e-05,
+      "loss": 2.9365,
+      "step": 28000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.487660792022297e-05,
+      "loss": 2.8829,
+      "step": 28005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.486844714401631e-05,
+      "loss": 2.9388,
+      "step": 28010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.486028548749019e-05,
+      "loss": 2.83,
+      "step": 28015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.485212295093353e-05,
+      "loss": 2.9706,
+      "step": 28020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.484395953463523e-05,
+      "loss": 3.0136,
+      "step": 28025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.483579523888432e-05,
+      "loss": 2.8873,
+      "step": 28030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.48276300639698e-05,
+      "loss": 3.0139,
+      "step": 28035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.481946401018068e-05,
+      "loss": 2.8255,
+      "step": 28040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.481129707780605e-05,
+      "loss": 2.9219,
+      "step": 28045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.480312926713502e-05,
+      "loss": 2.9725,
+      "step": 28050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.479496057845673e-05,
+      "loss": 2.9573,
+      "step": 28055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.478679101206031e-05,
+      "loss": 2.9326,
+      "step": 28060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.477862056823496e-05,
+      "loss": 3.037,
+      "step": 28065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.477044924726995e-05,
+      "loss": 3.0248,
+      "step": 28070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.476227704945449e-05,
+      "loss": 2.9389,
+      "step": 28075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.475410397507789e-05,
+      "loss": 2.9297,
+      "step": 28080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.474593002442946e-05,
+      "loss": 2.9612,
+      "step": 28085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.473775519779857e-05,
+      "loss": 2.9363,
+      "step": 28090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.472957949547454e-05,
+      "loss": 2.9064,
+      "step": 28095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.472140291774687e-05,
+      "loss": 2.9177,
+      "step": 28100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.471322546490494e-05,
+      "loss": 2.946,
+      "step": 28105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.470504713723824e-05,
+      "loss": 3.0095,
+      "step": 28110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.469686793503629e-05,
+      "loss": 2.91,
+      "step": 28115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.468868785858862e-05,
+      "loss": 2.9781,
+      "step": 28120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.468050690818479e-05,
+      "loss": 2.923,
+      "step": 28125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.467232508411439e-05,
+      "loss": 2.9755,
+      "step": 28130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.466414238666706e-05,
+      "loss": 2.8735,
+      "step": 28135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.465595881613247e-05,
+      "loss": 2.9386,
+      "step": 28140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.464777437280028e-05,
+      "loss": 2.949,
+      "step": 28145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.463958905696025e-05,
+      "loss": 3.0195,
+      "step": 28150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.46314028689021e-05,
+      "loss": 2.8568,
+      "step": 28155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.462321580891563e-05,
+      "loss": 2.9082,
+      "step": 28160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.461502787729065e-05,
+      "loss": 2.9077,
+      "step": 28165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.460683907431701e-05,
+      "loss": 2.9603,
+      "step": 28170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.459864940028458e-05,
+      "loss": 2.9181,
+      "step": 28175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.459045885548326e-05,
+      "loss": 2.8971,
+      "step": 28180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4582267440203e-05,
+      "loss": 2.8266,
+      "step": 28185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.457407515473376e-05,
+      "loss": 2.8676,
+      "step": 28190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.456588199936555e-05,
+      "loss": 2.9873,
+      "step": 28195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.45576879743884e-05,
+      "loss": 2.8211,
+      "step": 28200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.454949308009236e-05,
+      "loss": 2.9328,
+      "step": 28205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.454129731676751e-05,
+      "loss": 2.8129,
+      "step": 28210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4533100684704e-05,
+      "loss": 3.0176,
+      "step": 28215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.452490318419197e-05,
+      "loss": 2.9742,
+      "step": 28220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.45167048155216e-05,
+      "loss": 2.879,
+      "step": 28225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.450850557898311e-05,
+      "loss": 2.8455,
+      "step": 28230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.450030547486675e-05,
+      "loss": 2.9098,
+      "step": 28235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.449210450346277e-05,
+      "loss": 3.0313,
+      "step": 28240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.448390266506151e-05,
+      "loss": 2.8539,
+      "step": 28245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.447569995995329e-05,
+      "loss": 2.8726,
+      "step": 28250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.446749638842851e-05,
+      "loss": 2.9365,
+      "step": 28255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.44592919507775e-05,
+      "loss": 2.8525,
+      "step": 28260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.445108664729075e-05,
+      "loss": 2.7996,
+      "step": 28265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.44428804782587e-05,
+      "loss": 2.8893,
+      "step": 28270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.443467344397184e-05,
+      "loss": 2.9441,
+      "step": 28275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.442646554472069e-05,
+      "loss": 2.924,
+      "step": 28280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.441825678079581e-05,
+      "loss": 2.8862,
+      "step": 28285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.441004715248777e-05,
+      "loss": 2.9343,
+      "step": 28290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.440183666008718e-05,
+      "loss": 2.8672,
+      "step": 28295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.43936253038847e-05,
+      "loss": 2.9824,
+      "step": 28300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4385413084171e-05,
+      "loss": 2.9701,
+      "step": 28305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.437720000123676e-05,
+      "loss": 2.9021,
+      "step": 28310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.436898605537277e-05,
+      "loss": 2.884,
+      "step": 28315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.436077124686974e-05,
+      "loss": 2.872,
+      "step": 28320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.43525555760185e-05,
+      "loss": 2.9413,
+      "step": 28325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.434433904310986e-05,
+      "loss": 2.9795,
+      "step": 28330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.433612164843468e-05,
+      "loss": 2.8868,
+      "step": 28335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.432790339228385e-05,
+      "loss": 3.0209,
+      "step": 28340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.431968427494833e-05,
+      "loss": 2.9316,
+      "step": 28345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.431146429671898e-05,
+      "loss": 2.9062,
+      "step": 28350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.430324345788685e-05,
+      "loss": 2.8889,
+      "step": 28355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.429502175874293e-05,
+      "loss": 2.8908,
+      "step": 28360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.428679919957825e-05,
+      "loss": 2.9595,
+      "step": 28365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.42785757806839e-05,
+      "loss": 2.9824,
+      "step": 28370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.427035150235097e-05,
+      "loss": 2.8679,
+      "step": 28375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.426212636487058e-05,
+      "loss": 2.911,
+      "step": 28380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.425390036853392e-05,
+      "loss": 3.0341,
+      "step": 28385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.424567351363215e-05,
+      "loss": 2.8654,
+      "step": 28390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.423744580045651e-05,
+      "loss": 2.9037,
+      "step": 28395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.422921722929827e-05,
+      "loss": 2.9799,
+      "step": 28400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.422098780044869e-05,
+      "loss": 3.0163,
+      "step": 28405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.42127575141991e-05,
+      "loss": 2.8466,
+      "step": 28410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.42045263708408e-05,
+      "loss": 2.9271,
+      "step": 28415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.419629437066525e-05,
+      "loss": 2.9882,
+      "step": 28420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.418806151396377e-05,
+      "loss": 2.958,
+      "step": 28425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.417982780102785e-05,
+      "loss": 2.8691,
+      "step": 28430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.417159323214892e-05,
+      "loss": 2.8335,
+      "step": 28435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.41633578076185e-05,
+      "loss": 2.9498,
+      "step": 28440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.41551215277281e-05,
+      "loss": 2.9454,
+      "step": 28445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.41468843927693e-05,
+      "loss": 2.9625,
+      "step": 28450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.413864640303365e-05,
+      "loss": 2.9815,
+      "step": 28455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.41304075588128e-05,
+      "loss": 2.8747,
+      "step": 28460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.41221678603984e-05,
+      "loss": 3.0065,
+      "step": 28465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.411392730808209e-05,
+      "loss": 2.75,
+      "step": 28470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.410568590215563e-05,
+      "loss": 2.9409,
+      "step": 28475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.409744364291069e-05,
+      "loss": 2.9645,
+      "step": 28480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.408920053063909e-05,
+      "loss": 2.9542,
+      "step": 28485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.408095656563262e-05,
+      "loss": 2.9366,
+      "step": 28490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.40727117481831e-05,
+      "loss": 2.9023,
+      "step": 28495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.40644660785824e-05,
+      "loss": 2.8611,
+      "step": 28500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.405621955712241e-05,
+      "loss": 2.8626,
+      "step": 28505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.404797218409503e-05,
+      "loss": 2.9702,
+      "step": 28510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.403972395979222e-05,
+      "loss": 2.8793,
+      "step": 28515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.403147488450594e-05,
+      "loss": 2.9816,
+      "step": 28520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.402322495852825e-05,
+      "loss": 3.0229,
+      "step": 28525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.401497418215114e-05,
+      "loss": 3.0669,
+      "step": 28530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.400672255566672e-05,
+      "loss": 2.9572,
+      "step": 28535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.399847007936704e-05,
+      "loss": 2.9984,
+      "step": 28540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.399021675354427e-05,
+      "loss": 2.9457,
+      "step": 28545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.398196257849054e-05,
+      "loss": 2.8503,
+      "step": 28550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.397370755449807e-05,
+      "loss": 2.938,
+      "step": 28555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.396545168185908e-05,
+      "loss": 2.9272,
+      "step": 28560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.39571949608658e-05,
+      "loss": 2.8281,
+      "step": 28565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.394893739181051e-05,
+      "loss": 2.9124,
+      "step": 28570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.394067897498553e-05,
+      "loss": 2.9402,
+      "step": 28575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.39324197106832e-05,
+      "loss": 2.8289,
+      "step": 28580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.39241595991959e-05,
+      "loss": 2.9127,
+      "step": 28585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.391589864081599e-05,
+      "loss": 2.8165,
+      "step": 28590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.390763683583595e-05,
+      "loss": 2.9273,
+      "step": 28595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.38993741845482e-05,
+      "loss": 2.9389,
+      "step": 28600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.389111068724529e-05,
+      "loss": 2.8991,
+      "step": 28605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.388284634421965e-05,
+      "loss": 2.9849,
+      "step": 28610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.38745811557639e-05,
+      "loss": 2.8308,
+      "step": 28615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.386631512217059e-05,
+      "loss": 2.884,
+      "step": 28620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.385804824373234e-05,
+      "loss": 2.9518,
+      "step": 28625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.384978052074179e-05,
+      "loss": 2.8907,
+      "step": 28630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.38415119534916e-05,
+      "loss": 2.9615,
+      "step": 28635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.383324254227448e-05,
+      "loss": 3.0386,
+      "step": 28640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.382497228738316e-05,
+      "loss": 2.983,
+      "step": 28645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.38167011891104e-05,
+      "loss": 2.9205,
+      "step": 28650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.380842924774897e-05,
+      "loss": 2.8622,
+      "step": 28655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.380015646359171e-05,
+      "loss": 2.909,
+      "step": 28660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.379188283693146e-05,
+      "loss": 3.0164,
+      "step": 28665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.37836083680611e-05,
+      "loss": 2.9169,
+      "step": 28670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.377533305727354e-05,
+      "loss": 2.8858,
+      "step": 28675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.376705690486172e-05,
+      "loss": 2.8852,
+      "step": 28680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.375877991111859e-05,
+      "loss": 2.8966,
+      "step": 28685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.375050207633717e-05,
+      "loss": 2.925,
+      "step": 28690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.374222340081048e-05,
+      "loss": 2.8511,
+      "step": 28695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.373394388483159e-05,
+      "loss": 2.9116,
+      "step": 28700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.372566352869355e-05,
+      "loss": 2.8896,
+      "step": 28705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.371738233268953e-05,
+      "loss": 3.0067,
+      "step": 28710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.370910029711262e-05,
+      "loss": 2.9158,
+      "step": 28715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.370081742225602e-05,
+      "loss": 2.9033,
+      "step": 28720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.369253370841297e-05,
+      "loss": 2.857,
+      "step": 28725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.368424915587666e-05,
+      "loss": 2.8531,
+      "step": 28730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.367596376494037e-05,
+      "loss": 2.9158,
+      "step": 28735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.366767753589738e-05,
+      "loss": 3.0229,
+      "step": 28740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.365939046904106e-05,
+      "loss": 2.9089,
+      "step": 28745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.36511025646647e-05,
+      "loss": 2.8074,
+      "step": 28750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.364281382306175e-05,
+      "loss": 2.9116,
+      "step": 28755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.363452424452558e-05,
+      "loss": 2.9838,
+      "step": 28760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.362623382934963e-05,
+      "loss": 2.9924,
+      "step": 28765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.361794257782742e-05,
+      "loss": 2.8674,
+      "step": 28770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.36096504902524e-05,
+      "loss": 2.9027,
+      "step": 28775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.360135756691813e-05,
+      "loss": 2.7821,
+      "step": 28780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.359306380811814e-05,
+      "loss": 2.9686,
+      "step": 28785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.358476921414608e-05,
+      "loss": 2.9104,
+      "step": 28790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.35764737852955e-05,
+      "loss": 2.9781,
+      "step": 28795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.356817752186011e-05,
+      "loss": 2.9074,
+      "step": 28800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.355988042413357e-05,
+      "loss": 2.8906,
+      "step": 28805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.355158249240957e-05,
+      "loss": 2.9487,
+      "step": 28810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.354328372698189e-05,
+      "loss": 2.9026,
+      "step": 28815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.353498412814424e-05,
+      "loss": 2.9542,
+      "step": 28820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.352668369619047e-05,
+      "loss": 2.8573,
+      "step": 28825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.351838243141439e-05,
+      "loss": 2.9047,
+      "step": 28830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.351008033410984e-05,
+      "loss": 2.8528,
+      "step": 28835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.350177740457074e-05,
+      "loss": 2.9286,
+      "step": 28840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.349347364309097e-05,
+      "loss": 2.9089,
+      "step": 28845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.34851690499645e-05,
+      "loss": 2.9769,
+      "step": 28850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.34768636254853e-05,
+      "loss": 2.8591,
+      "step": 28855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.346855736994738e-05,
+      "loss": 2.9194,
+      "step": 28860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.346025028364477e-05,
+      "loss": 2.8392,
+      "step": 28865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.345194236687149e-05,
+      "loss": 3.0594,
+      "step": 28870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.34436336199217e-05,
+      "loss": 3.0519,
+      "step": 28875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.34353240430895e-05,
+      "loss": 2.9299,
+      "step": 28880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.342701363666901e-05,
+      "loss": 2.9604,
+      "step": 28885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.341870240095444e-05,
+      "loss": 2.9547,
+      "step": 28890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.341039033623999e-05,
+      "loss": 2.9438,
+      "step": 28895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.340207744281992e-05,
+      "loss": 2.8959,
+      "step": 28900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.339376372098848e-05,
+      "loss": 2.8271,
+      "step": 28905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.338544917103995e-05,
+      "loss": 2.8839,
+      "step": 28910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.33771337932687e-05,
+      "loss": 2.9338,
+      "step": 28915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.336881758796905e-05,
+      "loss": 3.0141,
+      "step": 28920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.33605005554354e-05,
+      "loss": 2.8166,
+      "step": 28925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.335218269596217e-05,
+      "loss": 2.9502,
+      "step": 28930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.334386400984379e-05,
+      "loss": 3.0691,
+      "step": 28935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.333554449737475e-05,
+      "loss": 2.927,
+      "step": 28940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.332722415884953e-05,
+      "loss": 2.8963,
+      "step": 28945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.331890299456269e-05,
+      "loss": 2.9549,
+      "step": 28950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.331058100480879e-05,
+      "loss": 2.8775,
+      "step": 28955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.33022581898824e-05,
+      "loss": 2.8245,
+      "step": 28960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.329393455007813e-05,
+      "loss": 2.8653,
+      "step": 28965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.328561008569065e-05,
+      "loss": 2.8681,
+      "step": 28970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.327728479701466e-05,
+      "loss": 2.9788,
+      "step": 28975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.326895868434482e-05,
+      "loss": 2.8794,
+      "step": 28980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.32606317479759e-05,
+      "loss": 2.9737,
+      "step": 28985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.325230398820265e-05,
+      "loss": 2.9541,
+      "step": 28990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.324397540531987e-05,
+      "loss": 2.9555,
+      "step": 28995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.32356459996224e-05,
+      "loss": 2.7762,
+      "step": 29000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.322731577140507e-05,
+      "loss": 2.9775,
+      "step": 29005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.321898472096276e-05,
+      "loss": 2.8562,
+      "step": 29010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.32106528485904e-05,
+      "loss": 2.8627,
+      "step": 29015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.32023201545829e-05,
+      "loss": 2.8909,
+      "step": 29020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.319398663923528e-05,
+      "loss": 2.9292,
+      "step": 29025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.318565230284249e-05,
+      "loss": 2.9005,
+      "step": 29030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.317731714569958e-05,
+      "loss": 2.8591,
+      "step": 29035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.31689811681016e-05,
+      "loss": 2.9002,
+      "step": 29040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.316064437034365e-05,
+      "loss": 2.9452,
+      "step": 29045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.315230675272082e-05,
+      "loss": 2.8793,
+      "step": 29050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.314396831552828e-05,
+      "loss": 2.9123,
+      "step": 29055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.31356290590612e-05,
+      "loss": 2.8659,
+      "step": 29060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.312728898361474e-05,
+      "loss": 2.8919,
+      "step": 29065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.311894808948419e-05,
+      "loss": 2.9551,
+      "step": 29070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.311060637696476e-05,
+      "loss": 2.9075,
+      "step": 29075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.310226384635178e-05,
+      "loss": 2.9353,
+      "step": 29080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.309392049794054e-05,
+      "loss": 2.8747,
+      "step": 29085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.30855763320264e-05,
+      "loss": 2.8627,
+      "step": 29090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.307723134890473e-05,
+      "loss": 2.8979,
+      "step": 29095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.306888554887093e-05,
+      "loss": 2.9405,
+      "step": 29100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.306053893222043e-05,
+      "loss": 2.8935,
+      "step": 29105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.305219149924874e-05,
+      "loss": 2.9763,
+      "step": 29110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.304384325025128e-05,
+      "loss": 2.9147,
+      "step": 29115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.30354941855236e-05,
+      "loss": 2.7694,
+      "step": 29120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.302714430536126e-05,
+      "loss": 2.8949,
+      "step": 29125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.301879361005984e-05,
+      "loss": 2.8964,
+      "step": 29130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.301044209991493e-05,
+      "loss": 2.9292,
+      "step": 29135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.300208977522216e-05,
+      "loss": 3.0472,
+      "step": 29140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.299373663627722e-05,
+      "loss": 2.9596,
+      "step": 29145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.298538268337579e-05,
+      "loss": 2.8014,
+      "step": 29150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.297702791681358e-05,
+      "loss": 2.9734,
+      "step": 29155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.296867233688636e-05,
+      "loss": 2.9091,
+      "step": 29160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.29603159438899e-05,
+      "loss": 3.0297,
+      "step": 29165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.295195873812e-05,
+      "loss": 3.0497,
+      "step": 29170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.294360071987252e-05,
+      "loss": 2.8843,
+      "step": 29175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.29352418894433e-05,
+      "loss": 2.8054,
+      "step": 29180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.292688224712826e-05,
+      "loss": 2.8863,
+      "step": 29185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.291852179322332e-05,
+      "loss": 2.8688,
+      "step": 29190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.291016052802439e-05,
+      "loss": 2.8654,
+      "step": 29195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.290179845182752e-05,
+      "loss": 2.9648,
+      "step": 29200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.289343556492867e-05,
+      "loss": 2.859,
+      "step": 29205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.288507186762389e-05,
+      "loss": 2.8917,
+      "step": 29210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.287670736020926e-05,
+      "loss": 2.9999,
+      "step": 29215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.286834204298085e-05,
+      "loss": 2.9131,
+      "step": 29220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.285997591623482e-05,
+      "loss": 2.9164,
+      "step": 29225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.285160898026727e-05,
+      "loss": 2.8334,
+      "step": 29230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.284324123537444e-05,
+      "loss": 2.9488,
+      "step": 29235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.28348726818525e-05,
+      "loss": 2.8616,
+      "step": 29240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.282650331999772e-05,
+      "loss": 2.9873,
+      "step": 29245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.281813315010633e-05,
+      "loss": 2.9563,
+      "step": 29250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.280976217247464e-05,
+      "loss": 3.0066,
+      "step": 29255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.280139038739901e-05,
+      "loss": 2.9123,
+      "step": 29260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.279301779517575e-05,
+      "loss": 3.0174,
+      "step": 29265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.278464439610127e-05,
+      "loss": 2.9415,
+      "step": 29270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.277627019047193e-05,
+      "loss": 2.9386,
+      "step": 29275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.276789517858423e-05,
+      "loss": 2.853,
+      "step": 29280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.275951936073462e-05,
+      "loss": 2.8994,
+      "step": 29285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.275114273721957e-05,
+      "loss": 2.9802,
+      "step": 29290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.274276530833564e-05,
+      "loss": 2.9745,
+      "step": 29295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.273438707437936e-05,
+      "loss": 2.9373,
+      "step": 29300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.272600803564732e-05,
+      "loss": 2.9159,
+      "step": 29305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.271762819243611e-05,
+      "loss": 2.9565,
+      "step": 29310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.27092475450424e-05,
+      "loss": 2.9111,
+      "step": 29315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.270086609376286e-05,
+      "loss": 2.8596,
+      "step": 29320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.269248383889415e-05,
+      "loss": 2.8576,
+      "step": 29325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.268410078073303e-05,
+      "loss": 2.9176,
+      "step": 29330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.267571691957621e-05,
+      "loss": 2.9672,
+      "step": 29335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.266733225572051e-05,
+      "loss": 2.989,
+      "step": 29340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.265894678946273e-05,
+      "loss": 2.8514,
+      "step": 29345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.265056052109969e-05,
+      "loss": 2.9141,
+      "step": 29350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.264217345092828e-05,
+      "loss": 2.9442,
+      "step": 29355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.263378557924537e-05,
+      "loss": 3.0106,
+      "step": 29360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.26253969063479e-05,
+      "loss": 2.8785,
+      "step": 29365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.261700743253283e-05,
+      "loss": 2.9473,
+      "step": 29370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.260861715809712e-05,
+      "loss": 2.9089,
+      "step": 29375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.260022608333776e-05,
+      "loss": 2.9381,
+      "step": 29380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.259183420855184e-05,
+      "loss": 2.912,
+      "step": 29385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.258344153403639e-05,
+      "loss": 2.9331,
+      "step": 29390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.25750480600885e-05,
+      "loss": 2.9539,
+      "step": 29395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.25666537870053e-05,
+      "loss": 2.8593,
+      "step": 29400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.255825871508393e-05,
+      "loss": 2.9327,
+      "step": 29405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.25498628446216e-05,
+      "loss": 2.9128,
+      "step": 29410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.254146617591547e-05,
+      "loss": 2.9985,
+      "step": 29415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.25330687092628e-05,
+      "loss": 2.9034,
+      "step": 29420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.252467044496085e-05,
+      "loss": 2.9169,
+      "step": 29425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.25162713833069e-05,
+      "loss": 2.7859,
+      "step": 29430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.250787152459827e-05,
+      "loss": 2.8961,
+      "step": 29435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.249947086913233e-05,
+      "loss": 2.831,
+      "step": 29440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.249106941720643e-05,
+      "loss": 2.8587,
+      "step": 29445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.2482667169118e-05,
+      "loss": 2.9023,
+      "step": 29450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.247426412516443e-05,
+      "loss": 2.9014,
+      "step": 29455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.246586028564321e-05,
+      "loss": 2.9217,
+      "step": 29460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.245745565085184e-05,
+      "loss": 2.9302,
+      "step": 29465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.24490502210878e-05,
+      "loss": 2.9024,
+      "step": 29470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.244064399664866e-05,
+      "loss": 2.9381,
+      "step": 29475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.243223697783197e-05,
+      "loss": 2.8686,
+      "step": 29480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.242382916493536e-05,
+      "loss": 2.9292,
+      "step": 29485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.241542055825644e-05,
+      "loss": 2.9875,
+      "step": 29490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.240701115809287e-05,
+      "loss": 2.9923,
+      "step": 29495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.239860096474233e-05,
+      "loss": 2.8487,
+      "step": 29500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.239018997850255e-05,
+      "loss": 2.9519,
+      "step": 29505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.238177819967126e-05,
+      "loss": 2.9417,
+      "step": 29510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.237336562854623e-05,
+      "loss": 2.9498,
+      "step": 29515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.236495226542523e-05,
+      "loss": 2.985,
+      "step": 29520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.235653811060613e-05,
+      "loss": 2.9075,
+      "step": 29525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.234812316438676e-05,
+      "loss": 2.9237,
+      "step": 29530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.233970742706501e-05,
+      "loss": 2.8501,
+      "step": 29535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.233129089893878e-05,
+      "loss": 2.9895,
+      "step": 29540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.232287358030601e-05,
+      "loss": 2.9533,
+      "step": 29545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.231445547146467e-05,
+      "loss": 2.9251,
+      "step": 29550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.230603657271274e-05,
+      "loss": 2.9009,
+      "step": 29555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.229761688434825e-05,
+      "loss": 2.9648,
+      "step": 29560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.228919640666926e-05,
+      "loss": 2.8492,
+      "step": 29565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.228077513997382e-05,
+      "loss": 2.9581,
+      "step": 29570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.227235308456006e-05,
+      "loss": 2.9362,
+      "step": 29575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.226393024072609e-05,
+      "loss": 2.8493,
+      "step": 29580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.225550660877012e-05,
+      "loss": 2.8939,
+      "step": 29585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.224708218899026e-05,
+      "loss": 2.8589,
+      "step": 29590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.223865698168479e-05,
+      "loss": 2.8989,
+      "step": 29595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.223023098715195e-05,
+      "loss": 2.9493,
+      "step": 29600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.222180420568997e-05,
+      "loss": 3.0184,
+      "step": 29605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.221337663759719e-05,
+      "loss": 2.8222,
+      "step": 29610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.220494828317192e-05,
+      "loss": 2.8514,
+      "step": 29615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.219651914271251e-05,
+      "loss": 2.8451,
+      "step": 29620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.218808921651736e-05,
+      "loss": 2.8551,
+      "step": 29625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.217965850488489e-05,
+      "loss": 2.9513,
+      "step": 29630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.217122700811349e-05,
+      "loss": 3.0594,
+      "step": 29635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.21627947265017e-05,
+      "loss": 2.9313,
+      "step": 29640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.215436166034795e-05,
+      "loss": 2.8376,
+      "step": 29645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.21459278099508e-05,
+      "loss": 2.8932,
+      "step": 29650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.213749317560878e-05,
+      "loss": 2.885,
+      "step": 29655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.212905775762048e-05,
+      "loss": 2.8201,
+      "step": 29660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.212062155628449e-05,
+      "loss": 2.851,
+      "step": 29665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.211218457189948e-05,
+      "loss": 3.0111,
+      "step": 29670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.210374680476406e-05,
+      "loss": 2.9116,
+      "step": 29675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.209530825517697e-05,
+      "loss": 2.9245,
+      "step": 29680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.208686892343689e-05,
+      "loss": 3.0946,
+      "step": 29685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.207842880984258e-05,
+      "loss": 2.8646,
+      "step": 29690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.206998791469282e-05,
+      "loss": 2.8728,
+      "step": 29695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.206154623828639e-05,
+      "loss": 2.9163,
+      "step": 29700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.205310378092211e-05,
+      "loss": 2.7843,
+      "step": 29705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.204466054289886e-05,
+      "loss": 2.9197,
+      "step": 29710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.203621652451553e-05,
+      "loss": 2.9149,
+      "step": 29715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.202777172607101e-05,
+      "loss": 2.9588,
+      "step": 29720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.201932614786423e-05,
+      "loss": 2.9868,
+      "step": 29725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.201087979019416e-05,
+      "loss": 2.9215,
+      "step": 29730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.200243265335981e-05,
+      "loss": 2.9484,
+      "step": 29735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.19939847376602e-05,
+      "loss": 2.9395,
+      "step": 29740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.198553604339436e-05,
+      "loss": 2.8608,
+      "step": 29745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.197708657086139e-05,
+      "loss": 2.8416,
+      "step": 29750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.196863632036036e-05,
+      "loss": 3.0265,
+      "step": 29755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.196018529219043e-05,
+      "loss": 3.0192,
+      "step": 29760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.195173348665075e-05,
+      "loss": 2.8725,
+      "step": 29765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.194328090404049e-05,
+      "loss": 2.9229,
+      "step": 29770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.193482754465888e-05,
+      "loss": 3.0057,
+      "step": 29775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.192637340880515e-05,
+      "loss": 2.9618,
+      "step": 29780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.19179184967786e-05,
+      "loss": 2.8864,
+      "step": 29785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.190946280887849e-05,
+      "loss": 2.9069,
+      "step": 29790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.190100634540417e-05,
+      "loss": 2.9189,
+      "step": 29795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.189254910665494e-05,
+      "loss": 2.9243,
+      "step": 29800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.188409109293024e-05,
+      "loss": 2.8964,
+      "step": 29805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.187563230452945e-05,
+      "loss": 2.8828,
+      "step": 29810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.1867172741752e-05,
+      "loss": 2.9696,
+      "step": 29815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.185871240489736e-05,
+      "loss": 2.9551,
+      "step": 29820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.185025129426499e-05,
+      "loss": 2.9532,
+      "step": 29825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.184178941015445e-05,
+      "loss": 2.8156,
+      "step": 29830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.183332675286525e-05,
+      "loss": 2.8666,
+      "step": 29835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.182486332269696e-05,
+      "loss": 2.9652,
+      "step": 29840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.181639911994921e-05,
+      "loss": 2.8096,
+      "step": 29845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.180793414492158e-05,
+      "loss": 2.9251,
+      "step": 29850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.179946839791377e-05,
+      "loss": 2.9929,
+      "step": 29855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.17910018792254e-05,
+      "loss": 2.9449,
+      "step": 29860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.178253458915624e-05,
+      "loss": 2.9011,
+      "step": 29865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.177406652800599e-05,
+      "loss": 2.944,
+      "step": 29870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.176559769607438e-05,
+      "loss": 2.8794,
+      "step": 29875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.175712809366127e-05,
+      "loss": 2.8848,
+      "step": 29880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.174865772106643e-05,
+      "loss": 3.0043,
+      "step": 29885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.174018657858972e-05,
+      "loss": 2.987,
+      "step": 29890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.1731714666531e-05,
+      "loss": 2.9567,
+      "step": 29895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.172324198519016e-05,
+      "loss": 2.938,
+      "step": 29900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.171476853486716e-05,
+      "loss": 2.8717,
+      "step": 29905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.17062943158619e-05,
+      "loss": 2.8823,
+      "step": 29910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.169781932847441e-05,
+      "loss": 2.9695,
+      "step": 29915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.168934357300467e-05,
+      "loss": 2.8692,
+      "step": 29920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.16808670497527e-05,
+      "loss": 2.9821,
+      "step": 29925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.16723897590186e-05,
+      "loss": 2.8691,
+      "step": 29930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.166391170110242e-05,
+      "loss": 2.9127,
+      "step": 29935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.16554328763043e-05,
+      "loss": 2.9791,
+      "step": 29940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.164695328492437e-05,
+      "loss": 2.7399,
+      "step": 29945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.163847292726283e-05,
+      "loss": 2.9916,
+      "step": 29950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.162999180361982e-05,
+      "loss": 2.8991,
+      "step": 29955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.162150991429561e-05,
+      "loss": 3.0544,
+      "step": 29960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.161302725959043e-05,
+      "loss": 2.9137,
+      "step": 29965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.160454383980454e-05,
+      "loss": 2.9366,
+      "step": 29970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.15960596552383e-05,
+      "loss": 2.9414,
+      "step": 29975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.1587574706192e-05,
+      "loss": 2.9007,
+      "step": 29980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.157908899296602e-05,
+      "loss": 2.8868,
+      "step": 29985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.157060251586072e-05,
+      "loss": 2.9794,
+      "step": 29990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.156211527517653e-05,
+      "loss": 2.9673,
+      "step": 29995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.155362727121389e-05,
+      "loss": 2.8317,
+      "step": 30000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.154513850427327e-05,
+      "loss": 2.9685,
+      "step": 30005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.153664897465516e-05,
+      "loss": 2.9912,
+      "step": 30010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.152815868266008e-05,
+      "loss": 2.932,
+      "step": 30015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.151966762858858e-05,
+      "loss": 2.9819,
+      "step": 30020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.151117581274124e-05,
+      "loss": 3.0265,
+      "step": 30025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.150268323541866e-05,
+      "loss": 2.8638,
+      "step": 30030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.149418989692145e-05,
+      "loss": 2.8968,
+      "step": 30035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.148569579755031e-05,
+      "loss": 2.8834,
+      "step": 30040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.147720093760588e-05,
+      "loss": 2.9045,
+      "step": 30045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.146870531738887e-05,
+      "loss": 2.88,
+      "step": 30050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.146020893720007e-05,
+      "loss": 2.9791,
+      "step": 30055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.145171179734018e-05,
+      "loss": 2.9694,
+      "step": 30060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.144321389811003e-05,
+      "loss": 2.9261,
+      "step": 30065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.14347152398104e-05,
+      "loss": 2.9523,
+      "step": 30070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.142621582274218e-05,
+      "loss": 2.9433,
+      "step": 30075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.141771564720623e-05,
+      "loss": 3.0046,
+      "step": 30080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.140921471350341e-05,
+      "loss": 3.0149,
+      "step": 30085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.140071302193471e-05,
+      "loss": 2.8853,
+      "step": 30090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.139221057280103e-05,
+      "loss": 2.8284,
+      "step": 30095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.138370736640336e-05,
+      "loss": 2.9463,
+      "step": 30100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.137520340304271e-05,
+      "loss": 2.8725,
+      "step": 30105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.136669868302011e-05,
+      "loss": 2.8936,
+      "step": 30110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.135819320663663e-05,
+      "loss": 2.8839,
+      "step": 30115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.134968697419334e-05,
+      "loss": 2.8686,
+      "step": 30120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.134117998599137e-05,
+      "loss": 2.9571,
+      "step": 30125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.133267224233184e-05,
+      "loss": 2.89,
+      "step": 30130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.132416374351594e-05,
+      "loss": 2.8883,
+      "step": 30135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.131565448984482e-05,
+      "loss": 2.8502,
+      "step": 30140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.130714448161974e-05,
+      "loss": 2.9626,
+      "step": 30145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.129863371914195e-05,
+      "loss": 2.9588,
+      "step": 30150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.129012220271269e-05,
+      "loss": 2.936,
+      "step": 30155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.128160993263329e-05,
+      "loss": 2.7968,
+      "step": 30160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.127309690920503e-05,
+      "loss": 2.9057,
+      "step": 30165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.126458313272932e-05,
+      "loss": 2.9188,
+      "step": 30170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.125606860350752e-05,
+      "loss": 3.0068,
+      "step": 30175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.124755332184103e-05,
+      "loss": 2.933,
+      "step": 30180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.123903728803126e-05,
+      "loss": 2.8213,
+      "step": 30185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.12305205023797e-05,
+      "loss": 2.8535,
+      "step": 30190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.122200296518785e-05,
+      "loss": 2.8975,
+      "step": 30195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.12134846767572e-05,
+      "loss": 2.8223,
+      "step": 30200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.120496563738928e-05,
+      "loss": 2.9244,
+      "step": 30205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.119644584738568e-05,
+      "loss": 2.8882,
+      "step": 30210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.118792530704796e-05,
+      "loss": 2.9055,
+      "step": 30215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.117940401667777e-05,
+      "loss": 2.953,
+      "step": 30220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.117088197657674e-05,
+      "loss": 2.9611,
+      "step": 30225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.116235918704657e-05,
+      "loss": 2.9437,
+      "step": 30230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.115383564838891e-05,
+      "loss": 2.854,
+      "step": 30235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.114531136090553e-05,
+      "loss": 2.836,
+      "step": 30240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.113678632489816e-05,
+      "loss": 2.9888,
+      "step": 30245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.112826054066856e-05,
+      "loss": 2.8903,
+      "step": 30250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.111973400851858e-05,
+      "loss": 2.9813,
+      "step": 30255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.111120672875001e-05,
+      "loss": 2.887,
+      "step": 30260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.110267870166473e-05,
+      "loss": 2.9734,
+      "step": 30265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.10941499275646e-05,
+      "loss": 2.9805,
+      "step": 30270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.108562040675158e-05,
+      "loss": 2.987,
+      "step": 30275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.107709013952757e-05,
+      "loss": 2.895,
+      "step": 30280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.106855912619454e-05,
+      "loss": 2.9934,
+      "step": 30285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.106002736705446e-05,
+      "loss": 2.9269,
+      "step": 30290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.105149486240939e-05,
+      "loss": 2.849,
+      "step": 30295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.104296161256134e-05,
+      "loss": 2.961,
+      "step": 30300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.103442761781237e-05,
+      "loss": 2.9246,
+      "step": 30305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.102589287846461e-05,
+      "loss": 2.8884,
+      "step": 30310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.101735739482016e-05,
+      "loss": 2.9045,
+      "step": 30315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.100882116718116e-05,
+      "loss": 2.8462,
+      "step": 30320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.100028419584981e-05,
+      "loss": 2.8328,
+      "step": 30325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.099174648112828e-05,
+      "loss": 2.9106,
+      "step": 30330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.098320802331883e-05,
+      "loss": 2.9721,
+      "step": 30335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.09746688227237e-05,
+      "loss": 2.9252,
+      "step": 30340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.096612887964517e-05,
+      "loss": 2.8905,
+      "step": 30345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.095758819438552e-05,
+      "loss": 2.9835,
+      "step": 30350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.094904676724711e-05,
+      "loss": 2.8393,
+      "step": 30355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.094050459853232e-05,
+      "loss": 2.8812,
+      "step": 30360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.093196168854349e-05,
+      "loss": 2.8758,
+      "step": 30365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.092341803758305e-05,
+      "loss": 2.9964,
+      "step": 30370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.091487364595342e-05,
+      "loss": 2.8417,
+      "step": 30375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.090632851395711e-05,
+      "loss": 2.9167,
+      "step": 30380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.089778264189656e-05,
+      "loss": 2.9276,
+      "step": 30385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.08892360300743e-05,
+      "loss": 2.9088,
+      "step": 30390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.088068867879289e-05,
+      "loss": 2.925,
+      "step": 30395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.087214058835489e-05,
+      "loss": 2.888,
+      "step": 30400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.086359175906287e-05,
+      "loss": 2.8995,
+      "step": 30405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.085504219121948e-05,
+      "loss": 3.0731,
+      "step": 30410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.084649188512735e-05,
+      "loss": 2.8927,
+      "step": 30415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.083794084108917e-05,
+      "loss": 2.8406,
+      "step": 30420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.082938905940762e-05,
+      "loss": 2.9564,
+      "step": 30425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.082083654038544e-05,
+      "loss": 2.8223,
+      "step": 30430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.081228328432536e-05,
+      "loss": 2.9299,
+      "step": 30435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.080372929153018e-05,
+      "loss": 2.8702,
+      "step": 30440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.079517456230268e-05,
+      "loss": 2.982,
+      "step": 30445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.078661909694574e-05,
+      "loss": 2.8083,
+      "step": 30450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.077806289576216e-05,
+      "loss": 2.8279,
+      "step": 30455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.076950595905483e-05,
+      "loss": 2.8669,
+      "step": 30460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.076094828712667e-05,
+      "loss": 2.8663,
+      "step": 30465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.075238988028061e-05,
+      "loss": 2.9318,
+      "step": 30470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.074383073881962e-05,
+      "loss": 2.8729,
+      "step": 30475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.073527086304668e-05,
+      "loss": 2.8968,
+      "step": 30480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.07267102532648e-05,
+      "loss": 2.9133,
+      "step": 30485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.071814890977699e-05,
+      "loss": 2.9584,
+      "step": 30490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.070958683288635e-05,
+      "loss": 2.8848,
+      "step": 30495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.070102402289598e-05,
+      "loss": 2.898,
+      "step": 30500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.069246048010895e-05,
+      "loss": 2.9328,
+      "step": 30505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.068389620482843e-05,
+      "loss": 2.9751,
+      "step": 30510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.067533119735758e-05,
+      "loss": 2.8194,
+      "step": 30515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.06667654579996e-05,
+      "loss": 3.0569,
+      "step": 30520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.06581989870577e-05,
+      "loss": 2.9046,
+      "step": 30525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.064963178483512e-05,
+      "loss": 2.8758,
+      "step": 30530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.064106385163516e-05,
+      "loss": 2.9394,
+      "step": 30535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.06324951877611e-05,
+      "loss": 2.902,
+      "step": 30540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.062392579351625e-05,
+      "loss": 2.8551,
+      "step": 30545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.061535566920397e-05,
+      "loss": 2.9254,
+      "step": 30550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.060678481512762e-05,
+      "loss": 3.0123,
+      "step": 30555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.059821323159062e-05,
+      "loss": 2.9214,
+      "step": 30560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.058964091889638e-05,
+      "loss": 2.9409,
+      "step": 30565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.058106787734838e-05,
+      "loss": 2.8777,
+      "step": 30570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.057249410725006e-05,
+      "loss": 2.9468,
+      "step": 30575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.056391960890494e-05,
+      "loss": 2.8538,
+      "step": 30580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.055534438261654e-05,
+      "loss": 2.8952,
+      "step": 30585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.054676842868846e-05,
+      "loss": 3.0147,
+      "step": 30590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.053819174742422e-05,
+      "loss": 2.864,
+      "step": 30595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.052961433912746e-05,
+      "loss": 2.8547,
+      "step": 30600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.052103620410181e-05,
+      "loss": 2.9817,
+      "step": 30605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.051245734265091e-05,
+      "loss": 2.8832,
+      "step": 30610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.050387775507845e-05,
+      "loss": 2.9165,
+      "step": 30615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.049529744168816e-05,
+      "loss": 2.9665,
+      "step": 30620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.048671640278375e-05,
+      "loss": 2.9529,
+      "step": 30625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.047813463866899e-05,
+      "loss": 2.857,
+      "step": 30630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.046955214964766e-05,
+      "loss": 2.8231,
+      "step": 30635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.04609689360236e-05,
+      "loss": 2.9612,
+      "step": 30640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.04523849981006e-05,
+      "loss": 2.9393,
+      "step": 30645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.044380033618258e-05,
+      "loss": 2.8486,
+      "step": 30650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.043521495057339e-05,
+      "loss": 2.97,
+      "step": 30655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.042662884157693e-05,
+      "loss": 2.8701,
+      "step": 30660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.041804200949717e-05,
+      "loss": 2.9308,
+      "step": 30665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.040945445463809e-05,
+      "loss": 2.8384,
+      "step": 30670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.040086617730363e-05,
+      "loss": 2.9154,
+      "step": 30675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.039227717779786e-05,
+      "loss": 2.8539,
+      "step": 30680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.03836874564248e-05,
+      "loss": 2.9339,
+      "step": 30685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.03750970134885e-05,
+      "loss": 2.9691,
+      "step": 30690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.03665058492931e-05,
+      "loss": 2.8477,
+      "step": 30695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.035791396414267e-05,
+      "loss": 2.8593,
+      "step": 30700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.034932135834136e-05,
+      "loss": 2.8363,
+      "step": 30705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.034072803219338e-05,
+      "loss": 2.9579,
+      "step": 30710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.033213398600287e-05,
+      "loss": 2.9809,
+      "step": 30715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.03235392200741e-05,
+      "loss": 2.9617,
+      "step": 30720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.03149437347113e-05,
+      "loss": 2.9589,
+      "step": 30725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.030634753021874e-05,
+      "loss": 2.9407,
+      "step": 30730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.029775060690071e-05,
+      "loss": 2.9504,
+      "step": 30735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.028915296506156e-05,
+      "loss": 2.8254,
+      "step": 30740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.02805546050056e-05,
+      "loss": 2.9001,
+      "step": 30745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.027195552703722e-05,
+      "loss": 3.0013,
+      "step": 30750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.026335573146084e-05,
+      "loss": 2.9238,
+      "step": 30755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.025475521858085e-05,
+      "loss": 2.898,
+      "step": 30760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.024615398870172e-05,
+      "loss": 2.9474,
+      "step": 30765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.023755204212792e-05,
+      "loss": 2.8724,
+      "step": 30770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.022894937916395e-05,
+      "loss": 2.9104,
+      "step": 30775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.022034600011436e-05,
+      "loss": 2.9829,
+      "step": 30780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.021174190528366e-05,
+      "loss": 2.8808,
+      "step": 30785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.020313709497645e-05,
+      "loss": 2.89,
+      "step": 30790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.019453156949732e-05,
+      "loss": 2.8489,
+      "step": 30795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.018592532915091e-05,
+      "loss": 2.922,
+      "step": 30800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.017731837424189e-05,
+      "loss": 2.9445,
+      "step": 30805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.01687107050749e-05,
+      "loss": 2.9034,
+      "step": 30810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.016010232195465e-05,
+      "loss": 2.8972,
+      "step": 30815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.01514932251859e-05,
+      "loss": 2.8073,
+      "step": 30820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.014288341507337e-05,
+      "loss": 2.8703,
+      "step": 30825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.013427289192187e-05,
+      "loss": 2.9409,
+      "step": 30830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.012566165603618e-05,
+      "loss": 2.9168,
+      "step": 30835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.011704970772113e-05,
+      "loss": 2.8388,
+      "step": 30840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.010843704728157e-05,
+      "loss": 2.9642,
+      "step": 30845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.009982367502241e-05,
+      "loss": 2.8497,
+      "step": 30850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.009120959124854e-05,
+      "loss": 2.9053,
+      "step": 30855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.008259479626485e-05,
+      "loss": 2.9451,
+      "step": 30860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.007397929037636e-05,
+      "loss": 2.8879,
+      "step": 30865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.006536307388803e-05,
+      "loss": 2.901,
+      "step": 30870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.005674614710484e-05,
+      "loss": 2.8949,
+      "step": 30875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.004812851033184e-05,
+      "loss": 2.901,
+      "step": 30880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.003951016387412e-05,
+      "loss": 3.002,
+      "step": 30885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.003089110803669e-05,
+      "loss": 2.9769,
+      "step": 30890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.002227134312471e-05,
+      "loss": 2.9087,
+      "step": 30895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.00136508694433e-05,
+      "loss": 2.8352,
+      "step": 30900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.000502968729759e-05,
+      "loss": 2.9029,
+      "step": 30905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.999640779699279e-05,
+      "loss": 2.9032,
+      "step": 30910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.998778519883411e-05,
+      "loss": 2.9317,
+      "step": 30915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.997916189312677e-05,
+      "loss": 2.9008,
+      "step": 30920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.997053788017602e-05,
+      "loss": 2.9986,
+      "step": 30925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.996191316028718e-05,
+      "loss": 2.8805,
+      "step": 30930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.99532877337655e-05,
+      "loss": 2.875,
+      "step": 30935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.994466160091635e-05,
+      "loss": 2.9465,
+      "step": 30940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.993603476204507e-05,
+      "loss": 2.9619,
+      "step": 30945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.992740721745704e-05,
+      "loss": 3.0094,
+      "step": 30950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.99187789674577e-05,
+      "loss": 2.8943,
+      "step": 30955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.991015001235242e-05,
+      "loss": 2.9054,
+      "step": 30960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.990152035244673e-05,
+      "loss": 2.9384,
+      "step": 30965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.989288998804606e-05,
+      "loss": 2.8936,
+      "step": 30970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.988425891945594e-05,
+      "loss": 3.0005,
+      "step": 30975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.987562714698189e-05,
+      "loss": 2.9571,
+      "step": 30980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.986699467092947e-05,
+      "loss": 2.7714,
+      "step": 30985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.985836149160428e-05,
+      "loss": 2.9387,
+      "step": 30990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.984972760931187e-05,
+      "loss": 2.853,
+      "step": 30995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.984109302435795e-05,
+      "loss": 2.9614,
+      "step": 31000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.983245773704811e-05,
+      "loss": 2.9257,
+      "step": 31005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.982382174768807e-05,
+      "loss": 3.0029,
+      "step": 31010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.981518505658351e-05,
+      "loss": 2.9978,
+      "step": 31015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.980654766404018e-05,
+      "loss": 2.9076,
+      "step": 31020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.979790957036383e-05,
+      "loss": 2.9335,
+      "step": 31025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.978927077586024e-05,
+      "loss": 2.9559,
+      "step": 31030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.978063128083522e-05,
+      "loss": 2.9022,
+      "step": 31035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.97719910855946e-05,
+      "loss": 2.9629,
+      "step": 31040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.97633501904442e-05,
+      "loss": 2.8279,
+      "step": 31045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.975470859568996e-05,
+      "loss": 2.9251,
+      "step": 31050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.974606630163773e-05,
+      "loss": 2.928,
+      "step": 31055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.973742330859347e-05,
+      "loss": 2.8712,
+      "step": 31060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.972877961686313e-05,
+      "loss": 2.9019,
+      "step": 31065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.972013522675269e-05,
+      "loss": 2.8459,
+      "step": 31070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.971149013856813e-05,
+      "loss": 2.961,
+      "step": 31075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.970284435261551e-05,
+      "loss": 2.8952,
+      "step": 31080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.969419786920086e-05,
+      "loss": 2.9913,
+      "step": 31085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.968555068863027e-05,
+      "loss": 2.9377,
+      "step": 31090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.967690281120984e-05,
+      "loss": 2.9321,
+      "step": 31095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.966825423724569e-05,
+      "loss": 2.953,
+      "step": 31100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.965960496704398e-05,
+      "loss": 2.8325,
+      "step": 31105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.965095500091087e-05,
+      "loss": 2.8712,
+      "step": 31110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.964230433915257e-05,
+      "loss": 2.9461,
+      "step": 31115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.963365298207533e-05,
+      "loss": 2.9154,
+      "step": 31120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.962500092998535e-05,
+      "loss": 2.8142,
+      "step": 31125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.961634818318896e-05,
+      "loss": 2.9432,
+      "step": 31130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.96076947419924e-05,
+      "loss": 3.0214,
+      "step": 31135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.959904060670206e-05,
+      "loss": 2.8958,
+      "step": 31140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.959038577762422e-05,
+      "loss": 2.9053,
+      "step": 31145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.958173025506529e-05,
+      "loss": 2.9763,
+      "step": 31150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.957307403933168e-05,
+      "loss": 2.8964,
+      "step": 31155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.956441713072976e-05,
+      "loss": 2.8784,
+      "step": 31160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.955575952956604e-05,
+      "loss": 2.93,
+      "step": 31165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.954710123614696e-05,
+      "loss": 2.8834,
+      "step": 31170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.953844225077902e-05,
+      "loss": 2.8815,
+      "step": 31175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.952978257376872e-05,
+      "loss": 2.9168,
+      "step": 31180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.952112220542263e-05,
+      "loss": 2.8659,
+      "step": 31185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.951246114604731e-05,
+      "loss": 2.9992,
+      "step": 31190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.950379939594935e-05,
+      "loss": 2.9015,
+      "step": 31195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.94951369554354e-05,
+      "loss": 2.9523,
+      "step": 31200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.948647382481202e-05,
+      "loss": 2.8744,
+      "step": 31205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.947781000438595e-05,
+      "loss": 2.9552,
+      "step": 31210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.946914549446387e-05,
+      "loss": 2.9297,
+      "step": 31215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.94604802953525e-05,
+      "loss": 2.9118,
+      "step": 31220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.945181440735854e-05,
+      "loss": 2.9117,
+      "step": 31225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.944314783078878e-05,
+      "loss": 2.8555,
+      "step": 31230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.943448056595003e-05,
+      "loss": 2.862,
+      "step": 31235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.942581261314906e-05,
+      "loss": 2.9019,
+      "step": 31240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.941714397269275e-05,
+      "loss": 2.8682,
+      "step": 31245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.940847464488794e-05,
+      "loss": 2.9571,
+      "step": 31250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.93998046300415e-05,
+      "loss": 2.9105,
+      "step": 31255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.939113392846035e-05,
+      "loss": 2.9394,
+      "step": 31260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.938246254045144e-05,
+      "loss": 2.9064,
+      "step": 31265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.937379046632174e-05,
+      "loss": 2.933,
+      "step": 31270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.93651177063782e-05,
+      "loss": 2.9508,
+      "step": 31275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.935644426092784e-05,
+      "loss": 2.9152,
+      "step": 31280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.93477701302777e-05,
+      "loss": 2.9745,
+      "step": 31285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.933909531473482e-05,
+      "loss": 2.95,
+      "step": 31290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.93304198146063e-05,
+      "loss": 2.8964,
+      "step": 31295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.932174363019922e-05,
+      "loss": 2.9592,
+      "step": 31300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.931306676182073e-05,
+      "loss": 2.9258,
+      "step": 31305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.930438920977797e-05,
+      "loss": 2.9416,
+      "step": 31310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.929571097437813e-05,
+      "loss": 2.8518,
+      "step": 31315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.92870320559284e-05,
+      "loss": 2.885,
+      "step": 31320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.927835245473602e-05,
+      "loss": 2.8831,
+      "step": 31325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.926967217110822e-05,
+      "loss": 2.9028,
+      "step": 31330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.926099120535228e-05,
+      "loss": 2.7874,
+      "step": 31335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.925230955777551e-05,
+      "loss": 2.8881,
+      "step": 31340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.924362722868521e-05,
+      "loss": 2.97,
+      "step": 31345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.923494421838875e-05,
+      "loss": 2.8367,
+      "step": 31350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.922626052719348e-05,
+      "loss": 2.9696,
+      "step": 31355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.92175761554068e-05,
+      "loss": 2.909,
+      "step": 31360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.920889110333615e-05,
+      "loss": 2.9156,
+      "step": 31365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.920020537128895e-05,
+      "loss": 2.9903,
+      "step": 31370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.919151895957267e-05,
+      "loss": 2.9708,
+      "step": 31375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.918283186849479e-05,
+      "loss": 2.9034,
+      "step": 31380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.917414409836285e-05,
+      "loss": 2.8978,
+      "step": 31385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.916545564948438e-05,
+      "loss": 2.9281,
+      "step": 31390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.915676652216693e-05,
+      "loss": 3.0772,
+      "step": 31395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.91480767167181e-05,
+      "loss": 2.8346,
+      "step": 31400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.913938623344548e-05,
+      "loss": 2.9993,
+      "step": 31405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.913069507265675e-05,
+      "loss": 2.9348,
+      "step": 31410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.912200323465952e-05,
+      "loss": 2.8296,
+      "step": 31415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.911331071976151e-05,
+      "loss": 2.9992,
+      "step": 31420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.91046175282704e-05,
+      "loss": 2.9563,
+      "step": 31425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.909592366049393e-05,
+      "loss": 2.9283,
+      "step": 31430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.908722911673986e-05,
+      "loss": 2.8843,
+      "step": 31435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.907853389731596e-05,
+      "loss": 2.86,
+      "step": 31440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.906983800253004e-05,
+      "loss": 2.9428,
+      "step": 31445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.906114143268992e-05,
+      "loss": 2.9009,
+      "step": 31450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.905244418810346e-05,
+      "loss": 2.7509,
+      "step": 31455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.904374626907851e-05,
+      "loss": 2.9048,
+      "step": 31460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9035047675923e-05,
+      "loss": 2.9435,
+      "step": 31465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.902634840894484e-05,
+      "loss": 2.8828,
+      "step": 31470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.901764846845197e-05,
+      "loss": 2.8436,
+      "step": 31475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.900894785475236e-05,
+      "loss": 2.9814,
+      "step": 31480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9000246568154e-05,
+      "loss": 2.8728,
+      "step": 31485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.899154460896494e-05,
+      "loss": 2.8582,
+      "step": 31490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.898284197749318e-05,
+      "loss": 2.9325,
+      "step": 31495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.897413867404678e-05,
+      "loss": 3.032,
+      "step": 31500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.896543469893387e-05,
+      "loss": 2.9834,
+      "step": 31505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.895673005246254e-05,
+      "loss": 3.0366,
+      "step": 31510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.894802473494093e-05,
+      "loss": 2.8753,
+      "step": 31515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.893931874667719e-05,
+      "loss": 2.919,
+      "step": 31520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.893061208797952e-05,
+      "loss": 2.8863,
+      "step": 31525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.89219047591561e-05,
+      "loss": 2.9731,
+      "step": 31530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.89131967605152e-05,
+      "loss": 3.0082,
+      "step": 31535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.890448809236504e-05,
+      "loss": 2.9545,
+      "step": 31540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.889577875501392e-05,
+      "loss": 3.0116,
+      "step": 31545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.888706874877015e-05,
+      "loss": 2.9256,
+      "step": 31550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.887835807394201e-05,
+      "loss": 2.9091,
+      "step": 31555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.88696467308379e-05,
+      "loss": 2.8431,
+      "step": 31560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.886093471976616e-05,
+      "loss": 2.9093,
+      "step": 31565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.885222204103521e-05,
+      "loss": 2.8009,
+      "step": 31570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.884350869495349e-05,
+      "loss": 2.863,
+      "step": 31575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.883479468182939e-05,
+      "loss": 2.8664,
+      "step": 31580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.88260800019714e-05,
+      "loss": 2.8964,
+      "step": 31585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.881736465568803e-05,
+      "loss": 2.9483,
+      "step": 31590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.880864864328776e-05,
+      "loss": 2.9391,
+      "step": 31595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.879993196507917e-05,
+      "loss": 2.9188,
+      "step": 31600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.879121462137078e-05,
+      "loss": 2.9546,
+      "step": 31605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.87824966124712e-05,
+      "loss": 2.9746,
+      "step": 31610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.877377793868904e-05,
+      "loss": 3.0175,
+      "step": 31615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.876505860033296e-05,
+      "loss": 2.8998,
+      "step": 31620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.875633859771155e-05,
+      "loss": 2.9554,
+      "step": 31625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.874761793113353e-05,
+      "loss": 2.9401,
+      "step": 31630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.87388966009076e-05,
+      "loss": 2.8578,
+      "step": 31635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.873017460734247e-05,
+      "loss": 2.8471,
+      "step": 31640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.872145195074694e-05,
+      "loss": 3.0387,
+      "step": 31645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.871272863142973e-05,
+      "loss": 2.924,
+      "step": 31650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.870400464969965e-05,
+      "loss": 2.8683,
+      "step": 31655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.869528000586553e-05,
+      "loss": 2.9745,
+      "step": 31660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.868655470023622e-05,
+      "loss": 2.945,
+      "step": 31665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.867782873312057e-05,
+      "loss": 2.8735,
+      "step": 31670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.866910210482748e-05,
+      "loss": 2.8971,
+      "step": 31675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.866037481566589e-05,
+      "loss": 2.9116,
+      "step": 31680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.865164686594468e-05,
+      "loss": 2.9266,
+      "step": 31685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.864291825597287e-05,
+      "loss": 2.8697,
+      "step": 31690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.86341889860594e-05,
+      "loss": 2.8774,
+      "step": 31695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.862545905651329e-05,
+      "loss": 2.9227,
+      "step": 31700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.861672846764357e-05,
+      "loss": 2.919,
+      "step": 31705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.860799721975933e-05,
+      "loss": 2.886,
+      "step": 31710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.85992653131696e-05,
+      "loss": 2.9173,
+      "step": 31715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.85905327481835e-05,
+      "loss": 2.9081,
+      "step": 31720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.858179952511017e-05,
+      "loss": 2.9306,
+      "step": 31725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.857306564425871e-05,
+      "loss": 2.9663,
+      "step": 31730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.856433110593836e-05,
+      "loss": 2.9069,
+      "step": 31735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.855559591045826e-05,
+      "loss": 2.9131,
+      "step": 31740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.854686005812765e-05,
+      "loss": 2.9223,
+      "step": 31745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.853812354925577e-05,
+      "loss": 2.9335,
+      "step": 31750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.852938638415187e-05,
+      "loss": 2.934,
+      "step": 31755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.852064856312527e-05,
+      "loss": 2.9928,
+      "step": 31760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.851191008648523e-05,
+      "loss": 2.8402,
+      "step": 31765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.850317095454114e-05,
+      "loss": 2.814,
+      "step": 31770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.849443116760231e-05,
+      "loss": 2.83,
+      "step": 31775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.848569072597814e-05,
+      "loss": 3.0176,
+      "step": 31780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.847694962997806e-05,
+      "loss": 2.8827,
+      "step": 31785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.846820787991144e-05,
+      "loss": 2.9517,
+      "step": 31790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.845946547608779e-05,
+      "loss": 2.9088,
+      "step": 31795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.845072241881651e-05,
+      "loss": 3.0063,
+      "step": 31800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.844197870840716e-05,
+      "loss": 2.9651,
+      "step": 31805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.843323434516923e-05,
+      "loss": 2.9244,
+      "step": 31810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.842448932941228e-05,
+      "loss": 2.9008,
+      "step": 31815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.841574366144587e-05,
+      "loss": 2.9529,
+      "step": 31820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.840699734157957e-05,
+      "loss": 2.9491,
+      "step": 31825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.839825037012302e-05,
+      "loss": 2.9322,
+      "step": 31830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.838950274738582e-05,
+      "loss": 2.9208,
+      "step": 31835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.838075447367767e-05,
+      "loss": 2.9469,
+      "step": 31840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.837200554930821e-05,
+      "loss": 2.9178,
+      "step": 31845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.836325597458714e-05,
+      "loss": 2.8573,
+      "step": 31850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.835450574982423e-05,
+      "loss": 2.914,
+      "step": 31855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.834575487532921e-05,
+      "loss": 2.9507,
+      "step": 31860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.833700335141185e-05,
+      "loss": 2.9966,
+      "step": 31865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.832825117838194e-05,
+      "loss": 2.9519,
+      "step": 31870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.831949835654931e-05,
+      "loss": 2.9724,
+      "step": 31875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.831074488622379e-05,
+      "loss": 2.9392,
+      "step": 31880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.830199076771523e-05,
+      "loss": 2.9674,
+      "step": 31885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.829323600133358e-05,
+      "loss": 2.8643,
+      "step": 31890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.828448058738869e-05,
+      "loss": 2.9328,
+      "step": 31895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.82757245261905e-05,
+      "loss": 2.9446,
+      "step": 31900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.826696781804898e-05,
+      "loss": 2.9819,
+      "step": 31905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.825821046327412e-05,
+      "loss": 2.8718,
+      "step": 31910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.82494524621759e-05,
+      "loss": 2.8615,
+      "step": 31915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.824069381506434e-05,
+      "loss": 2.8823,
+      "step": 31920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.823193452224955e-05,
+      "loss": 2.8444,
+      "step": 31925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.82231745840415e-05,
+      "loss": 2.9647,
+      "step": 31930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.821441400075037e-05,
+      "loss": 2.8585,
+      "step": 31935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.820565277268622e-05,
+      "loss": 2.9817,
+      "step": 31940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.819689090015921e-05,
+      "loss": 2.9303,
+      "step": 31945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.818812838347953e-05,
+      "loss": 2.8318,
+      "step": 31950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.817936522295732e-05,
+      "loss": 2.8766,
+      "step": 31955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.81706014189028e-05,
+      "loss": 2.8574,
+      "step": 31960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.816183697162622e-05,
+      "loss": 2.9619,
+      "step": 31965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.815307188143781e-05,
+      "loss": 2.8903,
+      "step": 31970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.814430614864785e-05,
+      "loss": 2.9444,
+      "step": 31975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.813553977356666e-05,
+      "loss": 2.8493,
+      "step": 31980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.812677275650454e-05,
+      "loss": 2.8557,
+      "step": 31985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.811800509777183e-05,
+      "loss": 2.8527,
+      "step": 31990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.810923679767893e-05,
+      "loss": 2.9664,
+      "step": 31995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.810046785653616e-05,
+      "loss": 2.8497,
+      "step": 32000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.809169827465401e-05,
+      "loss": 2.9941,
+      "step": 32005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.808292805234288e-05,
+      "loss": 2.9101,
+      "step": 32010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.807415718991324e-05,
+      "loss": 2.9132,
+      "step": 32015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.806538568767554e-05,
+      "loss": 2.9474,
+      "step": 32020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.80566135459403e-05,
+      "loss": 2.8246,
+      "step": 32025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.804784076501808e-05,
+      "loss": 2.8273,
+      "step": 32030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.803906734521937e-05,
+      "loss": 2.9181,
+      "step": 32035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.803029328685479e-05,
+      "loss": 2.9168,
+      "step": 32040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.802151859023489e-05,
+      "loss": 2.9618,
+      "step": 32045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.80127432556703e-05,
+      "loss": 2.9804,
+      "step": 32050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.800396728347168e-05,
+      "loss": 3.0305,
+      "step": 32055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.799519067394967e-05,
+      "loss": 2.9211,
+      "step": 32060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.798641342741498e-05,
+      "loss": 2.8567,
+      "step": 32065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.797763554417827e-05,
+      "loss": 2.9029,
+      "step": 32070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.796885702455031e-05,
+      "loss": 3.0384,
+      "step": 32075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.796007786884185e-05,
+      "loss": 2.9315,
+      "step": 32080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.795129807736363e-05,
+      "loss": 2.9927,
+      "step": 32085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.794251765042648e-05,
+      "loss": 2.8658,
+      "step": 32090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.793373658834119e-05,
+      "loss": 2.8246,
+      "step": 32095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.792495489141862e-05,
+      "loss": 3.0115,
+      "step": 32100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.791617255996963e-05,
+      "loss": 2.912,
+      "step": 32105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.790738959430511e-05,
+      "loss": 2.9383,
+      "step": 32110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.789860599473597e-05,
+      "loss": 2.8362,
+      "step": 32115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.788982176157314e-05,
+      "loss": 2.9029,
+      "step": 32120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.788103689512755e-05,
+      "loss": 2.9787,
+      "step": 32125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.78722513957102e-05,
+      "loss": 2.8881,
+      "step": 32130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.78634652636321e-05,
+      "loss": 2.958,
+      "step": 32135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.785467849920425e-05,
+      "loss": 2.977,
+      "step": 32140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.784589110273769e-05,
+      "loss": 2.9039,
+      "step": 32145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.78371030745435e-05,
+      "loss": 2.8776,
+      "step": 32150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.782831441493276e-05,
+      "loss": 2.9877,
+      "step": 32155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.78195251242166e-05,
+      "loss": 2.9085,
+      "step": 32160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.78107352027061e-05,
+      "loss": 2.9514,
+      "step": 32165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.780194465071247e-05,
+      "loss": 2.9127,
+      "step": 32170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.779315346854686e-05,
+      "loss": 2.914,
+      "step": 32175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.778436165652049e-05,
+      "loss": 2.9604,
+      "step": 32180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.777556921494454e-05,
+      "loss": 2.7847,
+      "step": 32185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.77667761441303e-05,
+      "loss": 2.9487,
+      "step": 32190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.775798244438901e-05,
+      "loss": 2.9266,
+      "step": 32195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.774918811603197e-05,
+      "loss": 2.8256,
+      "step": 32200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.774039315937051e-05,
+      "loss": 2.9982,
+      "step": 32205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.773159757471591e-05,
+      "loss": 2.8954,
+      "step": 32210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.772280136237957e-05,
+      "loss": 2.9615,
+      "step": 32215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.771400452267285e-05,
+      "loss": 2.7972,
+      "step": 32220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.770520705590716e-05,
+      "loss": 2.9331,
+      "step": 32225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.769640896239392e-05,
+      "loss": 2.8958,
+      "step": 32230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.768761024244455e-05,
+      "loss": 2.9598,
+      "step": 32235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.767881089637056e-05,
+      "loss": 2.9229,
+      "step": 32240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.767001092448339e-05,
+      "loss": 2.8666,
+      "step": 32245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.76612103270946e-05,
+      "loss": 2.8911,
+      "step": 32250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.765240910451567e-05,
+      "loss": 2.9265,
+      "step": 32255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.764360725705821e-05,
+      "loss": 2.9471,
+      "step": 32260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.763480478503375e-05,
+      "loss": 2.9925,
+      "step": 32265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.762600168875392e-05,
+      "loss": 2.8862,
+      "step": 32270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.761719796853034e-05,
+      "loss": 2.9615,
+      "step": 32275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.760839362467461e-05,
+      "loss": 2.886,
+      "step": 32280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.759958865749847e-05,
+      "loss": 2.9141,
+      "step": 32285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.759078306731354e-05,
+      "loss": 2.8905,
+      "step": 32290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.758197685443158e-05,
+      "loss": 3.0324,
+      "step": 32295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.757317001916427e-05,
+      "loss": 2.8531,
+      "step": 32300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.75643625618234e-05,
+      "loss": 2.7689,
+      "step": 32305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.755555448272075e-05,
+      "loss": 2.9306,
+      "step": 32310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.754674578216808e-05,
+      "loss": 2.9428,
+      "step": 32315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.753793646047727e-05,
+      "loss": 2.9179,
+      "step": 32320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.75291265179601e-05,
+      "loss": 2.9285,
+      "step": 32325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.752031595492847e-05,
+      "loss": 2.94,
+      "step": 32330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.751150477169426e-05,
+      "loss": 2.9316,
+      "step": 32335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.750269296856936e-05,
+      "loss": 2.8767,
+      "step": 32340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.749388054586571e-05,
+      "loss": 2.8706,
+      "step": 32345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.748506750389528e-05,
+      "loss": 2.9542,
+      "step": 32350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.747625384297002e-05,
+      "loss": 2.8606,
+      "step": 32355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.746743956340192e-05,
+      "loss": 2.8492,
+      "step": 32360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.745862466550303e-05,
+      "loss": 2.9344,
+      "step": 32365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.744980914958536e-05,
+      "loss": 3.0794,
+      "step": 32370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.744099301596097e-05,
+      "loss": 2.8982,
+      "step": 32375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.743217626494199e-05,
+      "loss": 2.9423,
+      "step": 32380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.742335889684044e-05,
+      "loss": 2.982,
+      "step": 32385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.741454091196853e-05,
+      "loss": 2.9148,
+      "step": 32390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.740572231063833e-05,
+      "loss": 2.9413,
+      "step": 32395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.739690309316209e-05,
+      "loss": 2.9175,
+      "step": 32400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.738808325985195e-05,
+      "loss": 2.9519,
+      "step": 32405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.737926281102012e-05,
+      "loss": 2.9339,
+      "step": 32410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.737044174697888e-05,
+      "loss": 2.9116,
+      "step": 32415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.736162006804045e-05,
+      "loss": 2.8696,
+      "step": 32420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.735279777451713e-05,
+      "loss": 2.8641,
+      "step": 32425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.73439748667212e-05,
+      "loss": 2.9802,
+      "step": 32430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.733515134496498e-05,
+      "loss": 2.9866,
+      "step": 32435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.732632720956085e-05,
+      "loss": 2.9002,
+      "step": 32440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.731750246082113e-05,
+      "loss": 2.9179,
+      "step": 32445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.730867709905825e-05,
+      "loss": 2.8927,
+      "step": 32450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.72998511245846e-05,
+      "loss": 2.9039,
+      "step": 32455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.72910245377126e-05,
+      "loss": 2.9146,
+      "step": 32460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.728219733875471e-05,
+      "loss": 2.804,
+      "step": 32465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.727336952802342e-05,
+      "loss": 2.9791,
+      "step": 32470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.726454110583123e-05,
+      "loss": 2.8933,
+      "step": 32475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.725571207249062e-05,
+      "loss": 2.8829,
+      "step": 32480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.724688242831416e-05,
+      "loss": 2.9495,
+      "step": 32485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.723805217361441e-05,
+      "loss": 2.922,
+      "step": 32490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.722922130870392e-05,
+      "loss": 2.8838,
+      "step": 32495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.722038983389535e-05,
+      "loss": 2.9668,
+      "step": 32500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.721155774950129e-05,
+      "loss": 2.7764,
+      "step": 32505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.720272505583438e-05,
+      "loss": 2.9426,
+      "step": 32510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.719389175320731e-05,
+      "loss": 2.9613,
+      "step": 32515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.718505784193277e-05,
+      "loss": 2.8995,
+      "step": 32520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.717622332232345e-05,
+      "loss": 2.8056,
+      "step": 32525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.716738819469212e-05,
+      "loss": 2.7784,
+      "step": 32530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.71585524593515e-05,
+      "loss": 2.8578,
+      "step": 32535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.714971611661438e-05,
+      "loss": 2.8837,
+      "step": 32540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.714087916679355e-05,
+      "loss": 2.8756,
+      "step": 32545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.713204161020184e-05,
+      "loss": 2.9525,
+      "step": 32550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.712320344715211e-05,
+      "loss": 2.9191,
+      "step": 32555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.711436467795718e-05,
+      "loss": 2.9093,
+      "step": 32560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.710552530292997e-05,
+      "loss": 3.0188,
+      "step": 32565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.709668532238335e-05,
+      "loss": 2.895,
+      "step": 32570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.708784473663029e-05,
+      "loss": 2.9457,
+      "step": 32575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.707900354598372e-05,
+      "loss": 2.8796,
+      "step": 32580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.70701617507566e-05,
+      "loss": 2.9575,
+      "step": 32585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.706131935126192e-05,
+      "loss": 2.7826,
+      "step": 32590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.70524763478127e-05,
+      "loss": 2.8611,
+      "step": 32595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.704363274072199e-05,
+      "loss": 2.9416,
+      "step": 32600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.70347885303028e-05,
+      "loss": 3.0155,
+      "step": 32605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.702594371686827e-05,
+      "loss": 2.963,
+      "step": 32610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.701709830073145e-05,
+      "loss": 2.8329,
+      "step": 32615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.700825228220548e-05,
+      "loss": 2.8859,
+      "step": 32620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.699940566160348e-05,
+      "loss": 2.8573,
+      "step": 32625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.699055843923864e-05,
+      "loss": 2.885,
+      "step": 32630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.698171061542413e-05,
+      "loss": 2.8804,
+      "step": 32635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.697286219047313e-05,
+      "loss": 2.9072,
+      "step": 32640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.696401316469891e-05,
+      "loss": 2.9694,
+      "step": 32645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.69551635384147e-05,
+      "loss": 2.9886,
+      "step": 32650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.694631331193375e-05,
+      "loss": 2.8365,
+      "step": 32655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.693746248556939e-05,
+      "loss": 2.8919,
+      "step": 32660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.692861105963488e-05,
+      "loss": 2.8433,
+      "step": 32665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.69197590344436e-05,
+      "loss": 2.9693,
+      "step": 32670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.691090641030886e-05,
+      "loss": 2.8592,
+      "step": 32675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.690205318754408e-05,
+      "loss": 2.8885,
+      "step": 32680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.689319936646263e-05,
+      "loss": 2.8895,
+      "step": 32685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.688434494737789e-05,
+      "loss": 2.9646,
+      "step": 32690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.687548993060337e-05,
+      "loss": 3.0169,
+      "step": 32695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.686663431645249e-05,
+      "loss": 2.936,
+      "step": 32700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.685777810523874e-05,
+      "loss": 2.9581,
+      "step": 32705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.684892129727562e-05,
+      "loss": 2.9107,
+      "step": 32710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.684006389287663e-05,
+      "loss": 3.0403,
+      "step": 32715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.683120589235535e-05,
+      "loss": 2.7647,
+      "step": 32720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.682234729602532e-05,
+      "loss": 3.0002,
+      "step": 32725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.681348810420015e-05,
+      "loss": 2.9779,
+      "step": 32730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.680462831719342e-05,
+      "loss": 2.8854,
+      "step": 32735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.679576793531877e-05,
+      "loss": 2.9397,
+      "step": 32740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.678690695888985e-05,
+      "loss": 2.8783,
+      "step": 32745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.677804538822034e-05,
+      "loss": 2.9206,
+      "step": 32750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.676918322362389e-05,
+      "loss": 2.8687,
+      "step": 32755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.676032046541427e-05,
+      "loss": 3.0146,
+      "step": 32760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.675145711390518e-05,
+      "loss": 2.9154,
+      "step": 32765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.674259316941038e-05,
+      "loss": 2.9631,
+      "step": 32770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.673372863224364e-05,
+      "loss": 2.8811,
+      "step": 32775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.672486350271876e-05,
+      "loss": 2.7772,
+      "step": 32780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.671599778114956e-05,
+      "loss": 2.8903,
+      "step": 32785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.670713146784988e-05,
+      "loss": 2.9667,
+      "step": 32790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.669826456313357e-05,
+      "loss": 2.9297,
+      "step": 32795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.668939706731453e-05,
+      "loss": 2.8244,
+      "step": 32800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.668052898070664e-05,
+      "loss": 2.8446,
+      "step": 32805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.667166030362383e-05,
+      "loss": 3.0025,
+      "step": 32810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.666279103638003e-05,
+      "loss": 3.0123,
+      "step": 32815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.665392117928922e-05,
+      "loss": 2.9781,
+      "step": 32820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.664505073266539e-05,
+      "loss": 2.8886,
+      "step": 32825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.663617969682252e-05,
+      "loss": 2.9753,
+      "step": 32830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.662730807207467e-05,
+      "loss": 3.012,
+      "step": 32835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.661843585873584e-05,
+      "loss": 2.9192,
+      "step": 32840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.660956305712015e-05,
+      "loss": 2.8682,
+      "step": 32845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.660068966754166e-05,
+      "loss": 2.8871,
+      "step": 32850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.659181569031447e-05,
+      "loss": 2.9395,
+      "step": 32855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.658294112575275e-05,
+      "loss": 2.9654,
+      "step": 32860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.657406597417062e-05,
+      "loss": 2.86,
+      "step": 32865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.656519023588226e-05,
+      "loss": 2.9544,
+      "step": 32870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.655631391120183e-05,
+      "loss": 2.8998,
+      "step": 32875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.654743700044362e-05,
+      "loss": 2.9394,
+      "step": 32880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.653855950392178e-05,
+      "loss": 2.8078,
+      "step": 32885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.652968142195062e-05,
+      "loss": 2.9287,
+      "step": 32890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.65208027548444e-05,
+      "loss": 2.8473,
+      "step": 32895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.651192350291742e-05,
+      "loss": 2.9442,
+      "step": 32900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.650304366648399e-05,
+      "loss": 3.0638,
+      "step": 32905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.649416324585844e-05,
+      "loss": 2.8785,
+      "step": 32910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.648528224135515e-05,
+      "loss": 2.888,
+      "step": 32915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.647640065328848e-05,
+      "loss": 2.9588,
+      "step": 32920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.646751848197285e-05,
+      "loss": 2.8099,
+      "step": 32925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.645863572772267e-05,
+      "loss": 2.9116,
+      "step": 32930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.644975239085237e-05,
+      "loss": 2.838,
+      "step": 32935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.64408684716764e-05,
+      "loss": 2.8845,
+      "step": 32940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.64319839705093e-05,
+      "loss": 2.8241,
+      "step": 32945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.642309888766552e-05,
+      "loss": 2.8859,
+      "step": 32950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.64142132234596e-05,
+      "loss": 2.8815,
+      "step": 32955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.640532697820609e-05,
+      "loss": 2.9139,
+      "step": 32960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.639644015221956e-05,
+      "loss": 3.0609,
+      "step": 32965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.638755274581456e-05,
+      "loss": 2.9368,
+      "step": 32970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.637866475930573e-05,
+      "loss": 2.9536,
+      "step": 32975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.636977619300768e-05,
+      "loss": 2.8347,
+      "step": 32980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.636088704723507e-05,
+      "loss": 2.9356,
+      "step": 32985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.635199732230254e-05,
+      "loss": 2.9035,
+      "step": 32990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.634310701852483e-05,
+      "loss": 2.866,
+      "step": 32995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.633421613621661e-05,
+      "loss": 2.9697,
+      "step": 33000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.632532467569261e-05,
+      "loss": 2.8761,
+      "step": 33005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.631643263726758e-05,
+      "loss": 2.9883,
+      "step": 33010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.63075400212563e-05,
+      "loss": 2.9545,
+      "step": 33015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.629864682797356e-05,
+      "loss": 2.9748,
+      "step": 33020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.628975305773415e-05,
+      "loss": 2.9162,
+      "step": 33025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.628085871085291e-05,
+      "loss": 2.9789,
+      "step": 33030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.62719637876447e-05,
+      "loss": 2.8239,
+      "step": 33035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.626306828842441e-05,
+      "loss": 2.8994,
+      "step": 33040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.62541722135069e-05,
+      "loss": 2.8763,
+      "step": 33045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.62452755632071e-05,
+      "loss": 2.9574,
+      "step": 33050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.623637833783993e-05,
+      "loss": 2.8765,
+      "step": 33055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.622748053772035e-05,
+      "loss": 2.9055,
+      "step": 33060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.621858216316333e-05,
+      "loss": 2.9105,
+      "step": 33065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.620968321448389e-05,
+      "loss": 2.7744,
+      "step": 33070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.620078369199701e-05,
+      "loss": 2.8973,
+      "step": 33075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.619188359601773e-05,
+      "loss": 2.8728,
+      "step": 33080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.61829829268611e-05,
+      "loss": 2.8757,
+      "step": 33085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.617408168484223e-05,
+      "loss": 2.9952,
+      "step": 33090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.61651798702762e-05,
+      "loss": 2.8643,
+      "step": 33095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.61562774834781e-05,
+      "loss": 2.8193,
+      "step": 33100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.614737452476309e-05,
+      "loss": 2.8858,
+      "step": 33105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.613847099444632e-05,
+      "loss": 2.9607,
+      "step": 33110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.612956689284297e-05,
+      "loss": 2.9727,
+      "step": 33115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.612066222026824e-05,
+      "loss": 2.9599,
+      "step": 33120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.611175697703733e-05,
+      "loss": 2.8303,
+      "step": 33125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.610285116346551e-05,
+      "loss": 2.9105,
+      "step": 33130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.609394477986799e-05,
+      "loss": 2.9532,
+      "step": 33135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.608503782656008e-05,
+      "loss": 2.8683,
+      "step": 33140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.607613030385707e-05,
+      "loss": 2.9274,
+      "step": 33145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.606722221207429e-05,
+      "loss": 2.8806,
+      "step": 33150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.605831355152708e-05,
+      "loss": 2.8986,
+      "step": 33155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.604940432253076e-05,
+      "loss": 2.9437,
+      "step": 33160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.604049452540074e-05,
+      "loss": 2.9189,
+      "step": 33165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.603158416045241e-05,
+      "loss": 2.8687,
+      "step": 33170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.602267322800119e-05,
+      "loss": 2.9835,
+      "step": 33175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.601376172836254e-05,
+      "loss": 2.9036,
+      "step": 33180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.600484966185186e-05,
+      "loss": 2.9593,
+      "step": 33185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.599593702878469e-05,
+      "loss": 2.9582,
+      "step": 33190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.59870238294765e-05,
+      "loss": 2.8799,
+      "step": 33195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.597811006424282e-05,
+      "loss": 2.883,
+      "step": 33200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.596919573339917e-05,
+      "loss": 2.9452,
+      "step": 33205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.596028083726113e-05,
+      "loss": 2.9211,
+      "step": 33210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.595136537614428e-05,
+      "loss": 2.9056,
+      "step": 33215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.59424493503642e-05,
+      "loss": 2.899,
+      "step": 33220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.593353276023651e-05,
+      "loss": 2.9865,
+      "step": 33225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.592461560607687e-05,
+      "loss": 2.9733,
+      "step": 33230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.59156978882009e-05,
+      "loss": 2.9919,
+      "step": 33235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.590677960692433e-05,
+      "loss": 2.9133,
+      "step": 33240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.589786076256285e-05,
+      "loss": 2.9033,
+      "step": 33245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.588894135543214e-05,
+      "loss": 2.8647,
+      "step": 33250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.588002138584795e-05,
+      "loss": 3.042,
+      "step": 33255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.587110085412607e-05,
+      "loss": 2.9397,
+      "step": 33260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.586217976058225e-05,
+      "loss": 2.9901,
+      "step": 33265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.585325810553229e-05,
+      "loss": 3.0254,
+      "step": 33270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.584433588929201e-05,
+      "loss": 2.959,
+      "step": 33275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.583541311217727e-05,
+      "loss": 2.9642,
+      "step": 33280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.582648977450387e-05,
+      "loss": 2.9824,
+      "step": 33285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.581756587658777e-05,
+      "loss": 2.8296,
+      "step": 33290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.580864141874481e-05,
+      "loss": 2.9056,
+      "step": 33295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.579971640129092e-05,
+      "loss": 2.907,
+      "step": 33300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.579079082454205e-05,
+      "loss": 2.9078,
+      "step": 33305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.578186468881413e-05,
+      "loss": 2.9193,
+      "step": 33310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.577293799442316e-05,
+      "loss": 2.9174,
+      "step": 33315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.576401074168515e-05,
+      "loss": 2.8899,
+      "step": 33320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.575508293091608e-05,
+      "loss": 2.897,
+      "step": 33325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.574615456243199e-05,
+      "loss": 2.927,
+      "step": 33330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.573722563654895e-05,
+      "loss": 2.9267,
+      "step": 33335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.572829615358304e-05,
+      "loss": 2.9236,
+      "step": 33340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.571936611385035e-05,
+      "loss": 3.0178,
+      "step": 33345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.5710435517667e-05,
+      "loss": 2.9537,
+      "step": 33350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.57015043653491e-05,
+      "loss": 2.8798,
+      "step": 33355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.569257265721284e-05,
+      "loss": 2.8915,
+      "step": 33360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.568364039357437e-05,
+      "loss": 2.9866,
+      "step": 33365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.567470757474987e-05,
+      "loss": 2.8136,
+      "step": 33370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.56657742010556e-05,
+      "loss": 3.0596,
+      "step": 33375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.565684027280775e-05,
+      "loss": 2.9237,
+      "step": 33380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.56479057903226e-05,
+      "loss": 2.8509,
+      "step": 33385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.56389707539164e-05,
+      "loss": 2.9415,
+      "step": 33390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.563003516390547e-05,
+      "loss": 2.8366,
+      "step": 33395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.56210990206061e-05,
+      "loss": 2.8816,
+      "step": 33400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.561216232433462e-05,
+      "loss": 2.8864,
+      "step": 33405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.56032250754074e-05,
+      "loss": 2.9719,
+      "step": 33410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.559428727414076e-05,
+      "loss": 2.8255,
+      "step": 33415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.558534892085116e-05,
+      "loss": 2.9403,
+      "step": 33420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.557641001585498e-05,
+      "loss": 2.8738,
+      "step": 33425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.556747055946862e-05,
+      "loss": 2.9348,
+      "step": 33430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.555853055200855e-05,
+      "loss": 3.0729,
+      "step": 33435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.554958999379125e-05,
+      "loss": 2.9366,
+      "step": 33440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.55406488851332e-05,
+      "loss": 2.9222,
+      "step": 33445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.553170722635091e-05,
+      "loss": 2.9459,
+      "step": 33450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.552276501776089e-05,
+      "loss": 2.902,
+      "step": 33455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.551382225967969e-05,
+      "loss": 3.0537,
+      "step": 33460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.55048789524239e-05,
+      "loss": 2.9016,
+      "step": 33465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.549593509631007e-05,
+      "loss": 2.9149,
+      "step": 33470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.548699069165481e-05,
+      "loss": 2.9062,
+      "step": 33475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.547804573877477e-05,
+      "loss": 3.1262,
+      "step": 33480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.546910023798656e-05,
+      "loss": 2.9397,
+      "step": 33485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.54601541896069e-05,
+      "loss": 2.8979,
+      "step": 33490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.545120759395238e-05,
+      "loss": 2.9784,
+      "step": 33495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.544226045133979e-05,
+      "loss": 2.9492,
+      "step": 33500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.543331276208578e-05,
+      "loss": 3.0133,
+      "step": 33505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.542436452650714e-05,
+      "loss": 2.9876,
+      "step": 33510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.541541574492061e-05,
+      "loss": 2.9355,
+      "step": 33515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.540646641764297e-05,
+      "loss": 2.8601,
+      "step": 33520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.539751654499101e-05,
+      "loss": 2.9849,
+      "step": 33525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.538856612728153e-05,
+      "loss": 2.8914,
+      "step": 33530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.537961516483143e-05,
+      "loss": 2.8737,
+      "step": 33535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.53706636579575e-05,
+      "loss": 2.8817,
+      "step": 33540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.536171160697666e-05,
+      "loss": 2.9393,
+      "step": 33545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.535275901220577e-05,
+      "loss": 2.944,
+      "step": 33550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.534380587396175e-05,
+      "loss": 2.8853,
+      "step": 33555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.533485219256156e-05,
+      "loss": 2.9045,
+      "step": 33560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.532589796832211e-05,
+      "loss": 2.9399,
+      "step": 33565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.531694320156042e-05,
+      "loss": 2.9252,
+      "step": 33570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.530798789259343e-05,
+      "loss": 2.9166,
+      "step": 33575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.529903204173817e-05,
+      "loss": 2.8618,
+      "step": 33580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.529007564931169e-05,
+      "loss": 2.9406,
+      "step": 33585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.528111871563101e-05,
+      "loss": 2.8923,
+      "step": 33590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.527216124101321e-05,
+      "loss": 2.914,
+      "step": 33595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.526320322577536e-05,
+      "loss": 2.992,
+      "step": 33600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.52542446702346e-05,
+      "loss": 2.9229,
+      "step": 33605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.524528557470801e-05,
+      "loss": 2.8694,
+      "step": 33610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.523632593951275e-05,
+      "loss": 2.8,
+      "step": 33615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.522736576496602e-05,
+      "loss": 2.8351,
+      "step": 33620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.521840505138494e-05,
+      "loss": 2.8189,
+      "step": 33625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.520944379908676e-05,
+      "loss": 2.8908,
+      "step": 33630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.520048200838866e-05,
+      "loss": 2.8337,
+      "step": 33635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.519151967960792e-05,
+      "loss": 2.9459,
+      "step": 33640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.518255681306175e-05,
+      "loss": 3.0163,
+      "step": 33645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.517359340906748e-05,
+      "loss": 2.9441,
+      "step": 33650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.516462946794237e-05,
+      "loss": 2.911,
+      "step": 33655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.515566499000375e-05,
+      "loss": 2.8933,
+      "step": 33660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.514669997556895e-05,
+      "loss": 2.8931,
+      "step": 33665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.513773442495532e-05,
+      "loss": 2.9274,
+      "step": 33670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.512876833848025e-05,
+      "loss": 2.8755,
+      "step": 33675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.511980171646107e-05,
+      "loss": 2.9004,
+      "step": 33680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.511083455921528e-05,
+      "loss": 2.9161,
+      "step": 33685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.510186686706023e-05,
+      "loss": 2.8918,
+      "step": 33690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.509289864031341e-05,
+      "loss": 2.9768,
+      "step": 33695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.50839298792923e-05,
+      "loss": 2.967,
+      "step": 33700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.507496058431434e-05,
+      "loss": 2.9006,
+      "step": 33705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.506599075569708e-05,
+      "loss": 2.9211,
+      "step": 33710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.505702039375801e-05,
+      "loss": 2.9669,
+      "step": 33715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.504804949881468e-05,
+      "loss": 2.8752,
+      "step": 33720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.503907807118466e-05,
+      "loss": 2.9349,
+      "step": 33725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.503010611118552e-05,
+      "loss": 2.8592,
+      "step": 33730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.502113361913488e-05,
+      "loss": 2.877,
+      "step": 33735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.501216059535033e-05,
+      "loss": 2.9673,
+      "step": 33740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.500318704014952e-05,
+      "loss": 2.8956,
+      "step": 33745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.49942129538501e-05,
+      "loss": 2.8102,
+      "step": 33750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.498523833676976e-05,
+      "loss": 2.8656,
+      "step": 33755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.497626318922619e-05,
+      "loss": 2.9996,
+      "step": 33760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.496728751153707e-05,
+      "loss": 2.8358,
+      "step": 33765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.495831130402017e-05,
+      "loss": 2.8609,
+      "step": 33770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.494933456699319e-05,
+      "loss": 2.8533,
+      "step": 33775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.494035730077397e-05,
+      "loss": 2.9665,
+      "step": 33780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.493137950568024e-05,
+      "loss": 2.8639,
+      "step": 33785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.492240118202983e-05,
+      "loss": 3.0289,
+      "step": 33790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.491342233014055e-05,
+      "loss": 2.8499,
+      "step": 33795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.490444295033025e-05,
+      "loss": 2.8949,
+      "step": 33800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.489546304291679e-05,
+      "loss": 2.8372,
+      "step": 33805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.488648260821805e-05,
+      "loss": 2.9587,
+      "step": 33810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.487750164655194e-05,
+      "loss": 2.9588,
+      "step": 33815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.486852015823636e-05,
+      "loss": 2.8207,
+      "step": 33820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.485953814358924e-05,
+      "loss": 2.8738,
+      "step": 33825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.485055560292858e-05,
+      "loss": 2.8939,
+      "step": 33830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.484157253657231e-05,
+      "loss": 2.922,
+      "step": 33835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.483258894483843e-05,
+      "loss": 2.9053,
+      "step": 33840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.482360482804495e-05,
+      "loss": 2.9809,
+      "step": 33845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.481462018650993e-05,
+      "loss": 2.9973,
+      "step": 33850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.480563502055136e-05,
+      "loss": 2.9943,
+      "step": 33855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.479664933048735e-05,
+      "loss": 2.9667,
+      "step": 33860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.478766311663597e-05,
+      "loss": 2.854,
+      "step": 33865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.477867637931534e-05,
+      "loss": 2.809,
+      "step": 33870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.476968911884356e-05,
+      "loss": 2.8215,
+      "step": 33875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.476070133553878e-05,
+      "loss": 2.9411,
+      "step": 33880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.475171302971916e-05,
+      "loss": 2.7941,
+      "step": 33885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.474272420170286e-05,
+      "loss": 2.846,
+      "step": 33890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.473373485180811e-05,
+      "loss": 2.9948,
+      "step": 33895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.472474498035311e-05,
+      "loss": 2.8994,
+      "step": 33900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.471575458765608e-05,
+      "loss": 2.9027,
+      "step": 33905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.470676367403529e-05,
+      "loss": 2.9343,
+      "step": 33910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.469777223980899e-05,
+      "loss": 2.9839,
+      "step": 33915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.468878028529548e-05,
+      "loss": 2.9223,
+      "step": 33920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.467978781081305e-05,
+      "loss": 2.8508,
+      "step": 33925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.467079481668006e-05,
+      "loss": 2.9609,
+      "step": 33930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.466180130321483e-05,
+      "loss": 2.9005,
+      "step": 33935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.465280727073572e-05,
+      "loss": 2.9194,
+      "step": 33940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.464381271956112e-05,
+      "loss": 2.9461,
+      "step": 33945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.463481765000943e-05,
+      "loss": 2.8836,
+      "step": 33950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.462582206239907e-05,
+      "loss": 2.9491,
+      "step": 33955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.461682595704845e-05,
+      "loss": 2.9803,
+      "step": 33960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.460782933427603e-05,
+      "loss": 2.8775,
+      "step": 33965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.459883219440033e-05,
+      "loss": 2.9151,
+      "step": 33970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.458983453773976e-05,
+      "loss": 2.9232,
+      "step": 33975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.45808363646129e-05,
+      "loss": 2.8113,
+      "step": 33980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.457183767533824e-05,
+      "loss": 2.928,
+      "step": 33985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.456283847023432e-05,
+      "loss": 3.0021,
+      "step": 33990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.455383874961973e-05,
+      "loss": 2.8836,
+      "step": 33995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.454483851381303e-05,
+      "loss": 2.8705,
+      "step": 34000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.453583776313285e-05,
+      "loss": 2.8939,
+      "step": 34005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.452683649789776e-05,
+      "loss": 2.9817,
+      "step": 34010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.451783471842642e-05,
+      "loss": 2.7749,
+      "step": 34015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.450883242503751e-05,
+      "loss": 2.9585,
+      "step": 34020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.449982961804965e-05,
+      "loss": 2.89,
+      "step": 34025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.449082629778158e-05,
+      "loss": 2.8725,
+      "step": 34030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.448182246455198e-05,
+      "loss": 2.9019,
+      "step": 34035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.44728181186796e-05,
+      "loss": 2.8782,
+      "step": 34040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.446381326048316e-05,
+      "loss": 2.8713,
+      "step": 34045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.445480789028144e-05,
+      "loss": 2.8629,
+      "step": 34050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.44458020083932e-05,
+      "loss": 2.8851,
+      "step": 34055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.44367956151373e-05,
+      "loss": 2.9035,
+      "step": 34060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.442778871083246e-05,
+      "loss": 2.9468,
+      "step": 34065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.44187812957976e-05,
+      "loss": 2.9034,
+      "step": 34070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.440977337035152e-05,
+      "loss": 2.8465,
+      "step": 34075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.440076493481313e-05,
+      "loss": 2.8978,
+      "step": 34080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.439175598950132e-05,
+      "loss": 2.8756,
+      "step": 34085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.438274653473497e-05,
+      "loss": 2.9166,
+      "step": 34090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.437373657083303e-05,
+      "loss": 2.8865,
+      "step": 34095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.436472609811441e-05,
+      "loss": 2.9185,
+      "step": 34100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.435571511689811e-05,
+      "loss": 2.9561,
+      "step": 34105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.434670362750312e-05,
+      "loss": 2.9314,
+      "step": 34110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.433769163024838e-05,
+      "loss": 2.9296,
+      "step": 34115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.432867912545297e-05,
+      "loss": 2.9771,
+      "step": 34120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.431966611343586e-05,
+      "loss": 2.8283,
+      "step": 34125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.431065259451616e-05,
+      "loss": 2.9681,
+      "step": 34130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.430163856901293e-05,
+      "loss": 2.9511,
+      "step": 34135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.429262403724522e-05,
+      "loss": 2.9594,
+      "step": 34140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.428360899953218e-05,
+      "loss": 2.8967,
+      "step": 34145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.42745934561929e-05,
+      "loss": 2.9442,
+      "step": 34150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.426557740754655e-05,
+      "loss": 2.8904,
+      "step": 34155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.425656085391227e-05,
+      "loss": 2.8658,
+      "step": 34160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.424754379560923e-05,
+      "loss": 2.9048,
+      "step": 34165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.423852623295666e-05,
+      "loss": 2.8787,
+      "step": 34170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.422950816627373e-05,
+      "loss": 2.995,
+      "step": 34175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.422048959587971e-05,
+      "loss": 2.8884,
+      "step": 34180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.421147052209383e-05,
+      "loss": 2.891,
+      "step": 34185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.420245094523536e-05,
+      "loss": 2.9486,
+      "step": 34190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.419343086562359e-05,
+      "loss": 2.7639,
+      "step": 34195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.41844102835778e-05,
+      "loss": 2.9172,
+      "step": 34200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.417538919941732e-05,
+      "loss": 2.8983,
+      "step": 34205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.41663676134615e-05,
+      "loss": 2.945,
+      "step": 34210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.41573455260297e-05,
+      "loss": 2.9067,
+      "step": 34215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.414832293744127e-05,
+      "loss": 2.8075,
+      "step": 34220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.413929984801563e-05,
+      "loss": 2.9352,
+      "step": 34225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.413027625807215e-05,
+      "loss": 2.9634,
+      "step": 34230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.41212521679303e-05,
+      "loss": 2.9096,
+      "step": 34235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.411222757790949e-05,
+      "loss": 2.8637,
+      "step": 34240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.41032024883292e-05,
+      "loss": 3.0012,
+      "step": 34245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.40941768995089e-05,
+      "loss": 3.0096,
+      "step": 34250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.40851508117681e-05,
+      "loss": 2.9446,
+      "step": 34255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.40761242254263e-05,
+      "loss": 3.0263,
+      "step": 34260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.406709714080305e-05,
+      "loss": 2.9731,
+      "step": 34265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.405806955821787e-05,
+      "loss": 2.804,
+      "step": 34270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.404904147799036e-05,
+      "loss": 2.9506,
+      "step": 34275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.404001290044008e-05,
+      "loss": 2.8657,
+      "step": 34280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.403098382588667e-05,
+      "loss": 2.9294,
+      "step": 34285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.40219542546497e-05,
+      "loss": 2.9308,
+      "step": 34290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.401292418704885e-05,
+      "loss": 2.8764,
+      "step": 34295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.400389362340375e-05,
+      "loss": 2.9378,
+      "step": 34300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.399486256403411e-05,
+      "loss": 2.7885,
+      "step": 34305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.398583100925958e-05,
+      "loss": 2.9973,
+      "step": 34310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.397679895939987e-05,
+      "loss": 2.9736,
+      "step": 34315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.396776641477474e-05,
+      "loss": 2.9365,
+      "step": 34320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.395873337570389e-05,
+      "loss": 2.8599,
+      "step": 34325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.394969984250713e-05,
+      "loss": 2.899,
+      "step": 34330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.39406658155042e-05,
+      "loss": 2.8473,
+      "step": 34335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.393163129501492e-05,
+      "loss": 2.8695,
+      "step": 34340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.392259628135908e-05,
+      "loss": 2.9352,
+      "step": 34345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.39135607748565e-05,
+      "loss": 2.764,
+      "step": 34350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.390452477582709e-05,
+      "loss": 2.9572,
+      "step": 34355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.389548828459067e-05,
+      "loss": 2.9367,
+      "step": 34360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.388645130146714e-05,
+      "loss": 2.9953,
+      "step": 34365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.387741382677634e-05,
+      "loss": 2.9857,
+      "step": 34370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.38683758608383e-05,
+      "loss": 2.9994,
+      "step": 34375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.385933740397286e-05,
+      "loss": 2.9331,
+      "step": 34380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.385029845650001e-05,
+      "loss": 2.9573,
+      "step": 34385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.384125901873973e-05,
+      "loss": 2.9284,
+      "step": 34390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.383221909101198e-05,
+      "loss": 2.8575,
+      "step": 34395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.38231786736368e-05,
+      "loss": 2.8623,
+      "step": 34400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.381413776693418e-05,
+      "loss": 2.9665,
+      "step": 34405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.380509637122416e-05,
+      "loss": 2.9554,
+      "step": 34410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.379605448682684e-05,
+      "loss": 2.9296,
+      "step": 34415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.378701211406223e-05,
+      "loss": 2.8741,
+      "step": 34420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.377796925325046e-05,
+      "loss": 2.867,
+      "step": 34425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.376892590471166e-05,
+      "loss": 2.918,
+      "step": 34430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.375988206876591e-05,
+      "loss": 2.9514,
+      "step": 34435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.375083774573336e-05,
+      "loss": 2.9398,
+      "step": 34440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.374179293593421e-05,
+      "loss": 2.9655,
+      "step": 34445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.37327476396886e-05,
+      "loss": 2.9623,
+      "step": 34450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.37237018573167e-05,
+      "loss": 2.9097,
+      "step": 34455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.371465558913881e-05,
+      "loss": 2.8478,
+      "step": 34460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.370560883547508e-05,
+      "loss": 2.9283,
+      "step": 34465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.369656159664578e-05,
+      "loss": 2.9808,
+      "step": 34470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.368751387297118e-05,
+      "loss": 2.8685,
+      "step": 34475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.367846566477156e-05,
+      "loss": 2.9297,
+      "step": 34480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.366941697236723e-05,
+      "loss": 3.0426,
+      "step": 34485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.366036779607848e-05,
+      "loss": 2.9303,
+      "step": 34490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.365131813622565e-05,
+      "loss": 2.8005,
+      "step": 34495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.364226799312908e-05,
+      "loss": 2.9657,
+      "step": 34500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.363321736710918e-05,
+      "loss": 2.8318,
+      "step": 34505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.362416625848627e-05,
+      "loss": 2.9293,
+      "step": 34510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.36151146675808e-05,
+      "loss": 2.8235,
+      "step": 34515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.360606259471317e-05,
+      "loss": 2.8843,
+      "step": 34520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.359701004020381e-05,
+      "loss": 2.8378,
+      "step": 34525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.35879570043732e-05,
+      "loss": 2.8136,
+      "step": 34530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.357890348754175e-05,
+      "loss": 2.8719,
+      "step": 34535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.356984949003e-05,
+      "loss": 2.8965,
+      "step": 34540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.356079501215844e-05,
+      "loss": 2.9103,
+      "step": 34545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.355174005424758e-05,
+      "loss": 2.9528,
+      "step": 34550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.354268461661796e-05,
+      "loss": 2.7572,
+      "step": 34555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.353362869959014e-05,
+      "loss": 2.8321,
+      "step": 34560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.35245723034847e-05,
+      "loss": 2.9173,
+      "step": 34565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.351551542862218e-05,
+      "loss": 2.9457,
+      "step": 34570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.350645807532325e-05,
+      "loss": 2.9283,
+      "step": 34575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.349740024390848e-05,
+      "loss": 2.9549,
+      "step": 34580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.348834193469855e-05,
+      "loss": 2.8301,
+      "step": 34585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.347928314801408e-05,
+      "loss": 2.9276,
+      "step": 34590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.347022388417577e-05,
+      "loss": 2.8998,
+      "step": 34595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.34611641435043e-05,
+      "loss": 2.9528,
+      "step": 34600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.345210392632036e-05,
+      "loss": 2.8908,
+      "step": 34605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.34430432329447e-05,
+      "loss": 2.9371,
+      "step": 34610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.343398206369802e-05,
+      "loss": 2.96,
+      "step": 34615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.342492041890114e-05,
+      "loss": 3.0633,
+      "step": 34620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.341585829887478e-05,
+      "loss": 2.8795,
+      "step": 34625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.340679570393973e-05,
+      "loss": 2.8969,
+      "step": 34630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.339773263441684e-05,
+      "loss": 2.9352,
+      "step": 34635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.33886690906269e-05,
+      "loss": 2.9527,
+      "step": 34640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.337960507289076e-05,
+      "loss": 2.8654,
+      "step": 34645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.337054058152929e-05,
+      "loss": 2.9194,
+      "step": 34650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.336147561686334e-05,
+      "loss": 2.9443,
+      "step": 34655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.335241017921381e-05,
+      "loss": 2.8805,
+      "step": 34660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.33433442689016e-05,
+      "loss": 2.8912,
+      "step": 34665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.333427788624766e-05,
+      "loss": 2.7868,
+      "step": 34670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.332521103157292e-05,
+      "loss": 2.9243,
+      "step": 34675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.331614370519834e-05,
+      "loss": 2.9079,
+      "step": 34680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.330707590744486e-05,
+      "loss": 2.9325,
+      "step": 34685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.329800763863353e-05,
+      "loss": 2.903,
+      "step": 34690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.328893889908533e-05,
+      "loss": 2.8985,
+      "step": 34695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.327986968912126e-05,
+      "loss": 2.8961,
+      "step": 34700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.327080000906241e-05,
+      "loss": 2.9515,
+      "step": 34705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.32617298592298e-05,
+      "loss": 3.1082,
+      "step": 34710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.325265923994452e-05,
+      "loss": 2.9163,
+      "step": 34715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.324358815152766e-05,
+      "loss": 2.8831,
+      "step": 34720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.323451659430033e-05,
+      "loss": 2.9545,
+      "step": 34725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.322544456858365e-05,
+      "loss": 2.9,
+      "step": 34730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.321637207469877e-05,
+      "loss": 2.8834,
+      "step": 34735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.320729911296684e-05,
+      "loss": 2.9965,
+      "step": 34740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.319822568370903e-05,
+      "loss": 2.9824,
+      "step": 34745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.318915178724653e-05,
+      "loss": 2.9255,
+      "step": 34750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.318007742390056e-05,
+      "loss": 2.7977,
+      "step": 34755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.317100259399233e-05,
+      "loss": 2.9035,
+      "step": 34760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.31619272978431e-05,
+      "loss": 2.8852,
+      "step": 34765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.31528515357741e-05,
+      "loss": 2.9529,
+      "step": 34770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.314377530810663e-05,
+      "loss": 2.9199,
+      "step": 34775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.313469861516197e-05,
+      "loss": 2.8136,
+      "step": 34780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.312562145726141e-05,
+      "loss": 2.9379,
+      "step": 34785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.31165438347263e-05,
+      "loss": 2.9497,
+      "step": 34790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.310746574787796e-05,
+      "loss": 2.9325,
+      "step": 34795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.309838719703775e-05,
+      "loss": 2.8773,
+      "step": 34800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.308930818252706e-05,
+      "loss": 2.9109,
+      "step": 34805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.308022870466725e-05,
+      "loss": 2.8345,
+      "step": 34810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.307114876377972e-05,
+      "loss": 2.934,
+      "step": 34815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.306206836018593e-05,
+      "loss": 2.9398,
+      "step": 34820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.305298749420729e-05,
+      "loss": 2.9252,
+      "step": 34825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.304390616616527e-05,
+      "loss": 2.8983,
+      "step": 34830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.30348243763813e-05,
+      "loss": 3.0828,
+      "step": 34835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.302574212517692e-05,
+      "loss": 2.9109,
+      "step": 34840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.301665941287362e-05,
+      "loss": 2.8901,
+      "step": 34845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.30075762397929e-05,
+      "loss": 2.8145,
+      "step": 34850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.29984926062563e-05,
+      "loss": 2.8307,
+      "step": 34855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.298940851258539e-05,
+      "loss": 2.9926,
+      "step": 34860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.298032395910172e-05,
+      "loss": 2.9533,
+      "step": 34865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.297123894612686e-05,
+      "loss": 3.0023,
+      "step": 34870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.296215347398245e-05,
+      "loss": 2.9609,
+      "step": 34875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.29530675429901e-05,
+      "loss": 2.9064,
+      "step": 34880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.294398115347141e-05,
+      "loss": 2.8641,
+      "step": 34885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.293489430574807e-05,
+      "loss": 2.935,
+      "step": 34890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.292580700014172e-05,
+      "loss": 2.9441,
+      "step": 34895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.291671923697405e-05,
+      "loss": 2.9035,
+      "step": 34900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.290763101656674e-05,
+      "loss": 2.929,
+      "step": 34905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.289854233924156e-05,
+      "loss": 2.9035,
+      "step": 34910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.288945320532016e-05,
+      "loss": 2.8942,
+      "step": 34915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.288036361512435e-05,
+      "loss": 2.9464,
+      "step": 34920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.287127356897588e-05,
+      "loss": 2.902,
+      "step": 34925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.28621830671965e-05,
+      "loss": 3.0157,
+      "step": 34930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.285309211010804e-05,
+      "loss": 2.9301,
+      "step": 34935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.284400069803229e-05,
+      "loss": 2.9237,
+      "step": 34940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.283490883129108e-05,
+      "loss": 2.8346,
+      "step": 34945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.282581651020629e-05,
+      "loss": 2.7809,
+      "step": 34950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.281672373509971e-05,
+      "loss": 2.9296,
+      "step": 34955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.280763050629328e-05,
+      "loss": 2.9476,
+      "step": 34960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.279853682410884e-05,
+      "loss": 2.7784,
+      "step": 34965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.278944268886835e-05,
+      "loss": 2.9017,
+      "step": 34970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.278034810089368e-05,
+      "loss": 2.9364,
+      "step": 34975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.277125306050682e-05,
+      "loss": 2.7928,
+      "step": 34980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.276215756802969e-05,
+      "loss": 2.8446,
+      "step": 34985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.275306162378427e-05,
+      "loss": 2.8968,
+      "step": 34990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.274396522809257e-05,
+      "loss": 2.9727,
+      "step": 34995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.273486838127655e-05,
+      "loss": 2.7918,
+      "step": 35000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.272577108365828e-05,
+      "loss": 2.8971,
+      "step": 35005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.271667333555977e-05,
+      "loss": 2.9587,
+      "step": 35010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.270757513730304e-05,
+      "loss": 2.9569,
+      "step": 35015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.269847648921022e-05,
+      "loss": 2.9272,
+      "step": 35020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.268937739160336e-05,
+      "loss": 2.9587,
+      "step": 35025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.268027784480455e-05,
+      "loss": 2.9501,
+      "step": 35030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.267117784913594e-05,
+      "loss": 2.9382,
+      "step": 35035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.266207740491962e-05,
+      "loss": 2.9742,
+      "step": 35040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.265297651247778e-05,
+      "loss": 2.978,
+      "step": 35045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.264387517213253e-05,
+      "loss": 2.9018,
+      "step": 35050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.263477338420612e-05,
+      "loss": 2.9115,
+      "step": 35055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.262567114902065e-05,
+      "loss": 2.915,
+      "step": 35060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.261656846689843e-05,
+      "loss": 2.9436,
+      "step": 35065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.26074653381616e-05,
+      "loss": 2.8954,
+      "step": 35070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.259836176313247e-05,
+      "loss": 2.8579,
+      "step": 35075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.258925774213328e-05,
+      "loss": 2.9587,
+      "step": 35080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.258015327548627e-05,
+      "loss": 2.904,
+      "step": 35085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.257104836351378e-05,
+      "loss": 2.8612,
+      "step": 35090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.256194300653806e-05,
+      "loss": 2.8465,
+      "step": 35095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.255283720488148e-05,
+      "loss": 2.9208,
+      "step": 35100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.254373095886634e-05,
+      "loss": 2.9465,
+      "step": 35105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.253462426881501e-05,
+      "loss": 2.8483,
+      "step": 35110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.252551713504986e-05,
+      "loss": 2.9105,
+      "step": 35115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.251640955789328e-05,
+      "loss": 2.859,
+      "step": 35120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.250730153766767e-05,
+      "loss": 2.8989,
+      "step": 35125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.249819307469542e-05,
+      "loss": 2.9258,
+      "step": 35130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.248908416929898e-05,
+      "loss": 2.8248,
+      "step": 35135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.247997482180078e-05,
+      "loss": 2.7602,
+      "step": 35140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.24708650325233e-05,
+      "loss": 2.9486,
+      "step": 35145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.246175480178903e-05,
+      "loss": 2.908,
+      "step": 35150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.245264412992044e-05,
+      "loss": 2.8762,
+      "step": 35155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.244353301724004e-05,
+      "loss": 2.9286,
+      "step": 35160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.243442146407037e-05,
+      "loss": 2.9988,
+      "step": 35165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.242530947073398e-05,
+      "loss": 2.8292,
+      "step": 35170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.241619703755337e-05,
+      "loss": 2.9993,
+      "step": 35175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.240708416485116e-05,
+      "loss": 2.9701,
+      "step": 35180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.239797085294993e-05,
+      "loss": 2.9817,
+      "step": 35185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.238885710217228e-05,
+      "loss": 2.8559,
+      "step": 35190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.237974291284084e-05,
+      "loss": 2.8247,
+      "step": 35195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.237062828527822e-05,
+      "loss": 2.8326,
+      "step": 35200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.236151321980706e-05,
+      "loss": 2.7624,
+      "step": 35205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.235239771675007e-05,
+      "loss": 2.9101,
+      "step": 35210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.234328177642989e-05,
+      "loss": 2.9011,
+      "step": 35215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.233416539916924e-05,
+      "loss": 2.8961,
+      "step": 35220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.232504858529082e-05,
+      "loss": 2.879,
+      "step": 35225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.231593133511736e-05,
+      "loss": 2.9384,
+      "step": 35230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.23068136489716e-05,
+      "loss": 3.0096,
+      "step": 35235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.229769552717629e-05,
+      "loss": 2.9416,
+      "step": 35240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.228857697005421e-05,
+      "loss": 3.0123,
+      "step": 35245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.227945797792814e-05,
+      "loss": 3.004,
+      "step": 35250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.227033855112092e-05,
+      "loss": 2.8708,
+      "step": 35255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.226121868995531e-05,
+      "loss": 2.836,
+      "step": 35260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.22520983947542e-05,
+      "loss": 2.8946,
+      "step": 35265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.224297766584039e-05,
+      "loss": 2.9542,
+      "step": 35270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.223385650353681e-05,
+      "loss": 2.933,
+      "step": 35275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.222473490816627e-05,
+      "loss": 2.8622,
+      "step": 35280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.22156128800517e-05,
+      "loss": 2.9692,
+      "step": 35285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.220649041951601e-05,
+      "loss": 2.9771,
+      "step": 35290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.219736752688214e-05,
+      "loss": 2.8656,
+      "step": 35295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.2188244202473e-05,
+      "loss": 3.0075,
+      "step": 35300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.217912044661158e-05,
+      "loss": 2.7848,
+      "step": 35305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.216999625962081e-05,
+      "loss": 2.9148,
+      "step": 35310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.216087164182372e-05,
+      "loss": 2.8263,
+      "step": 35315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.215174659354328e-05,
+      "loss": 2.8802,
+      "step": 35320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.214262111510254e-05,
+      "loss": 2.9858,
+      "step": 35325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.21334952068245e-05,
+      "loss": 2.9376,
+      "step": 35330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.212436886903226e-05,
+      "loss": 2.8747,
+      "step": 35335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.211524210204883e-05,
+      "loss": 2.904,
+      "step": 35340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.210611490619731e-05,
+      "loss": 2.9789,
+      "step": 35345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.209698728180078e-05,
+      "loss": 2.9554,
+      "step": 35350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.208785922918238e-05,
+      "loss": 2.8725,
+      "step": 35355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.20787307486652e-05,
+      "loss": 2.8908,
+      "step": 35360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.20696018405724e-05,
+      "loss": 2.9053,
+      "step": 35365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.206047250522715e-05,
+      "loss": 2.8853,
+      "step": 35370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.205134274295258e-05,
+      "loss": 2.9124,
+      "step": 35375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.20422125540719e-05,
+      "loss": 2.9585,
+      "step": 35380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.203308193890832e-05,
+      "loss": 2.8858,
+      "step": 35385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.202395089778502e-05,
+      "loss": 2.9453,
+      "step": 35390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.201481943102526e-05,
+      "loss": 2.9896,
+      "step": 35395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.200568753895227e-05,
+      "loss": 2.9391,
+      "step": 35400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.199655522188933e-05,
+      "loss": 2.8348,
+      "step": 35405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.198742248015969e-05,
+      "loss": 2.9765,
+      "step": 35410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.197828931408665e-05,
+      "loss": 2.9382,
+      "step": 35415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.196915572399352e-05,
+      "loss": 2.9073,
+      "step": 35420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.196002171020362e-05,
+      "loss": 2.8542,
+      "step": 35425
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.195088727304029e-05,
+      "loss": 2.8927,
+      "step": 35430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.194175241282686e-05,
+      "loss": 2.8267,
+      "step": 35435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.193261712988673e-05,
+      "loss": 2.9968,
+      "step": 35440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.192348142454324e-05,
+      "loss": 2.9433,
+      "step": 35445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.19143452971198e-05,
+      "loss": 2.9001,
+      "step": 35450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.190520874793983e-05,
+      "loss": 3.0494,
+      "step": 35455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.189607177732676e-05,
+      "loss": 2.8958,
+      "step": 35460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.188693438560403e-05,
+      "loss": 2.9008,
+      "step": 35465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.187779657309506e-05,
+      "loss": 2.8193,
+      "step": 35470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.186865834012337e-05,
+      "loss": 2.8817,
+      "step": 35475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.18595196870124e-05,
+      "loss": 2.9017,
+      "step": 35480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.185038061408569e-05,
+      "loss": 2.8598,
+      "step": 35485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.184124112166672e-05,
+      "loss": 2.8339,
+      "step": 35490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.183210121007903e-05,
+      "loss": 2.9158,
+      "step": 35495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.182296087964617e-05,
+      "loss": 2.9442,
+      "step": 35500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.18138201306917e-05,
+      "loss": 2.8842,
+      "step": 35505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.180467896353919e-05,
+      "loss": 2.962,
+      "step": 35510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.179553737851223e-05,
+      "loss": 2.9473,
+      "step": 35515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.178639537593442e-05,
+      "loss": 2.8989,
+      "step": 35520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.177725295612938e-05,
+      "loss": 2.9536,
+      "step": 35525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.176811011942075e-05,
+      "loss": 2.8651,
+      "step": 35530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.175896686613219e-05,
+      "loss": 2.9091,
+      "step": 35535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.17498231965873e-05,
+      "loss": 2.9515,
+      "step": 35540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.174067911110985e-05,
+      "loss": 2.809,
+      "step": 35545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.173153461002344e-05,
+      "loss": 2.9239,
+      "step": 35550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.172238969365185e-05,
+      "loss": 2.9895,
+      "step": 35555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.171324436231876e-05,
+      "loss": 2.9937,
+      "step": 35560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.170409861634791e-05,
+      "loss": 2.8631,
+      "step": 35565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.169495245606307e-05,
+      "loss": 2.917,
+      "step": 35570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1685805881788e-05,
+      "loss": 2.8904,
+      "step": 35575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.167665889384646e-05,
+      "loss": 2.871,
+      "step": 35580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.166751149256227e-05,
+      "loss": 2.8615,
+      "step": 35585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.165836367825922e-05,
+      "loss": 2.974,
+      "step": 35590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.164921545126114e-05,
+      "loss": 2.8881,
+      "step": 35595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.164006681189187e-05,
+      "loss": 2.9251,
+      "step": 35600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.163091776047527e-05,
+      "loss": 2.787,
+      "step": 35605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.162176829733519e-05,
+      "loss": 2.9561,
+      "step": 35610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.161261842279555e-05,
+      "loss": 2.8519,
+      "step": 35615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.16034681371802e-05,
+      "loss": 2.9613,
+      "step": 35620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.159431744081308e-05,
+      "loss": 2.8569,
+      "step": 35625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.158516633401809e-05,
+      "loss": 2.8901,
+      "step": 35630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.15760148171192e-05,
+      "loss": 2.8713,
+      "step": 35635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.156686289044037e-05,
+      "loss": 2.9174,
+      "step": 35640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.155771055430553e-05,
+      "loss": 2.9254,
+      "step": 35645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.15485578090387e-05,
+      "loss": 2.8204,
+      "step": 35650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.153940465496386e-05,
+      "loss": 2.9748,
+      "step": 35655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.153025109240504e-05,
+      "loss": 2.8858,
+      "step": 35660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.152109712168625e-05,
+      "loss": 2.9126,
+      "step": 35665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.151194274313153e-05,
+      "loss": 2.9721,
+      "step": 35670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.150278795706495e-05,
+      "loss": 2.9127,
+      "step": 35675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.149363276381058e-05,
+      "loss": 2.8781,
+      "step": 35680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.148447716369249e-05,
+      "loss": 2.8995,
+      "step": 35685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.14753211570348e-05,
+      "loss": 2.8649,
+      "step": 35690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.146616474416159e-05,
+      "loss": 3.0396,
+      "step": 35695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.145700792539704e-05,
+      "loss": 2.9092,
+      "step": 35700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.144785070106525e-05,
+      "loss": 2.9678,
+      "step": 35705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.143869307149039e-05,
+      "loss": 3.0147,
+      "step": 35710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.142953503699664e-05,
+      "loss": 3.0115,
+      "step": 35715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.142037659790818e-05,
+      "loss": 2.9343,
+      "step": 35720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.141121775454921e-05,
+      "loss": 2.9534,
+      "step": 35725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.140205850724394e-05,
+      "loss": 3.0012,
+      "step": 35730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.13928988563166e-05,
+      "loss": 2.9616,
+      "step": 35735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.138373880209144e-05,
+      "loss": 2.8834,
+      "step": 35740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.137457834489271e-05,
+      "loss": 2.9481,
+      "step": 35745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.136541748504468e-05,
+      "loss": 2.9699,
+      "step": 35750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.135625622287164e-05,
+      "loss": 2.8175,
+      "step": 35755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.13470945586979e-05,
+      "loss": 2.9846,
+      "step": 35760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.133793249284776e-05,
+      "loss": 2.9044,
+      "step": 35765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.132877002564555e-05,
+      "loss": 2.9547,
+      "step": 35770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.131960715741561e-05,
+      "loss": 2.955,
+      "step": 35775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.131044388848232e-05,
+      "loss": 2.9879,
+      "step": 35780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.130128021917001e-05,
+      "loss": 2.8742,
+      "step": 35785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.129211614980311e-05,
+      "loss": 2.8247,
+      "step": 35790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.128295168070597e-05,
+      "loss": 2.9735,
+      "step": 35795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.127378681220305e-05,
+      "loss": 2.9008,
+      "step": 35800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.126462154461875e-05,
+      "loss": 2.8441,
+      "step": 35805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.125545587827752e-05,
+      "loss": 2.9692,
+      "step": 35810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.124628981350382e-05,
+      "loss": 2.8261,
+      "step": 35815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.123712335062211e-05,
+      "loss": 2.9706,
+      "step": 35820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.12279564899569e-05,
+      "loss": 2.9643,
+      "step": 35825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.121878923183265e-05,
+      "loss": 2.9409,
+      "step": 35830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.120962157657389e-05,
+      "loss": 2.9722,
+      "step": 35835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.120045352450518e-05,
+      "loss": 2.9089,
+      "step": 35840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.119128507595098e-05,
+      "loss": 2.8327,
+      "step": 35845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.118211623123592e-05,
+      "loss": 2.9421,
+      "step": 35850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.117294699068454e-05,
+      "loss": 2.9118,
+      "step": 35855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.116377735462141e-05,
+      "loss": 2.8417,
+      "step": 35860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.115460732337114e-05,
+      "loss": 3.0031,
+      "step": 35865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.114543689725836e-05,
+      "loss": 2.8887,
+      "step": 35870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.113626607660765e-05,
+      "loss": 2.909,
+      "step": 35875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.112709486174368e-05,
+      "loss": 2.9342,
+      "step": 35880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.11179232529911e-05,
+      "loss": 2.9224,
+      "step": 35885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.110875125067457e-05,
+      "loss": 2.9006,
+      "step": 35890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.109957885511876e-05,
+      "loss": 2.9288,
+      "step": 35895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.109040606664838e-05,
+      "loss": 3.0638,
+      "step": 35900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.108123288558814e-05,
+      "loss": 2.8758,
+      "step": 35905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.107205931226276e-05,
+      "loss": 2.912,
+      "step": 35910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.106288534699695e-05,
+      "loss": 2.8976,
+      "step": 35915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.10537109901155e-05,
+      "loss": 2.9587,
+      "step": 35920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.104453624194314e-05,
+      "loss": 2.805,
+      "step": 35925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.103536110280468e-05,
+      "loss": 2.9512,
+      "step": 35930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.102618557302488e-05,
+      "loss": 2.9392,
+      "step": 35935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1017009652928544e-05,
+      "loss": 2.9841,
+      "step": 35940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1007833342840535e-05,
+      "loss": 2.9687,
+      "step": 35945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.099865664308563e-05,
+      "loss": 2.8055,
+      "step": 35950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.098947955398872e-05,
+      "loss": 2.9647,
+      "step": 35955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.098030207587463e-05,
+      "loss": 2.9269,
+      "step": 35960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0971124209068266e-05,
+      "loss": 2.8707,
+      "step": 35965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.096194595389448e-05,
+      "loss": 2.9195,
+      "step": 35970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.095276731067821e-05,
+      "loss": 2.886,
+      "step": 35975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.094358827974436e-05,
+      "loss": 2.8722,
+      "step": 35980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0934408861417836e-05,
+      "loss": 2.9694,
+      "step": 35985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0925229056023616e-05,
+      "loss": 2.8724,
+      "step": 35990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.091604886388662e-05,
+      "loss": 2.8529,
+      "step": 35995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.090686828533184e-05,
+      "loss": 2.9825,
+      "step": 36000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.089768732068426e-05,
+      "loss": 2.929,
+      "step": 36005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.088850597026888e-05,
+      "loss": 2.9083,
+      "step": 36010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.087932423441071e-05,
+      "loss": 2.9542,
+      "step": 36015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.087014211343476e-05,
+      "loss": 2.8646,
+      "step": 36020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.086095960766608e-05,
+      "loss": 3.0433,
+      "step": 36025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0851776717429724e-05,
+      "loss": 2.866,
+      "step": 36030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0842593443050754e-05,
+      "loss": 2.9735,
+      "step": 36035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.083340978485425e-05,
+      "loss": 2.85,
+      "step": 36040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.082422574316529e-05,
+      "loss": 2.9199,
+      "step": 36045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.081504131830901e-05,
+      "loss": 2.9769,
+      "step": 36050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.08058565106105e-05,
+      "loss": 2.884,
+      "step": 36055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.079667132039493e-05,
+      "loss": 2.8549,
+      "step": 36060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0787485747987405e-05,
+      "loss": 2.8875,
+      "step": 36065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.077829979371312e-05,
+      "loss": 2.899,
+      "step": 36070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.076911345789722e-05,
+      "loss": 2.8531,
+      "step": 36075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0759926740864915e-05,
+      "loss": 2.8944,
+      "step": 36080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.07507396429414e-05,
+      "loss": 2.9609,
+      "step": 36085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0741552164451884e-05,
+      "loss": 2.817,
+      "step": 36090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.073236430572161e-05,
+      "loss": 2.956,
+      "step": 36095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.072317606707578e-05,
+      "loss": 2.7936,
+      "step": 36100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0713987448839716e-05,
+      "loss": 2.8659,
+      "step": 36105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0704798451338616e-05,
+      "loss": 2.982,
+      "step": 36110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.069560907489781e-05,
+      "loss": 2.9742,
+      "step": 36115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.068641931984257e-05,
+      "loss": 2.8944,
+      "step": 36120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0677229186498205e-05,
+      "loss": 2.9703,
+      "step": 36125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.066803867519007e-05,
+      "loss": 2.8894,
+      "step": 36130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0658847786243444e-05,
+      "loss": 2.7907,
+      "step": 36135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.064965651998373e-05,
+      "loss": 2.897,
+      "step": 36140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.064046487673625e-05,
+      "loss": 2.9486,
+      "step": 36145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0631272856826397e-05,
+      "loss": 2.9583,
+      "step": 36150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0622080460579554e-05,
+      "loss": 2.9219,
+      "step": 36155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.061288768832112e-05,
+      "loss": 3.1646,
+      "step": 36160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.060369454037654e-05,
+      "loss": 2.9332,
+      "step": 36165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0594501017071204e-05,
+      "loss": 2.9987,
+      "step": 36170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0585307118730583e-05,
+      "loss": 2.8228,
+      "step": 36175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.057611284568011e-05,
+      "loss": 2.9433,
+      "step": 36180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0566918198245256e-05,
+      "loss": 2.924,
+      "step": 36185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0557723176751525e-05,
+      "loss": 2.9657,
+      "step": 36190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.054852778152439e-05,
+      "loss": 2.9003,
+      "step": 36195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.053933201288936e-05,
+      "loss": 2.8987,
+      "step": 36200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0530135871171964e-05,
+      "loss": 2.9244,
+      "step": 36205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.052093935669775e-05,
+      "loss": 3.0316,
+      "step": 36210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0511742469792235e-05,
+      "loss": 2.8533,
+      "step": 36215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0502545210781e-05,
+      "loss": 3.0109,
+      "step": 36220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.049334757998962e-05,
+      "loss": 2.8934,
+      "step": 36225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.048414957774368e-05,
+      "loss": 2.8954,
+      "step": 36230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.047495120436878e-05,
+      "loss": 2.886,
+      "step": 36235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.046575246019053e-05,
+      "loss": 2.8464,
+      "step": 36240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.045655334553454e-05,
+      "loss": 2.8245,
+      "step": 36245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0447353860726487e-05,
+      "loss": 2.9147,
+      "step": 36250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0438154006092004e-05,
+      "loss": 2.9139,
+      "step": 36255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0428953781956766e-05,
+      "loss": 2.973,
+      "step": 36260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.041975318864643e-05,
+      "loss": 2.8813,
+      "step": 36265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.041055222648672e-05,
+      "loss": 2.8941,
+      "step": 36270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.040135089580331e-05,
+      "loss": 2.8835,
+      "step": 36275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.039214919692194e-05,
+      "loss": 2.8718,
+      "step": 36280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0382947130168347e-05,
+      "loss": 2.9932,
+      "step": 36285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.037374469586824e-05,
+      "loss": 2.8884,
+      "step": 36290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.03645418943474e-05,
+      "loss": 2.9168,
+      "step": 36295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.035533872593161e-05,
+      "loss": 2.9293,
+      "step": 36300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.034613519094665e-05,
+      "loss": 2.8474,
+      "step": 36305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.033693128971829e-05,
+      "loss": 2.8438,
+      "step": 36310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0327727022572366e-05,
+      "loss": 2.9381,
+      "step": 36315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0318522389834685e-05,
+      "loss": 2.9483,
+      "step": 36320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.030931739183109e-05,
+      "loss": 3.0081,
+      "step": 36325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.030011202888744e-05,
+      "loss": 2.9482,
+      "step": 36330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0290906301329564e-05,
+      "loss": 2.88,
+      "step": 36335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.028170020948336e-05,
+      "loss": 2.9202,
+      "step": 36340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.027249375367471e-05,
+      "loss": 2.9725,
+      "step": 36345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.026328693422951e-05,
+      "loss": 2.9612,
+      "step": 36350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.025407975147369e-05,
+      "loss": 2.8957,
+      "step": 36355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.024487220573315e-05,
+      "loss": 2.9458,
+      "step": 36360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.023566429733385e-05,
+      "loss": 2.8979,
+      "step": 36365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.022645602660173e-05,
+      "loss": 2.7857,
+      "step": 36370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0217247393862764e-05,
+      "loss": 2.8849,
+      "step": 36375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0208038399442904e-05,
+      "loss": 2.9669,
+      "step": 36380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.019882904366817e-05,
+      "loss": 2.929,
+      "step": 36385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.018961932686454e-05,
+      "loss": 2.963,
+      "step": 36390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.018040924935805e-05,
+      "loss": 2.954,
+      "step": 36395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0171198811474706e-05,
+      "loss": 2.9079,
+      "step": 36400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.016198801354056e-05,
+      "loss": 2.9297,
+      "step": 36405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.015277685588169e-05,
+      "loss": 2.89,
+      "step": 36410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0143565338824125e-05,
+      "loss": 3.0121,
+      "step": 36415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.013435346269396e-05,
+      "loss": 2.9304,
+      "step": 36420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.012514122781728e-05,
+      "loss": 3.0077,
+      "step": 36425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.011592863452019e-05,
+      "loss": 2.8922,
+      "step": 36430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0106715683128825e-05,
+      "loss": 2.9502,
+      "step": 36435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.009750237396929e-05,
+      "loss": 2.8601,
+      "step": 36440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.008828870736774e-05,
+      "loss": 2.7603,
+      "step": 36445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.007907468365033e-05,
+      "loss": 2.9285,
+      "step": 36450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.006986030314322e-05,
+      "loss": 3.0467,
+      "step": 36455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.00606455661726e-05,
+      "loss": 2.9058,
+      "step": 36460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.005143047306466e-05,
+      "loss": 2.8821,
+      "step": 36465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.004221502414561e-05,
+      "loss": 2.887,
+      "step": 36470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.003299921974166e-05,
+      "loss": 2.94,
+      "step": 36475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.002378306017905e-05,
+      "loss": 2.9248,
+      "step": 36480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0014566545784e-05,
+      "loss": 2.8819,
+      "step": 36485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.000534967688279e-05,
+      "loss": 2.9072,
+      "step": 36490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9996132453801676e-05,
+      "loss": 2.9792,
+      "step": 36495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.998691487686695e-05,
+      "loss": 2.8334,
+      "step": 36500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.997769694640492e-05,
+      "loss": 2.9579,
+      "step": 36505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.996847866274185e-05,
+      "loss": 2.9561,
+      "step": 36510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.995926002620409e-05,
+      "loss": 2.966,
+      "step": 36515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9950041037117945e-05,
+      "loss": 2.9988,
+      "step": 36520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.994082169580979e-05,
+      "loss": 2.8449,
+      "step": 36525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.993160200260596e-05,
+      "loss": 2.9115,
+      "step": 36530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.992238195783283e-05,
+      "loss": 2.9212,
+      "step": 36535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.991316156181679e-05,
+      "loss": 2.9261,
+      "step": 36540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9903940814884195e-05,
+      "loss": 2.8571,
+      "step": 36545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.98947197173615e-05,
+      "loss": 2.9233,
+      "step": 36550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.98854982695751e-05,
+      "loss": 2.853,
+      "step": 36555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9876276471851435e-05,
+      "loss": 2.8747,
+      "step": 36560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9867054324516925e-05,
+      "loss": 2.8314,
+      "step": 36565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9857831827898046e-05,
+      "loss": 2.9816,
+      "step": 36570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.984860898232128e-05,
+      "loss": 3.0054,
+      "step": 36575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.983938578811306e-05,
+      "loss": 2.9914,
+      "step": 36580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.983016224559992e-05,
+      "loss": 2.9431,
+      "step": 36585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.982093835510834e-05,
+      "loss": 2.9093,
+      "step": 36590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.981171411696485e-05,
+      "loss": 2.9373,
+      "step": 36595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9802489531495975e-05,
+      "loss": 2.9853,
+      "step": 36600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9793264599028264e-05,
+      "loss": 2.912,
+      "step": 36605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9784039319888276e-05,
+      "loss": 2.9267,
+      "step": 36610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.977481369440254e-05,
+      "loss": 2.9008,
+      "step": 36615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.976558772289769e-05,
+      "loss": 2.9762,
+      "step": 36620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.975636140570027e-05,
+      "loss": 2.9154,
+      "step": 36625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.97471347431369e-05,
+      "loss": 2.8697,
+      "step": 36630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.973790773553421e-05,
+      "loss": 2.9223,
+      "step": 36635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9728680383218795e-05,
+      "loss": 2.9451,
+      "step": 36640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.971945268651731e-05,
+      "loss": 2.863,
+      "step": 36645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.971022464575642e-05,
+      "loss": 2.7904,
+      "step": 36650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.970099626126278e-05,
+      "loss": 2.8734,
+      "step": 36655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9691767533363053e-05,
+      "loss": 2.8136,
+      "step": 36660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.968253846238395e-05,
+      "loss": 2.8531,
+      "step": 36665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.967330904865215e-05,
+      "loss": 2.9558,
+      "step": 36670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.966407929249438e-05,
+      "loss": 2.8903,
+      "step": 36675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.965484919423736e-05,
+      "loss": 2.8649,
+      "step": 36680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.964561875420782e-05,
+      "loss": 2.8721,
+      "step": 36685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.963638797273253e-05,
+      "loss": 3.0096,
+      "step": 36690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.962715685013821e-05,
+      "loss": 2.8797,
+      "step": 36695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9617925386751685e-05,
+      "loss": 2.8094,
+      "step": 36700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.96086935828997e-05,
+      "loss": 2.9699,
+      "step": 36705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9599461438909066e-05,
+      "loss": 2.9541,
+      "step": 36710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.959022895510661e-05,
+      "loss": 2.9376,
+      "step": 36715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.958099613181911e-05,
+      "loss": 2.9429,
+      "step": 36720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.957176296937345e-05,
+      "loss": 3.0038,
+      "step": 36725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.956252946809644e-05,
+      "loss": 2.8716,
+      "step": 36730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.955329562831493e-05,
+      "loss": 2.9097,
+      "step": 36735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.954406145035583e-05,
+      "loss": 2.885,
+      "step": 36740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9534826934545975e-05,
+      "loss": 2.9103,
+      "step": 36745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.95255920812123e-05,
+      "loss": 2.9093,
+      "step": 36750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.951635689068168e-05,
+      "loss": 2.8738,
+      "step": 36755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.950712136328106e-05,
+      "loss": 2.8769,
+      "step": 36760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.949788549933734e-05,
+      "loss": 2.9039,
+      "step": 36765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.948864929917747e-05,
+      "loss": 2.9052,
+      "step": 36770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.947941276312841e-05,
+      "loss": 2.8267,
+      "step": 36775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.947017589151711e-05,
+      "loss": 2.7422,
+      "step": 36780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9460938684670564e-05,
+      "loss": 2.9193,
+      "step": 36785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.945170114291574e-05,
+      "loss": 2.8534,
+      "step": 36790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.944246326657967e-05,
+      "loss": 2.8454,
+      "step": 36795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.943322505598934e-05,
+      "loss": 2.9112,
+      "step": 36800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9423986511471786e-05,
+      "loss": 2.9088,
+      "step": 36805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.941474763335402e-05,
+      "loss": 2.9106,
+      "step": 36810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.940550842196312e-05,
+      "loss": 2.8416,
+      "step": 36815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9396268877626136e-05,
+      "loss": 2.9848,
+      "step": 36820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.938702900067014e-05,
+      "loss": 2.7837,
+      "step": 36825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9377788791422205e-05,
+      "loss": 3.0173,
+      "step": 36830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.936854825020943e-05,
+      "loss": 2.9042,
+      "step": 36835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.935930737735893e-05,
+      "loss": 2.8973,
+      "step": 36840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.935006617319781e-05,
+      "loss": 2.857,
+      "step": 36845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.934082463805321e-05,
+      "loss": 2.8922,
+      "step": 36850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.933158277225227e-05,
+      "loss": 2.9049,
+      "step": 36855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.932234057612214e-05,
+      "loss": 2.8742,
+      "step": 36860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.931309804998999e-05,
+      "loss": 2.9807,
+      "step": 36865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.930385519418298e-05,
+      "loss": 2.8809,
+      "step": 36870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.929461200902831e-05,
+      "loss": 2.886,
+      "step": 36875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.92853684948532e-05,
+      "loss": 2.8829,
+      "step": 36880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.927612465198483e-05,
+      "loss": 2.9231,
+      "step": 36885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.926688048075044e-05,
+      "loss": 2.7495,
+      "step": 36890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9257635981477265e-05,
+      "loss": 2.8911,
+      "step": 36895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9248391154492544e-05,
+      "loss": 2.8541,
+      "step": 36900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.923914600012354e-05,
+      "loss": 2.924,
+      "step": 36905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9229900518697525e-05,
+      "loss": 2.8485,
+      "step": 36910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.922065471054177e-05,
+      "loss": 2.8886,
+      "step": 36915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.921140857598357e-05,
+      "loss": 2.9009,
+      "step": 36920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.920216211535026e-05,
+      "loss": 2.9729,
+      "step": 36925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.91929153289691e-05,
+      "loss": 3.0147,
+      "step": 36930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9183668217167463e-05,
+      "loss": 2.9474,
+      "step": 36935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.917442078027265e-05,
+      "loss": 2.9577,
+      "step": 36940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9165173018612065e-05,
+      "loss": 2.9398,
+      "step": 36945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.915592493251303e-05,
+      "loss": 2.9833,
+      "step": 36950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.914667652230291e-05,
+      "loss": 2.8729,
+      "step": 36955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.913742778830913e-05,
+      "loss": 2.8207,
+      "step": 36960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.912817873085905e-05,
+      "loss": 2.7696,
+      "step": 36965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9118929350280115e-05,
+      "loss": 2.8585,
+      "step": 36970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.91096796468997e-05,
+      "loss": 2.9741,
+      "step": 36975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.910042962104525e-05,
+      "loss": 2.9202,
+      "step": 36980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9091179273044225e-05,
+      "loss": 2.9933,
+      "step": 36985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.908192860322406e-05,
+      "loss": 2.9308,
+      "step": 36990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.907267761191224e-05,
+      "loss": 2.8384,
+      "step": 36995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.906342629943622e-05,
+      "loss": 2.8655,
+      "step": 37000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9054174666123505e-05,
+      "loss": 2.8533,
+      "step": 37005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.904492271230158e-05,
+      "loss": 2.8505,
+      "step": 37010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.903567043829795e-05,
+      "loss": 3.054,
+      "step": 37015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.902641784444016e-05,
+      "loss": 2.8955,
+      "step": 37020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.901716493105573e-05,
+      "loss": 2.9925,
+      "step": 37025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9007911698472216e-05,
+      "loss": 2.8675,
+      "step": 37030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.899865814701715e-05,
+      "loss": 2.9364,
+      "step": 37035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.89894042770181e-05,
+      "loss": 2.9043,
+      "step": 37040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.898015008880267e-05,
+      "loss": 2.8574,
+      "step": 37045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8970895582698435e-05,
+      "loss": 2.8687,
+      "step": 37050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8961640759033e-05,
+      "loss": 2.8734,
+      "step": 37055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8952385618133956e-05,
+      "loss": 3.0561,
+      "step": 37060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.894313016032896e-05,
+      "loss": 2.8561,
+      "step": 37065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.893387438594562e-05,
+      "loss": 2.8426,
+      "step": 37070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8924618295311604e-05,
+      "loss": 2.9786,
+      "step": 37075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.891536188875455e-05,
+      "loss": 2.8874,
+      "step": 37080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.890610516660213e-05,
+      "loss": 2.9354,
+      "step": 37085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8896848129182024e-05,
+      "loss": 2.9377,
+      "step": 37090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.888759077682192e-05,
+      "loss": 2.9814,
+      "step": 37095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.887833310984954e-05,
+      "loss": 2.8389,
+      "step": 37100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8869075128592565e-05,
+      "loss": 2.9874,
+      "step": 37105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.885981683337875e-05,
+      "loss": 2.8298,
+      "step": 37110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.885055822453579e-05,
+      "loss": 2.9949,
+      "step": 37115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.884129930239146e-05,
+      "loss": 2.9478,
+      "step": 37120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.883204006727353e-05,
+      "loss": 2.9489,
+      "step": 37125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8822780519509736e-05,
+      "loss": 3.0177,
+      "step": 37130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8813520659427865e-05,
+      "loss": 2.9127,
+      "step": 37135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.880426048735572e-05,
+      "loss": 2.857,
+      "step": 37140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.87950000036211e-05,
+      "loss": 2.8591,
+      "step": 37145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.878573920855181e-05,
+      "loss": 2.8892,
+      "step": 37150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8776478102475674e-05,
+      "loss": 3.0472,
+      "step": 37155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8767216685720536e-05,
+      "loss": 2.8443,
+      "step": 37160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.875795495861423e-05,
+      "loss": 2.9315,
+      "step": 37165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8748692921484616e-05,
+      "loss": 2.9526,
+      "step": 37170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.873943057465956e-05,
+      "loss": 2.8021,
+      "step": 37175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.873016791846695e-05,
+      "loss": 2.8881,
+      "step": 37180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8720904953234656e-05,
+      "loss": 2.8938,
+      "step": 37185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8711641679290596e-05,
+      "loss": 2.8383,
+      "step": 37190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8702378096962676e-05,
+      "loss": 2.8999,
+      "step": 37195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.86931142065788e-05,
+      "loss": 2.8545,
+      "step": 37200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.868385000846695e-05,
+      "loss": 2.835,
+      "step": 37205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.867458550295501e-05,
+      "loss": 2.9603,
+      "step": 37210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.866532069037098e-05,
+      "loss": 2.9127,
+      "step": 37215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.865605557104279e-05,
+      "loss": 2.8585,
+      "step": 37220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.864679014529845e-05,
+      "loss": 2.9721,
+      "step": 37225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.863752441346593e-05,
+      "loss": 2.9105,
+      "step": 37230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8628258375873216e-05,
+      "loss": 2.9734,
+      "step": 37235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.861899203284834e-05,
+      "loss": 2.8137,
+      "step": 37240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8609725384719306e-05,
+      "loss": 2.8969,
+      "step": 37245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8600458431814165e-05,
+      "loss": 2.8808,
+      "step": 37250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8591191174460934e-05,
+      "loss": 3.016,
+      "step": 37255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.858192361298768e-05,
+      "loss": 2.8486,
+      "step": 37260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.857265574772246e-05,
+      "loss": 2.8979,
+      "step": 37265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.856338757899334e-05,
+      "loss": 2.9418,
+      "step": 37270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.855411910712844e-05,
+      "loss": 2.9317,
+      "step": 37275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.85448503324558e-05,
+      "loss": 2.9491,
+      "step": 37280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.853558125530356e-05,
+      "loss": 2.9082,
+      "step": 37285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8526311875999827e-05,
+      "loss": 2.9589,
+      "step": 37290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8517042194872754e-05,
+      "loss": 2.9105,
+      "step": 37295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.850777221225044e-05,
+      "loss": 2.9293,
+      "step": 37300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.849850192846105e-05,
+      "loss": 2.8143,
+      "step": 37305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.848923134383275e-05,
+      "loss": 2.9081,
+      "step": 37310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8479960458693704e-05,
+      "loss": 2.8822,
+      "step": 37315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.847068927337209e-05,
+      "loss": 2.9006,
+      "step": 37320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.84614177881961e-05,
+      "loss": 2.9044,
+      "step": 37325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.845214600349394e-05,
+      "loss": 2.8879,
+      "step": 37330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.844287391959381e-05,
+      "loss": 2.7597,
+      "step": 37335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8433601536823954e-05,
+      "loss": 2.8885,
+      "step": 37340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.84243288555126e-05,
+      "loss": 3.039,
+      "step": 37345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.841505587598797e-05,
+      "loss": 2.9008,
+      "step": 37350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.840578259857835e-05,
+      "loss": 2.8424,
+      "step": 37355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.839650902361198e-05,
+      "loss": 2.9118,
+      "step": 37360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.838723515141715e-05,
+      "loss": 2.9492,
+      "step": 37365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.837796098232215e-05,
+      "loss": 2.9564,
+      "step": 37370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.836868651665526e-05,
+      "loss": 3.0389,
+      "step": 37375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.83594117547448e-05,
+      "loss": 2.8927,
+      "step": 37380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8350136696919065e-05,
+      "loss": 2.8652,
+      "step": 37385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.834086134350643e-05,
+      "loss": 2.8852,
+      "step": 37390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.833158569483519e-05,
+      "loss": 2.9091,
+      "step": 37395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.832230975123372e-05,
+      "loss": 2.956,
+      "step": 37400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8313033513030366e-05,
+      "loss": 2.8411,
+      "step": 37405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8303756980553504e-05,
+      "loss": 2.868,
+      "step": 37410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.829448015413152e-05,
+      "loss": 2.7869,
+      "step": 37415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.828520303409278e-05,
+      "loss": 2.9744,
+      "step": 37420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8275925620765724e-05,
+      "loss": 3.001,
+      "step": 37425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.826664791447872e-05,
+      "loss": 2.975,
+      "step": 37430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.825736991556022e-05,
+      "loss": 2.9074,
+      "step": 37435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.824809162433865e-05,
+      "loss": 2.8827,
+      "step": 37440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.823881304114245e-05,
+      "loss": 2.9693,
+      "step": 37445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.822953416630008e-05,
+      "loss": 2.9695,
+      "step": 37450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.822025500013999e-05,
+      "loss": 2.9306,
+      "step": 37455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8210975542990675e-05,
+      "loss": 2.965,
+      "step": 37460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.82016957951806e-05,
+      "loss": 2.9021,
+      "step": 37465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8192415757038254e-05,
+      "loss": 2.8817,
+      "step": 37470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818313542889217e-05,
+      "loss": 2.8794,
+      "step": 37475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8173854811070825e-05,
+      "loss": 2.9092,
+      "step": 37480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8164573903902775e-05,
+      "loss": 2.9448,
+      "step": 37485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815529270771653e-05,
+      "loss": 2.8741,
+      "step": 37490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814601122284068e-05,
+      "loss": 2.9163,
+      "step": 37495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8136729449603734e-05,
+      "loss": 2.8327,
+      "step": 37500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8127447388334266e-05,
+      "loss": 2.8612,
+      "step": 37505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.811816503936088e-05,
+      "loss": 2.9278,
+      "step": 37510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.810888240301213e-05,
+      "loss": 2.9108,
+      "step": 37515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.809959947961664e-05,
+      "loss": 2.8943,
+      "step": 37520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8090316269502986e-05,
+      "loss": 2.8089,
+      "step": 37525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.808103277299981e-05,
+      "loss": 2.9782,
+      "step": 37530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8071748990435727e-05,
+      "loss": 2.8616,
+      "step": 37535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.806246492213937e-05,
+      "loss": 2.8736,
+      "step": 37540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8053180568439414e-05,
+      "loss": 2.9147,
+      "step": 37545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.804389592966448e-05,
+      "loss": 2.857,
+      "step": 37550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.803461100614327e-05,
+      "loss": 2.861,
+      "step": 37555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.802532579820442e-05,
+      "loss": 3.02,
+      "step": 37560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8016040306176665e-05,
+      "loss": 2.9708,
+      "step": 37565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.800675453038866e-05,
+      "loss": 2.9661,
+      "step": 37570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.799746847116914e-05,
+      "loss": 2.9023,
+      "step": 37575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.798818212884679e-05,
+      "loss": 2.9054,
+      "step": 37580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.797889550375038e-05,
+      "loss": 2.9402,
+      "step": 37585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.796960859620863e-05,
+      "loss": 2.8467,
+      "step": 37590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.796032140655029e-05,
+      "loss": 2.9492,
+      "step": 37595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.795103393510412e-05,
+      "loss": 2.9086,
+      "step": 37600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.794174618219886e-05,
+      "loss": 2.9785,
+      "step": 37605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.793245814816333e-05,
+      "loss": 2.9201,
+      "step": 37610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.79231698333263e-05,
+      "loss": 2.9886,
+      "step": 37615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7913881238016545e-05,
+      "loss": 3.0072,
+      "step": 37620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7904592362562905e-05,
+      "loss": 2.9139,
+      "step": 37625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.789530320729417e-05,
+      "loss": 2.9223,
+      "step": 37630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.78860137725392e-05,
+      "loss": 2.9628,
+      "step": 37635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.787672405862681e-05,
+      "loss": 2.9707,
+      "step": 37640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.786743406588584e-05,
+      "loss": 2.9116,
+      "step": 37645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.785814379464518e-05,
+      "loss": 2.9109,
+      "step": 37650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7848853245233656e-05,
+      "loss": 2.8845,
+      "step": 37655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.783956241798019e-05,
+      "loss": 2.9451,
+      "step": 37660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7830271313213615e-05,
+      "loss": 2.9132,
+      "step": 37665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7820979931262856e-05,
+      "loss": 2.8817,
+      "step": 37670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.781168827245683e-05,
+      "loss": 2.869,
+      "step": 37675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.780239633712443e-05,
+      "loss": 2.9374,
+      "step": 37680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.77931041255946e-05,
+      "loss": 2.9983,
+      "step": 37685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.778381163819626e-05,
+      "loss": 2.9188,
+      "step": 37690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.777451887525838e-05,
+      "loss": 2.9612,
+      "step": 37695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7765225837109884e-05,
+      "loss": 2.937,
+      "step": 37700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.775593252407976e-05,
+      "loss": 2.9689,
+      "step": 37705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.774663893649696e-05,
+      "loss": 2.894,
+      "step": 37710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.773734507469049e-05,
+      "loss": 2.9203,
+      "step": 37715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.772805093898933e-05,
+      "loss": 2.8063,
+      "step": 37720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.771875652972249e-05,
+      "loss": 2.9015,
+      "step": 37725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.770946184721898e-05,
+      "loss": 2.9073,
+      "step": 37730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.770016689180783e-05,
+      "loss": 2.8615,
+      "step": 37735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.769087166381807e-05,
+      "loss": 2.9437,
+      "step": 37740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.768157616357873e-05,
+      "loss": 2.9249,
+      "step": 37745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.767228039141888e-05,
+      "loss": 2.8794,
+      "step": 37750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.766298434766758e-05,
+      "loss": 2.8633,
+      "step": 37755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7653688032653886e-05,
+      "loss": 2.9057,
+      "step": 37760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.76443914467069e-05,
+      "loss": 2.8831,
+      "step": 37765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.763509459015568e-05,
+      "loss": 2.9021,
+      "step": 37770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.762579746332936e-05,
+      "loss": 2.9267,
+      "step": 37775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.761650006655702e-05,
+      "loss": 2.9196,
+      "step": 37780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7607202400167806e-05,
+      "loss": 2.9046,
+      "step": 37785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7597904464490836e-05,
+      "loss": 2.9647,
+      "step": 37790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.758860625985524e-05,
+      "loss": 2.8236,
+      "step": 37795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.757930778659017e-05,
+      "loss": 2.951,
+      "step": 37800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.75700090450248e-05,
+      "loss": 2.9261,
+      "step": 37805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.756071003548829e-05,
+      "loss": 2.9043,
+      "step": 37810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7551410758309776e-05,
+      "loss": 2.962,
+      "step": 37815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7542111213818495e-05,
+      "loss": 2.923,
+      "step": 37820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.753281140234362e-05,
+      "loss": 2.9845,
+      "step": 37825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.752351132421435e-05,
+      "loss": 2.8345,
+      "step": 37830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.751421097975992e-05,
+      "loss": 2.9808,
+      "step": 37835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7504910369309536e-05,
+      "loss": 2.9523,
+      "step": 37840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7495609493192445e-05,
+      "loss": 2.9221,
+      "step": 37845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.748630835173787e-05,
+      "loss": 3.0083,
+      "step": 37850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.747700694527507e-05,
+      "loss": 2.8758,
+      "step": 37855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.746770527413332e-05,
+      "loss": 2.9078,
+      "step": 37860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.745840333864186e-05,
+      "loss": 3.0111,
+      "step": 37865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7449101139130015e-05,
+      "loss": 2.9585,
+      "step": 37870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7439798675927026e-05,
+      "loss": 2.9238,
+      "step": 37875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7430495949362216e-05,
+      "loss": 2.9088,
+      "step": 37880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.742119295976489e-05,
+      "loss": 2.8018,
+      "step": 37885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.741188970746436e-05,
+      "loss": 2.9291,
+      "step": 37890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.740258619278998e-05,
+      "loss": 2.864,
+      "step": 37895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.739328241607105e-05,
+      "loss": 2.9093,
+      "step": 37900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7383978377636925e-05,
+      "loss": 2.8627,
+      "step": 37905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.737467407781696e-05,
+      "loss": 2.919,
+      "step": 37910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7365369516940515e-05,
+      "loss": 2.9226,
+      "step": 37915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.735606469533699e-05,
+      "loss": 2.9497,
+      "step": 37920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.734675961333572e-05,
+      "loss": 2.8001,
+      "step": 37925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.733745427126612e-05,
+      "loss": 2.8385,
+      "step": 37930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.73281486694576e-05,
+      "loss": 2.9706,
+      "step": 37935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.731884280823957e-05,
+      "loss": 2.8699,
+      "step": 37940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.730953668794143e-05,
+      "loss": 2.95,
+      "step": 37945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.730023030889262e-05,
+      "loss": 2.9854,
+      "step": 37950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.729092367142257e-05,
+      "loss": 2.9531,
+      "step": 37955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.728161677586073e-05,
+      "loss": 2.9092,
+      "step": 37960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7272309622536556e-05,
+      "loss": 3.0203,
+      "step": 37965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.726300221177952e-05,
+      "loss": 2.9385,
+      "step": 37970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.725369454391907e-05,
+      "loss": 2.8807,
+      "step": 37975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7244386619284706e-05,
+      "loss": 2.86,
+      "step": 37980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7235078438205946e-05,
+      "loss": 2.9307,
+      "step": 37985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.722577000101223e-05,
+      "loss": 2.8729,
+      "step": 37990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.721646130803313e-05,
+      "loss": 2.935,
+      "step": 37995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7207152359598126e-05,
+      "loss": 2.8613,
+      "step": 38000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.719784315603676e-05,
+      "loss": 2.9663,
+      "step": 38005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.718853369767857e-05,
+      "loss": 2.8431,
+      "step": 38010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7179223984853095e-05,
+      "loss": 2.9182,
+      "step": 38015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.716991401788988e-05,
+      "loss": 2.8974,
+      "step": 38020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7160603797118506e-05,
+      "loss": 2.921,
+      "step": 38025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7151293322868546e-05,
+      "loss": 2.7851,
+      "step": 38030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.714198259546959e-05,
+      "loss": 2.8768,
+      "step": 38035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.71326716152512e-05,
+      "loss": 2.8862,
+      "step": 38040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.712336038254301e-05,
+      "loss": 2.8535,
+      "step": 38045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.71140488976746e-05,
+      "loss": 2.9406,
+      "step": 38050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.71047371609756e-05,
+      "loss": 2.9089,
+      "step": 38055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.709542517277564e-05,
+      "loss": 2.8545,
+      "step": 38060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7086112933404356e-05,
+      "loss": 2.8214,
+      "step": 38065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.707680044319138e-05,
+      "loss": 2.909,
+      "step": 38070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.706748770246637e-05,
+      "loss": 2.8879,
+      "step": 38075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7058174711558997e-05,
+      "loss": 2.966,
+      "step": 38080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.704886147079893e-05,
+      "loss": 2.8955,
+      "step": 38085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.703954798051585e-05,
+      "loss": 2.8999,
+      "step": 38090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.703023424103943e-05,
+      "loss": 2.9012,
+      "step": 38095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7020920252699374e-05,
+      "loss": 2.9701,
+      "step": 38100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.701160601582541e-05,
+      "loss": 2.9536,
+      "step": 38105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.700229153074722e-05,
+      "loss": 2.8541,
+      "step": 38110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.699297679779456e-05,
+      "loss": 2.8796,
+      "step": 38115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.698366181729714e-05,
+      "loss": 2.9423,
+      "step": 38120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.69743465895847e-05,
+      "loss": 2.8686,
+      "step": 38125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.696503111498701e-05,
+      "loss": 2.9098,
+      "step": 38130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6955715393833805e-05,
+      "loss": 2.8536,
+      "step": 38135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.694639942645489e-05,
+      "loss": 2.8156,
+      "step": 38140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.693708321318e-05,
+      "loss": 2.8586,
+      "step": 38145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.692776675433894e-05,
+      "loss": 2.9319,
+      "step": 38150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6918450050261496e-05,
+      "loss": 2.9036,
+      "step": 38155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.690913310127748e-05,
+      "loss": 2.9553,
+      "step": 38160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.68998159077167e-05,
+      "loss": 2.882,
+      "step": 38165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.689049846990897e-05,
+      "loss": 2.9337,
+      "step": 38170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.688118078818413e-05,
+      "loss": 2.929,
+      "step": 38175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.687186286287199e-05,
+      "loss": 2.873,
+      "step": 38180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6862544694302444e-05,
+      "loss": 2.8809,
+      "step": 38185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6853226282805305e-05,
+      "loss": 2.8802,
+      "step": 38190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.684390762871045e-05,
+      "loss": 3.0047,
+      "step": 38195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.683458873234775e-05,
+      "loss": 2.8953,
+      "step": 38200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6825269594047084e-05,
+      "loss": 2.9609,
+      "step": 38205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6815950214138334e-05,
+      "loss": 2.8409,
+      "step": 38210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6806630592951416e-05,
+      "loss": 2.8808,
+      "step": 38215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6797310730816225e-05,
+      "loss": 2.9478,
+      "step": 38220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.678799062806265e-05,
+      "loss": 2.9715,
+      "step": 38225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.677867028502066e-05,
+      "loss": 2.9671,
+      "step": 38230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.676934970202016e-05,
+      "loss": 2.9187,
+      "step": 38235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.676002887939109e-05,
+      "loss": 2.8356,
+      "step": 38240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6750707817463425e-05,
+      "loss": 2.8684,
+      "step": 38245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6741386516567075e-05,
+      "loss": 2.872,
+      "step": 38250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.673206497703204e-05,
+      "loss": 2.8774,
+      "step": 38255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.672274319918828e-05,
+      "loss": 2.9954,
+      "step": 38260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6713421183365776e-05,
+      "loss": 2.9044,
+      "step": 38265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6704098929894534e-05,
+      "loss": 2.8621,
+      "step": 38270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.669477643910452e-05,
+      "loss": 2.9997,
+      "step": 38275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.668545371132579e-05,
+      "loss": 2.9042,
+      "step": 38280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.667613074688831e-05,
+      "loss": 3.0349,
+      "step": 38285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6666807546122145e-05,
+      "loss": 2.9199,
+      "step": 38290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.665748410935731e-05,
+      "loss": 2.8542,
+      "step": 38295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6648160436923834e-05,
+      "loss": 2.9548,
+      "step": 38300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6638836529151784e-05,
+      "loss": 2.8692,
+      "step": 38305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.662951238637121e-05,
+      "loss": 2.9136,
+      "step": 38310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.662018800891219e-05,
+      "loss": 2.9309,
+      "step": 38315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.661086339710476e-05,
+      "loss": 2.8888,
+      "step": 38320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6601538551279055e-05,
+      "loss": 2.888,
+      "step": 38325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.659221347176512e-05,
+      "loss": 2.8546,
+      "step": 38330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6582888158893085e-05,
+      "loss": 2.9312,
+      "step": 38335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.657356261299305e-05,
+      "loss": 2.9274,
+      "step": 38340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6564236834395115e-05,
+      "loss": 2.9947,
+      "step": 38345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.655491082342943e-05,
+      "loss": 2.9601,
+      "step": 38350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.65455845804261e-05,
+      "loss": 3.0021,
+      "step": 38355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.653625810571529e-05,
+      "loss": 2.9961,
+      "step": 38360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.652693139962712e-05,
+      "loss": 2.8948,
+      "step": 38365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.651760446249178e-05,
+      "loss": 3.0174,
+      "step": 38370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.65082772946394e-05,
+      "loss": 2.8852,
+      "step": 38375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.649894989640018e-05,
+      "loss": 2.8829,
+      "step": 38380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.64896222681043e-05,
+      "loss": 2.854,
+      "step": 38385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.648029441008193e-05,
+      "loss": 2.8965,
+      "step": 38390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.647096632266328e-05,
+      "loss": 2.9584,
+      "step": 38395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6461638006178555e-05,
+      "loss": 2.9511,
+      "step": 38400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.645230946095795e-05,
+      "loss": 2.96,
+      "step": 38405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6442980687331736e-05,
+      "loss": 2.865,
+      "step": 38410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.643365168563008e-05,
+      "loss": 3.003,
+      "step": 38415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.642432245618326e-05,
+      "loss": 2.9045,
+      "step": 38420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6414992999321515e-05,
+      "loss": 2.9366,
+      "step": 38425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.64056633153751e-05,
+      "loss": 2.868,
+      "step": 38430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6396333404674264e-05,
+      "loss": 2.9089,
+      "step": 38435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.638700326754929e-05,
+      "loss": 2.7398,
+      "step": 38440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.637767290433046e-05,
+      "loss": 2.922,
+      "step": 38445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.636834231534803e-05,
+      "loss": 2.9312,
+      "step": 38450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.635901150093235e-05,
+      "loss": 2.8783,
+      "step": 38455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6349680461413655e-05,
+      "loss": 2.9924,
+      "step": 38460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.634034919712231e-05,
+      "loss": 2.9591,
+      "step": 38465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.633101770838859e-05,
+      "loss": 2.8616,
+      "step": 38470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.632168599554286e-05,
+      "loss": 3.0016,
+      "step": 38475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.631235405891542e-05,
+      "loss": 2.9687,
+      "step": 38480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.630302189883664e-05,
+      "loss": 2.9811,
+      "step": 38485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.629368951563685e-05,
+      "loss": 2.8947,
+      "step": 38490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.628435690964642e-05,
+      "loss": 2.9554,
+      "step": 38495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6275024081195716e-05,
+      "loss": 2.873,
+      "step": 38500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6265691030615085e-05,
+      "loss": 2.8632,
+      "step": 38505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6256357758234934e-05,
+      "loss": 2.8319,
+      "step": 38510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.624702426438566e-05,
+      "loss": 2.9604,
+      "step": 38515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.623769054939762e-05,
+      "loss": 2.967,
+      "step": 38520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.622835661360128e-05,
+      "loss": 2.8113,
+      "step": 38525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6219022457326985e-05,
+      "loss": 2.8457,
+      "step": 38530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.620968808090521e-05,
+      "loss": 2.8823,
+      "step": 38535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6200353484666346e-05,
+      "loss": 2.9062,
+      "step": 38540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6191018668940855e-05,
+      "loss": 2.9178,
+      "step": 38545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.618168363405917e-05,
+      "loss": 2.9094,
+      "step": 38550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6172348380351724e-05,
+      "loss": 3.0012,
+      "step": 38555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6163012908149014e-05,
+      "loss": 2.9075,
+      "step": 38560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6153677217781476e-05,
+      "loss": 2.8777,
+      "step": 38565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.614434130957959e-05,
+      "loss": 3.0064,
+      "step": 38570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.613500518387385e-05,
+      "loss": 2.8895,
+      "step": 38575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.612566884099473e-05,
+      "loss": 2.9514,
+      "step": 38580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6116332281272744e-05,
+      "loss": 2.9568,
+      "step": 38585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.61069955050384e-05,
+      "loss": 2.7652,
+      "step": 38590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6097658512622187e-05,
+      "loss": 2.8886,
+      "step": 38595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.608832130435465e-05,
+      "loss": 2.8231,
+      "step": 38600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.607898388056631e-05,
+      "loss": 2.9853,
+      "step": 38605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.606964624158769e-05,
+      "loss": 2.8414,
+      "step": 38610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.606030838774934e-05,
+      "loss": 2.9798,
+      "step": 38615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6050970319381823e-05,
+      "loss": 2.926,
+      "step": 38620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.604163203681568e-05,
+      "loss": 2.9032,
+      "step": 38625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.603229354038151e-05,
+      "loss": 2.9411,
+      "step": 38630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.602295483040985e-05,
+      "loss": 2.9748,
+      "step": 38635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.60136159072313e-05,
+      "loss": 2.9822,
+      "step": 38640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6004276771176433e-05,
+      "loss": 2.9414,
+      "step": 38645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.599493742257586e-05,
+      "loss": 2.9509,
+      "step": 38650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.59855978617602e-05,
+      "loss": 2.8694,
+      "step": 38655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.597625808906003e-05,
+      "loss": 2.8197,
+      "step": 38660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5966918104806e-05,
+      "loss": 3.0196,
+      "step": 38665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5957577909328695e-05,
+      "loss": 2.8918,
+      "step": 38670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.594823750295881e-05,
+      "loss": 2.9807,
+      "step": 38675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.593889688602694e-05,
+      "loss": 2.9667,
+      "step": 38680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.592955605886373e-05,
+      "loss": 3.0255,
+      "step": 38685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5920215021799885e-05,
+      "loss": 2.9401,
+      "step": 38690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5910873775166016e-05,
+      "loss": 2.872,
+      "step": 38695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5901532319292824e-05,
+      "loss": 2.8179,
+      "step": 38700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.589219065451098e-05,
+      "loss": 2.8651,
+      "step": 38705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.588284878115116e-05,
+      "loss": 2.8563,
+      "step": 38710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.587350669954409e-05,
+      "loss": 2.8735,
+      "step": 38715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.586416441002043e-05,
+      "loss": 2.9282,
+      "step": 38720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.585482191291089e-05,
+      "loss": 2.8598,
+      "step": 38725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.584547920854622e-05,
+      "loss": 2.914,
+      "step": 38730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.583613629725714e-05,
+      "loss": 2.9205,
+      "step": 38735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.582679317937434e-05,
+      "loss": 2.8616,
+      "step": 38740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.58174498552286e-05,
+      "loss": 2.9619,
+      "step": 38745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5808106325150634e-05,
+      "loss": 2.7686,
+      "step": 38750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.579876258947121e-05,
+      "loss": 2.8478,
+      "step": 38755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.578941864852109e-05,
+      "loss": 2.813,
+      "step": 38760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.578007450263103e-05,
+      "loss": 2.9957,
+      "step": 38765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.57707301521318e-05,
+      "loss": 2.9462,
+      "step": 38770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.57613855973542e-05,
+      "loss": 2.8751,
+      "step": 38775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.575204083862902e-05,
+      "loss": 2.9227,
+      "step": 38780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.574269587628702e-05,
+      "loss": 2.7829,
+      "step": 38785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.573335071065904e-05,
+      "loss": 2.843,
+      "step": 38790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.572400534207589e-05,
+      "loss": 2.8843,
+      "step": 38795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.571465977086836e-05,
+      "loss": 2.9119,
+      "step": 38800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5705313997367295e-05,
+      "loss": 2.9616,
+      "step": 38805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.569596802190351e-05,
+      "loss": 2.9585,
+      "step": 38810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.568662184480786e-05,
+      "loss": 2.9334,
+      "step": 38815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5677275466411186e-05,
+      "loss": 2.9254,
+      "step": 38820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.566792888704433e-05,
+      "loss": 2.9485,
+      "step": 38825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.565858210703817e-05,
+      "loss": 2.9242,
+      "step": 38830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.564923512672357e-05,
+      "loss": 2.9262,
+      "step": 38835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.563988794643139e-05,
+      "loss": 2.7887,
+      "step": 38840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.563054056649253e-05,
+      "loss": 2.8505,
+      "step": 38845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5621192987237866e-05,
+      "loss": 2.9535,
+      "step": 38850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.561184520899829e-05,
+      "loss": 2.9196,
+      "step": 38855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5602497232104714e-05,
+      "loss": 2.8381,
+      "step": 38860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5593149056888036e-05,
+      "loss": 2.8623,
+      "step": 38865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.558380068367919e-05,
+      "loss": 2.9379,
+      "step": 38870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.557445211280909e-05,
+      "loss": 2.8807,
+      "step": 38875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.556510334460866e-05,
+      "loss": 2.8702,
+      "step": 38880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.555575437940886e-05,
+      "loss": 2.9291,
+      "step": 38885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5546405217540606e-05,
+      "loss": 2.835,
+      "step": 38890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5537055859334865e-05,
+      "loss": 2.8461,
+      "step": 38895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5527706305122596e-05,
+      "loss": 2.8338,
+      "step": 38900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5518356555234766e-05,
+      "loss": 2.8348,
+      "step": 38905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5509006610002334e-05,
+      "loss": 2.9308,
+      "step": 38910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.549965646975628e-05,
+      "loss": 2.9519,
+      "step": 38915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.549030613482761e-05,
+      "loss": 2.9271,
+      "step": 38920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.54809556055473e-05,
+      "loss": 2.9987,
+      "step": 38925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5471604882246353e-05,
+      "loss": 2.9179,
+      "step": 38930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5462253965255786e-05,
+      "loss": 2.9543,
+      "step": 38935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.54529028549066e-05,
+      "loss": 2.872,
+      "step": 38940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.544355155152983e-05,
+      "loss": 2.9168,
+      "step": 38945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.543420005545648e-05,
+      "loss": 2.9155,
+      "step": 38950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.542484836701759e-05,
+      "loss": 2.902,
+      "step": 38955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5415496486544216e-05,
+      "loss": 2.8562,
+      "step": 38960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.540614441436739e-05,
+      "loss": 2.947,
+      "step": 38965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.539679215081819e-05,
+      "loss": 2.9029,
+      "step": 38970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.538743969622764e-05,
+      "loss": 2.8107,
+      "step": 38975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5378087050926855e-05,
+      "loss": 2.9297,
+      "step": 38980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5368734215246864e-05,
+      "loss": 2.9081,
+      "step": 38985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5359381189518775e-05,
+      "loss": 2.7933,
+      "step": 38990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.535002797407367e-05,
+      "loss": 2.9468,
+      "step": 38995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.534067456924264e-05,
+      "loss": 2.8945,
+      "step": 39000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5331320975356804e-05,
+      "loss": 2.913,
+      "step": 39005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5321967192747246e-05,
+      "loss": 2.986,
+      "step": 39010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.531261322174509e-05,
+      "loss": 2.8588,
+      "step": 39015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.530325906268146e-05,
+      "loss": 2.8698,
+      "step": 39020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.529390471588749e-05,
+      "loss": 2.8724,
+      "step": 39025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5284550181694306e-05,
+      "loss": 2.9366,
+      "step": 39030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.527519546043305e-05,
+      "loss": 3.0149,
+      "step": 39035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.526584055243488e-05,
+      "loss": 2.9878,
+      "step": 39040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.525648545803094e-05,
+      "loss": 2.9673,
+      "step": 39045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.52471301775524e-05,
+      "loss": 3.0611,
+      "step": 39050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.523777471133041e-05,
+      "loss": 2.9062,
+      "step": 39055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.522841905969616e-05,
+      "loss": 2.8616,
+      "step": 39060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.521906322298083e-05,
+      "loss": 2.8715,
+      "step": 39065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5209707201515616e-05,
+      "loss": 2.8759,
+      "step": 39070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.52003509956317e-05,
+      "loss": 2.957,
+      "step": 39075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.519099460566028e-05,
+      "loss": 2.8962,
+      "step": 39080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5181638031932584e-05,
+      "loss": 2.9173,
+      "step": 39085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5172281274779794e-05,
+      "loss": 2.986,
+      "step": 39090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.516292433453315e-05,
+      "loss": 2.8654,
+      "step": 39095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5153567211523896e-05,
+      "loss": 2.8546,
+      "step": 39100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.514420990608322e-05,
+      "loss": 2.835,
+      "step": 39105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5134852418542406e-05,
+      "loss": 2.9277,
+      "step": 39110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5125494749232656e-05,
+      "loss": 2.8952,
+      "step": 39115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.511613689848527e-05,
+      "loss": 2.9003,
+      "step": 39120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.510677886663148e-05,
+      "loss": 2.9257,
+      "step": 39125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.509742065400256e-05,
+      "loss": 2.9495,
+      "step": 39130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.508806226092978e-05,
+      "loss": 2.9421,
+      "step": 39135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.507870368774442e-05,
+      "loss": 2.9736,
+      "step": 39140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.506934493477775e-05,
+      "loss": 2.8598,
+      "step": 39145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5059986002361074e-05,
+      "loss": 2.9178,
+      "step": 39150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.50506268908257e-05,
+      "loss": 2.9907,
+      "step": 39155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.504126760050291e-05,
+      "loss": 2.8362,
+      "step": 39160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.503190813172402e-05,
+      "loss": 2.9375,
+      "step": 39165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.502254848482035e-05,
+      "loss": 2.8701,
+      "step": 39170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.501318866012322e-05,
+      "loss": 2.9198,
+      "step": 39175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.500382865796397e-05,
+      "loss": 3.0389,
+      "step": 39180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.499446847867392e-05,
+      "loss": 2.9117,
+      "step": 39185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.498510812258443e-05,
+      "loss": 2.8322,
+      "step": 39190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4975747590026825e-05,
+      "loss": 2.8825,
+      "step": 39195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.496638688133247e-05,
+      "loss": 2.8644,
+      "step": 39200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4957025996832734e-05,
+      "loss": 2.9517,
+      "step": 39205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4947664936858966e-05,
+      "loss": 2.8239,
+      "step": 39210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.493830370174254e-05,
+      "loss": 2.8351,
+      "step": 39215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.492894229181484e-05,
+      "loss": 2.9448,
+      "step": 39220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.491958070740728e-05,
+      "loss": 2.9297,
+      "step": 39225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.49102189488512e-05,
+      "loss": 3.0029,
+      "step": 39230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.490085701647805e-05,
+      "loss": 2.8797,
+      "step": 39235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4891494910619167e-05,
+      "loss": 2.9321,
+      "step": 39240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.488213263160602e-05,
+      "loss": 3.0159,
+      "step": 39245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.487277017977e-05,
+      "loss": 2.949,
+      "step": 39250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.486340755544254e-05,
+      "loss": 2.9539,
+      "step": 39255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.485404475895505e-05,
+      "loss": 2.829,
+      "step": 39260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4844681790638966e-05,
+      "loss": 3.0433,
+      "step": 39265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4835318650825764e-05,
+      "loss": 2.8908,
+      "step": 39270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4825955339846834e-05,
+      "loss": 2.9326,
+      "step": 39275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.481659185803367e-05,
+      "loss": 2.9567,
+      "step": 39280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.480722820571773e-05,
+      "loss": 2.8709,
+      "step": 39285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.479786438323045e-05,
+      "loss": 2.8734,
+      "step": 39290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.478850039090333e-05,
+      "loss": 2.9655,
+      "step": 39295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.477913622906782e-05,
+      "loss": 2.9544,
+      "step": 39300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.476977189805541e-05,
+      "loss": 2.9563,
+      "step": 39305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.476040739819761e-05,
+      "loss": 2.9185,
+      "step": 39310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.475104272982589e-05,
+      "loss": 2.7891,
+      "step": 39315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.474167789327177e-05,
+      "loss": 2.8379,
+      "step": 39320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.473231288886673e-05,
+      "loss": 2.9808,
+      "step": 39325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4722947716942306e-05,
+      "loss": 2.8793,
+      "step": 39330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.471358237783001e-05,
+      "loss": 2.8812,
+      "step": 39335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.470421687186136e-05,
+      "loss": 2.8906,
+      "step": 39340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.469485119936789e-05,
+      "loss": 2.9611,
+      "step": 39345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.468548536068113e-05,
+      "loss": 2.8418,
+      "step": 39350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4676119356132646e-05,
+      "loss": 2.8912,
+      "step": 39355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4666753186053946e-05,
+      "loss": 2.9819,
+      "step": 39360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4657386850776614e-05,
+      "loss": 2.8476,
+      "step": 39365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.46480203506322e-05,
+      "loss": 2.9334,
+      "step": 39370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.463865368595228e-05,
+      "loss": 2.9772,
+      "step": 39375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4629286857068394e-05,
+      "loss": 2.9703,
+      "step": 39380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4619919864312156e-05,
+      "loss": 2.9328,
+      "step": 39385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4610552708015126e-05,
+      "loss": 2.8895,
+      "step": 39390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.46011853885089e-05,
+      "loss": 2.8405,
+      "step": 39395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.459181790612508e-05,
+      "loss": 2.9404,
+      "step": 39400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.458245026119524e-05,
+      "loss": 3.0248,
+      "step": 39405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.457308245405101e-05,
+      "loss": 2.9314,
+      "step": 39410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.456371448502399e-05,
+      "loss": 2.8899,
+      "step": 39415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4554346354445795e-05,
+      "loss": 2.984,
+      "step": 39420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.454497806264807e-05,
+      "loss": 3.0094,
+      "step": 39425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.453560960996241e-05,
+      "loss": 2.8832,
+      "step": 39430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.452624099672049e-05,
+      "loss": 2.9133,
+      "step": 39435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4516872223253914e-05,
+      "loss": 2.9449,
+      "step": 39440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4507503289894344e-05,
+      "loss": 2.8773,
+      "step": 39445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4498134196973425e-05,
+      "loss": 2.8999,
+      "step": 39450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.448876494482282e-05,
+      "loss": 2.9702,
+      "step": 39455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.447939553377419e-05,
+      "loss": 2.9748,
+      "step": 39460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.44700259641592e-05,
+      "loss": 2.9291,
+      "step": 39465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4460656236309526e-05,
+      "loss": 3.0138,
+      "step": 39470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.445128635055685e-05,
+      "loss": 2.914,
+      "step": 39475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4441916307232845e-05,
+      "loss": 2.9744,
+      "step": 39480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.443254610666923e-05,
+      "loss": 2.8969,
+      "step": 39485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.442317574919766e-05,
+      "loss": 2.9421,
+      "step": 39490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.441380523514988e-05,
+      "loss": 2.9348,
+      "step": 39495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.440443456485758e-05,
+      "loss": 2.9973,
+      "step": 39500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4395063738652455e-05,
+      "loss": 2.9117,
+      "step": 39505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4385692756866224e-05,
+      "loss": 2.7862,
+      "step": 39510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.437632161983065e-05,
+      "loss": 2.9954,
+      "step": 39515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.436695032787743e-05,
+      "loss": 2.9478,
+      "step": 39520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.43575788813383e-05,
+      "loss": 2.9504,
+      "step": 39525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.434820728054502e-05,
+      "loss": 2.964,
+      "step": 39530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4338835525829303e-05,
+      "loss": 2.9383,
+      "step": 39535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4329463617522934e-05,
+      "loss": 2.9072,
+      "step": 39540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4320091555957655e-05,
+      "loss": 2.9232,
+      "step": 39545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.431071934146522e-05,
+      "loss": 2.9884,
+      "step": 39550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.430134697437741e-05,
+      "loss": 2.9779,
+      "step": 39555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4291974455025974e-05,
+      "loss": 2.9341,
+      "step": 39560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.428260178374274e-05,
+      "loss": 2.898,
+      "step": 39565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.427322896085945e-05,
+      "loss": 2.8986,
+      "step": 39570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.42638559867079e-05,
+      "loss": 2.9989,
+      "step": 39575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.425448286161988e-05,
+      "loss": 2.8006,
+      "step": 39580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.42451095859272e-05,
+      "loss": 2.8528,
+      "step": 39585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.423573615996168e-05,
+      "loss": 2.8394,
+      "step": 39590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.42263625840551e-05,
+      "loss": 2.8503,
+      "step": 39595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.42169888585393e-05,
+      "loss": 2.9601,
+      "step": 39600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.420761498374607e-05,
+      "loss": 2.8659,
+      "step": 39605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.419824096000726e-05,
+      "loss": 2.906,
+      "step": 39610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4188866787654704e-05,
+      "loss": 2.8477,
+      "step": 39615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.417949246702023e-05,
+      "loss": 2.8904,
+      "step": 39620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.41701179984357e-05,
+      "loss": 2.9391,
+      "step": 39625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4160743382232924e-05,
+      "loss": 2.913,
+      "step": 39630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.415136861874378e-05,
+      "loss": 2.9024,
+      "step": 39635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.414199370830012e-05,
+      "loss": 2.8174,
+      "step": 39640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4132618651233814e-05,
+      "loss": 2.8927,
+      "step": 39645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4123243447876716e-05,
+      "loss": 2.8618,
+      "step": 39650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.41138680985607e-05,
+      "loss": 2.9224,
+      "step": 39655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.410449260361766e-05,
+      "loss": 2.8478,
+      "step": 39660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.409511696337947e-05,
+      "loss": 2.8954,
+      "step": 39665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.408574117817802e-05,
+      "loss": 2.9927,
+      "step": 39670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.407636524834521e-05,
+      "loss": 2.9178,
+      "step": 39675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4066989174212925e-05,
+      "loss": 3.0282,
+      "step": 39680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4057612956113066e-05,
+      "loss": 2.8586,
+      "step": 39685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.404823659437757e-05,
+      "loss": 2.869,
+      "step": 39690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4038860089338326e-05,
+      "loss": 2.9595,
+      "step": 39695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.402948344132727e-05,
+      "loss": 2.866,
+      "step": 39700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.402010665067631e-05,
+      "loss": 2.88,
+      "step": 39705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4010729717717376e-05,
+      "loss": 3.0093,
+      "step": 39710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.400135264278242e-05,
+      "loss": 3.0159,
+      "step": 39715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.399197542620337e-05,
+      "loss": 2.9275,
+      "step": 39720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.398259806831217e-05,
+      "loss": 2.8932,
+      "step": 39725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3973220569440784e-05,
+      "loss": 3.0074,
+      "step": 39730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.396384292992114e-05,
+      "loss": 2.9537,
+      "step": 39735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.395446515008522e-05,
+      "loss": 2.8433,
+      "step": 39740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3945087230264976e-05,
+      "loss": 2.882,
+      "step": 39745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.393570917079238e-05,
+      "loss": 2.9964,
+      "step": 39750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.392633097199942e-05,
+      "loss": 2.9111,
+      "step": 39755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3916952634218055e-05,
+      "loss": 2.8659,
+      "step": 39760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.390757415778027e-05,
+      "loss": 3.0062,
+      "step": 39765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.389819554301807e-05,
+      "loss": 2.9252,
+      "step": 39770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.388881679026345e-05,
+      "loss": 2.9927,
+      "step": 39775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.38794378998484e-05,
+      "loss": 2.8829,
+      "step": 39780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.387005887210491e-05,
+      "loss": 2.8966,
+      "step": 39785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3860679707365014e-05,
+      "loss": 2.9137,
+      "step": 39790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.38513004059607e-05,
+      "loss": 2.9483,
+      "step": 39795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.384192096822403e-05,
+      "loss": 2.9194,
+      "step": 39800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3832541394486956e-05,
+      "loss": 2.8545,
+      "step": 39805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.382316168508158e-05,
+      "loss": 2.8406,
+      "step": 39810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.381378184033988e-05,
+      "loss": 2.9521,
+      "step": 39815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3804401860593944e-05,
+      "loss": 2.888,
+      "step": 39820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3795021746175765e-05,
+      "loss": 2.8845,
+      "step": 39825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3785641497417416e-05,
+      "loss": 2.9789,
+      "step": 39830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.377626111465095e-05,
+      "loss": 2.9504,
+      "step": 39835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3766880598208414e-05,
+      "loss": 2.8571,
+      "step": 39840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.375749994842189e-05,
+      "loss": 2.8724,
+      "step": 39845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3748119165623415e-05,
+      "loss": 2.8672,
+      "step": 39850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.373873825014506e-05,
+      "loss": 2.8477,
+      "step": 39855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.372935720231892e-05,
+      "loss": 2.8981,
+      "step": 39860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.371997602247707e-05,
+      "loss": 2.9327,
+      "step": 39865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.371059471095161e-05,
+      "loss": 2.941,
+      "step": 39870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3701213268074593e-05,
+      "loss": 2.8725,
+      "step": 39875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.369183169417815e-05,
+      "loss": 2.8906,
+      "step": 39880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.368244998959435e-05,
+      "loss": 2.9037,
+      "step": 39885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.367306815465531e-05,
+      "loss": 2.8437,
+      "step": 39890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.366368618969314e-05,
+      "loss": 2.9128,
+      "step": 39895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.365430409503995e-05,
+      "loss": 2.8053,
+      "step": 39900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3644921871027855e-05,
+      "loss": 2.9577,
+      "step": 39905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.363553951798898e-05,
+      "loss": 2.9038,
+      "step": 39910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.362615703625545e-05,
+      "loss": 2.7936,
+      "step": 39915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.36167744261594e-05,
+      "loss": 2.9377,
+      "step": 39920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.360739168803297e-05,
+      "loss": 2.9109,
+      "step": 39925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.359800882220828e-05,
+      "loss": 2.9265,
+      "step": 39930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.358862582901749e-05,
+      "loss": 2.9385,
+      "step": 39935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.357924270879276e-05,
+      "loss": 2.825,
+      "step": 39940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.35698594618662e-05,
+      "loss": 2.9822,
+      "step": 39945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3560476088570025e-05,
+      "loss": 2.864,
+      "step": 39950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.355109258923634e-05,
+      "loss": 2.9031,
+      "step": 39955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.354170896419737e-05,
+      "loss": 2.9224,
+      "step": 39960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.353232521378524e-05,
+      "loss": 2.8657,
+      "step": 39965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.352294133833215e-05,
+      "loss": 2.8771,
+      "step": 39970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3513557338170285e-05,
+      "loss": 2.9672,
+      "step": 39975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3504173213631814e-05,
+      "loss": 2.9861,
+      "step": 39980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.349478896504894e-05,
+      "loss": 2.7771,
+      "step": 39985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.348540459275383e-05,
+      "loss": 3.0076,
+      "step": 39990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3476020097078706e-05,
+      "loss": 2.8939,
+      "step": 39995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.346663547835576e-05,
+      "loss": 2.8722,
+      "step": 40000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.34572507369172e-05,
+      "loss": 2.9714,
+      "step": 40005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.344786587309525e-05,
+      "loss": 2.9159,
+      "step": 40010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.343848088722211e-05,
+      "loss": 2.9107,
+      "step": 40015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.342909577963001e-05,
+      "loss": 2.9396,
+      "step": 40020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.341971055065117e-05,
+      "loss": 2.8359,
+      "step": 40025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3410325200617816e-05,
+      "loss": 3.0078,
+      "step": 40030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.340093972986217e-05,
+      "loss": 2.944,
+      "step": 40035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3391554138716485e-05,
+      "loss": 2.8924,
+      "step": 40040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3382168427513e-05,
+      "loss": 2.8412,
+      "step": 40045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.337278259658396e-05,
+      "loss": 2.8663,
+      "step": 40050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3363396646261596e-05,
+      "loss": 2.9944,
+      "step": 40055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.335401057687819e-05,
+      "loss": 2.8754,
+      "step": 40060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3344624388765995e-05,
+      "loss": 3.015,
+      "step": 40065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.333523808225724e-05,
+      "loss": 2.9211,
+      "step": 40070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.332585165768423e-05,
+      "loss": 2.8372,
+      "step": 40075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3316465115379245e-05,
+      "loss": 2.9193,
+      "step": 40080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.330707845567451e-05,
+      "loss": 3.0104,
+      "step": 40085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3297691678902354e-05,
+      "loss": 3.0343,
+      "step": 40090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.328830478539502e-05,
+      "loss": 2.9139,
+      "step": 40095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.32789177754848e-05,
+      "loss": 2.8394,
+      "step": 40100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.326953064950402e-05,
+      "loss": 2.8348,
+      "step": 40105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.326014340778494e-05,
+      "loss": 2.9856,
+      "step": 40110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3250756050659877e-05,
+      "loss": 2.8584,
+      "step": 40115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.324136857846113e-05,
+      "loss": 2.9356,
+      "step": 40120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.323198099152101e-05,
+      "loss": 2.9938,
+      "step": 40125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.322259329017182e-05,
+      "loss": 2.7805,
+      "step": 40130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3213205474745874e-05,
+      "loss": 2.9447,
+      "step": 40135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.320381754557552e-05,
+      "loss": 2.83,
+      "step": 40140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3194429502993046e-05,
+      "loss": 3.0096,
+      "step": 40145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.318504134733079e-05,
+      "loss": 2.8826,
+      "step": 40150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.317565307892109e-05,
+      "loss": 2.9744,
+      "step": 40155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.316626469809629e-05,
+      "loss": 2.9111,
+      "step": 40160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.315687620518871e-05,
+      "loss": 2.9982,
+      "step": 40165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3147487600530723e-05,
+      "loss": 2.9362,
+      "step": 40170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.313809888445463e-05,
+      "loss": 2.9937,
+      "step": 40175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.312871005729282e-05,
+      "loss": 2.9344,
+      "step": 40180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3119321119377655e-05,
+      "loss": 2.9145,
+      "step": 40185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3109932071041445e-05,
+      "loss": 2.8539,
+      "step": 40190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3100542912616614e-05,
+      "loss": 2.8503,
+      "step": 40195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3091153644435476e-05,
+      "loss": 2.87,
+      "step": 40200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.308176426683046e-05,
+      "loss": 2.9429,
+      "step": 40205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3072374780133884e-05,
+      "loss": 2.9386,
+      "step": 40210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.306298518467815e-05,
+      "loss": 3.036,
+      "step": 40215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3053595480795656e-05,
+      "loss": 2.8432,
+      "step": 40220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3044205668818784e-05,
+      "loss": 3.0163,
+      "step": 40225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.30348157490799e-05,
+      "loss": 2.8913,
+      "step": 40230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.302542572191141e-05,
+      "loss": 2.981,
+      "step": 40235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.301603558764573e-05,
+      "loss": 2.8556,
+      "step": 40240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3006645346615245e-05,
+      "loss": 2.8744,
+      "step": 40245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.299725499915235e-05,
+      "loss": 2.9407,
+      "step": 40250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.298786454558948e-05,
+      "loss": 2.9359,
+      "step": 40255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.297847398625903e-05,
+      "loss": 2.9248,
+      "step": 40260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.296908332149343e-05,
+      "loss": 2.8628,
+      "step": 40265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.295969255162508e-05,
+      "loss": 2.9414,
+      "step": 40270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2950301676986444e-05,
+      "loss": 2.9685,
+      "step": 40275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2940910697909904e-05,
+      "loss": 2.8849,
+      "step": 40280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.293151961472791e-05,
+      "loss": 2.8472,
+      "step": 40285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.292212842777291e-05,
+      "loss": 2.8614,
+      "step": 40290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.291273713737732e-05,
+      "loss": 2.927,
+      "step": 40295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.29033457438736e-05,
+      "loss": 3.0145,
+      "step": 40300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.289395424759419e-05,
+      "loss": 2.8465,
+      "step": 40305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.288456264887156e-05,
+      "loss": 2.9013,
+      "step": 40310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.287517094803812e-05,
+      "loss": 2.9301,
+      "step": 40315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.286577914542636e-05,
+      "loss": 2.9548,
+      "step": 40320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.285638724136874e-05,
+      "loss": 2.9958,
+      "step": 40325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2846995236197716e-05,
+      "loss": 2.8651,
+      "step": 40330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2837603130245775e-05,
+      "loss": 2.9869,
+      "step": 40335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.282821092384534e-05,
+      "loss": 2.9414,
+      "step": 40340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2818818617328934e-05,
+      "loss": 2.9305,
+      "step": 40345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.280942621102901e-05,
+      "loss": 2.8631,
+      "step": 40350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.280003370527806e-05,
+      "loss": 2.8812,
+      "step": 40355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.279064110040859e-05,
+      "loss": 2.8836,
+      "step": 40360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2781248396753045e-05,
+      "loss": 2.9294,
+      "step": 40365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.277185559464397e-05,
+      "loss": 2.7999,
+      "step": 40370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.27624626944138e-05,
+      "loss": 2.8508,
+      "step": 40375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2753069696395093e-05,
+      "loss": 2.909,
+      "step": 40380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2743676600920324e-05,
+      "loss": 2.9201,
+      "step": 40385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.273428340832199e-05,
+      "loss": 2.9354,
+      "step": 40390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2724890118932635e-05,
+      "loss": 2.8943,
+      "step": 40395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2715496733084725e-05,
+      "loss": 2.9079,
+      "step": 40400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.270610325111083e-05,
+      "loss": 2.8814,
+      "step": 40405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.269670967334341e-05,
+      "loss": 2.8109,
+      "step": 40410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2687316000115064e-05,
+      "loss": 2.8215,
+      "step": 40415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.267792223175825e-05,
+      "loss": 2.9191,
+      "step": 40420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.266852836860553e-05,
+      "loss": 2.9095,
+      "step": 40425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2659134410989444e-05,
+      "loss": 2.8694,
+      "step": 40430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2649740359242503e-05,
+      "loss": 2.8359,
+      "step": 40435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.264034621369729e-05,
+      "loss": 2.8253,
+      "step": 40440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2630951974686305e-05,
+      "loss": 2.8745,
+      "step": 40445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2621557642542104e-05,
+      "loss": 2.8942,
+      "step": 40450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.261216321759725e-05,
+      "loss": 2.9215,
+      "step": 40455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.26027687001843e-05,
+      "loss": 2.8883,
+      "step": 40460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.259337409063582e-05,
+      "loss": 2.8958,
+      "step": 40465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2583979389284334e-05,
+      "loss": 2.8856,
+      "step": 40470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.257458459646243e-05,
+      "loss": 2.8772,
+      "step": 40475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2565189712502674e-05,
+      "loss": 2.872,
+      "step": 40480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2555794737737616e-05,
+      "loss": 2.9295,
+      "step": 40485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.254639967249987e-05,
+      "loss": 2.909,
+      "step": 40490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.253700451712198e-05,
+      "loss": 2.8641,
+      "step": 40495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2527609271936526e-05,
+      "loss": 2.8133,
+      "step": 40500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.251821393727609e-05,
+      "loss": 2.9166,
+      "step": 40505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.25088185134733e-05,
+      "loss": 2.9581,
+      "step": 40510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2499423000860684e-05,
+      "loss": 2.858,
+      "step": 40515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.249002739977087e-05,
+      "loss": 2.9403,
+      "step": 40520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.248063171053645e-05,
+      "loss": 3.0387,
+      "step": 40525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2471235933490016e-05,
+      "loss": 2.9363,
+      "step": 40530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.246184006896417e-05,
+      "loss": 2.9886,
+      "step": 40535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.245244411729152e-05,
+      "loss": 2.9673,
+      "step": 40540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2443048078804656e-05,
+      "loss": 2.9219,
+      "step": 40545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.243365195383622e-05,
+      "loss": 2.9631,
+      "step": 40550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.242425574271882e-05,
+      "loss": 2.988,
+      "step": 40555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2414859445785044e-05,
+      "loss": 2.9045,
+      "step": 40560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2405463063367535e-05,
+      "loss": 3.0129,
+      "step": 40565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.239606659579892e-05,
+      "loss": 2.9308,
+      "step": 40570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.238667004341182e-05,
+      "loss": 2.9238,
+      "step": 40575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.237727340653886e-05,
+      "loss": 2.8877,
+      "step": 40580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.236787668551267e-05,
+      "loss": 2.8997,
+      "step": 40585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.235847988066589e-05,
+      "loss": 2.9224,
+      "step": 40590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.234908299233114e-05,
+      "loss": 2.917,
+      "step": 40595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2339686020841095e-05,
+      "loss": 2.9112,
+      "step": 40600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2330288966528395e-05,
+      "loss": 2.935,
+      "step": 40605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.232089182972565e-05,
+      "loss": 2.861,
+      "step": 40610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2311494610765554e-05,
+      "loss": 2.9359,
+      "step": 40615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2302097309980727e-05,
+      "loss": 2.8474,
+      "step": 40620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.229269992770383e-05,
+      "loss": 2.8837,
+      "step": 40625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2283302464267534e-05,
+      "loss": 2.7955,
+      "step": 40630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.22739049200045e-05,
+      "loss": 2.929,
+      "step": 40635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.226450729524739e-05,
+      "loss": 2.9457,
+      "step": 40640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2255109590328844e-05,
+      "loss": 2.9834,
+      "step": 40645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.224571180558158e-05,
+      "loss": 2.8896,
+      "step": 40650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.223631394133822e-05,
+      "loss": 2.8876,
+      "step": 40655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.222691599793148e-05,
+      "loss": 2.921,
+      "step": 40660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.221751797569404e-05,
+      "loss": 2.9561,
+      "step": 40665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.220811987495854e-05,
+      "loss": 2.861,
+      "step": 40670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.219872169605771e-05,
+      "loss": 2.8743,
+      "step": 40675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.21893234393242e-05,
+      "loss": 2.8463,
+      "step": 40680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.217992510509072e-05,
+      "loss": 2.8846,
+      "step": 40685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.217052669368996e-05,
+      "loss": 2.7723,
+      "step": 40690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.21611282054546e-05,
+      "loss": 3.0088,
+      "step": 40695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.215172964071735e-05,
+      "loss": 2.8152,
+      "step": 40700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2142330999810904e-05,
+      "loss": 2.8799,
+      "step": 40705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2132932283067994e-05,
+      "loss": 2.864,
+      "step": 40710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2123533490821285e-05,
+      "loss": 2.8423,
+      "step": 40715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.211413462340351e-05,
+      "loss": 2.9613,
+      "step": 40720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2104735681147364e-05,
+      "loss": 2.8465,
+      "step": 40725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2095336664385574e-05,
+      "loss": 2.9426,
+      "step": 40730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.208593757345086e-05,
+      "loss": 2.9464,
+      "step": 40735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.207653840867592e-05,
+      "loss": 2.831,
+      "step": 40740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2067139170393486e-05,
+      "loss": 2.7727,
+      "step": 40745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2057739858936296e-05,
+      "loss": 2.7693,
+      "step": 40750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.204834047463706e-05,
+      "loss": 2.8384,
+      "step": 40755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2038941017828505e-05,
+      "loss": 2.8845,
+      "step": 40760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.202954148884337e-05,
+      "loss": 3.0128,
+      "step": 40765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2020141888014406e-05,
+      "loss": 2.845,
+      "step": 40770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2010742215674316e-05,
+      "loss": 2.8075,
+      "step": 40775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2001342472155865e-05,
+      "loss": 2.8801,
+      "step": 40780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.199194265779178e-05,
+      "loss": 2.9138,
+      "step": 40785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.198254277291482e-05,
+      "loss": 2.9777,
+      "step": 40790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.197314281785771e-05,
+      "loss": 2.9095,
+      "step": 40795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.196374279295324e-05,
+      "loss": 2.9226,
+      "step": 40800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.195434269853412e-05,
+      "loss": 2.7852,
+      "step": 40805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1944942534933114e-05,
+      "loss": 2.9162,
+      "step": 40810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1935542302483e-05,
+      "loss": 2.8495,
+      "step": 40815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1926142001516506e-05,
+      "loss": 2.885,
+      "step": 40820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.191674163236643e-05,
+      "loss": 2.9555,
+      "step": 40825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.19073411953655e-05,
+      "loss": 2.8881,
+      "step": 40830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.189794069084649e-05,
+      "loss": 2.8518,
+      "step": 40835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.18885401191422e-05,
+      "loss": 2.8297,
+      "step": 40840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.187913948058536e-05,
+      "loss": 2.8873,
+      "step": 40845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.186973877550878e-05,
+      "loss": 2.9657,
+      "step": 40850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.186033800424521e-05,
+      "loss": 2.8774,
+      "step": 40855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1850937167127444e-05,
+      "loss": 2.9234,
+      "step": 40860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.184153626448825e-05,
+      "loss": 2.8557,
+      "step": 40865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.183213529666041e-05,
+      "loss": 2.9911,
+      "step": 40870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1822734263976735e-05,
+      "loss": 2.8576,
+      "step": 40875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.181333316676997e-05,
+      "loss": 2.9622,
+      "step": 40880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1803932005372955e-05,
+      "loss": 2.9415,
+      "step": 40885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1794530780118444e-05,
+      "loss": 2.9377,
+      "step": 40890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1785129491339234e-05,
+      "loss": 2.8742,
+      "step": 40895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.177572813936814e-05,
+      "loss": 2.879,
+      "step": 40900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.176632672453795e-05,
+      "loss": 2.8508,
+      "step": 40905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1756925247181474e-05,
+      "loss": 3.0187,
+      "step": 40910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1747523707631506e-05,
+      "loss": 2.9273,
+      "step": 40915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1738122106220864e-05,
+      "loss": 2.9247,
+      "step": 40920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.172872044328233e-05,
+      "loss": 2.8821,
+      "step": 40925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.171931871914875e-05,
+      "loss": 2.912,
+      "step": 40930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.170991693415289e-05,
+      "loss": 2.8664,
+      "step": 40935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.170051508862761e-05,
+      "loss": 2.8184,
+      "step": 40940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.169111318290569e-05,
+      "loss": 3.0344,
+      "step": 40945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.168171121731997e-05,
+      "loss": 2.8947,
+      "step": 40950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.167230919220328e-05,
+      "loss": 2.9678,
+      "step": 40955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1662907107888425e-05,
+      "loss": 3.0044,
+      "step": 40960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.165350496470823e-05,
+      "loss": 2.8954,
+      "step": 40965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.164410276299553e-05,
+      "loss": 2.7232,
+      "step": 40970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1634700503083146e-05,
+      "loss": 2.981,
+      "step": 40975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.162529818530392e-05,
+      "loss": 2.8747,
+      "step": 40980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1615895809990666e-05,
+      "loss": 2.9373,
+      "step": 40985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.160649337747623e-05,
+      "loss": 2.993,
+      "step": 40990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.159709088809346e-05,
+      "loss": 2.7919,
+      "step": 40995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.158768834217519e-05,
+      "loss": 2.912,
+      "step": 41000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1578285740054244e-05,
+      "loss": 2.9365,
+      "step": 41005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.156888308206348e-05,
+      "loss": 3.0659,
+      "step": 41010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.155948036853575e-05,
+      "loss": 2.924,
+      "step": 41015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1550077599803895e-05,
+      "loss": 2.862,
+      "step": 41020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.154067477620076e-05,
+      "loss": 2.8886,
+      "step": 41025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.153127189805919e-05,
+      "loss": 2.8992,
+      "step": 41030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.152186896571204e-05,
+      "loss": 2.8795,
+      "step": 41035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.151246597949217e-05,
+      "loss": 2.9585,
+      "step": 41040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.150306293973245e-05,
+      "loss": 2.9523,
+      "step": 41045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.149365984676573e-05,
+      "loss": 2.9103,
+      "step": 41050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.148425670092486e-05,
+      "loss": 2.827,
+      "step": 41055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1474853502542705e-05,
+      "loss": 2.904,
+      "step": 41060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.146545025195214e-05,
+      "loss": 2.7796,
+      "step": 41065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.145604694948602e-05,
+      "loss": 2.9091,
+      "step": 41070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.144664359547723e-05,
+      "loss": 2.9535,
+      "step": 41075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.143724019025861e-05,
+      "loss": 2.9338,
+      "step": 41080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.142783673416306e-05,
+      "loss": 2.9143,
+      "step": 41085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.141843322752342e-05,
+      "loss": 2.9396,
+      "step": 41090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.140902967067261e-05,
+      "loss": 2.8779,
+      "step": 41095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1399626063943486e-05,
+      "loss": 2.9149,
+      "step": 41100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.139022240766892e-05,
+      "loss": 2.9689,
+      "step": 41105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.13808187021818e-05,
+      "loss": 2.8727,
+      "step": 41110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1371414947815e-05,
+      "loss": 2.8265,
+      "step": 41115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.136201114490142e-05,
+      "loss": 2.9028,
+      "step": 41120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.135260729377392e-05,
+      "loss": 2.8933,
+      "step": 41125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1343203394765426e-05,
+      "loss": 2.9334,
+      "step": 41130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.133379944820879e-05,
+      "loss": 2.9435,
+      "step": 41135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1324395454436916e-05,
+      "loss": 2.9052,
+      "step": 41140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.13149914137827e-05,
+      "loss": 2.9146,
+      "step": 41145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.130558732657903e-05,
+      "loss": 2.9118,
+      "step": 41150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.129618319315882e-05,
+      "loss": 2.7913,
+      "step": 41155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.128677901385493e-05,
+      "loss": 2.8951,
+      "step": 41160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.127737478900031e-05,
+      "loss": 2.8725,
+      "step": 41165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.126797051892781e-05,
+      "loss": 2.8907,
+      "step": 41170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.125856620397036e-05,
+      "loss": 2.8709,
+      "step": 41175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.124916184446087e-05,
+      "loss": 2.8989,
+      "step": 41180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.123975744073223e-05,
+      "loss": 2.836,
+      "step": 41185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.123035299311734e-05,
+      "loss": 3.0186,
+      "step": 41190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.122094850194913e-05,
+      "loss": 2.9064,
+      "step": 41195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1211543967560505e-05,
+      "loss": 2.8539,
+      "step": 41200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.120213939028435e-05,
+      "loss": 2.8629,
+      "step": 41205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.119273477045362e-05,
+      "loss": 2.8779,
+      "step": 41210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.118333010840121e-05,
+      "loss": 2.9828,
+      "step": 41215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1173925404460024e-05,
+      "loss": 2.9019,
+      "step": 41220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1164520658963e-05,
+      "loss": 2.9367,
+      "step": 41225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.115511587224304e-05,
+      "loss": 2.8928,
+      "step": 41230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1145711044633084e-05,
+      "loss": 2.8937,
+      "step": 41235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.113630617646602e-05,
+      "loss": 2.9959,
+      "step": 41240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.112690126807481e-05,
+      "loss": 2.9503,
+      "step": 41245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.111749631979236e-05,
+      "loss": 2.8232,
+      "step": 41250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.110809133195159e-05,
+      "loss": 2.9072,
+      "step": 41255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.109868630488546e-05,
+      "loss": 2.8682,
+      "step": 41260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.108928123892685e-05,
+      "loss": 3.0085,
+      "step": 41265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1079876134408746e-05,
+      "loss": 2.9409,
+      "step": 41270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.107047099166403e-05,
+      "loss": 2.8939,
+      "step": 41275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1061065811025654e-05,
+      "loss": 2.9054,
+      "step": 41280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1051660592826564e-05,
+      "loss": 2.856,
+      "step": 41285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.104225533739967e-05,
+      "loss": 3.0415,
+      "step": 41290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.103285004507794e-05,
+      "loss": 2.9389,
+      "step": 41295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.102344471619428e-05,
+      "loss": 2.9099,
+      "step": 41300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.101403935108167e-05,
+      "loss": 2.9881,
+      "step": 41305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.100463395007301e-05,
+      "loss": 2.9028,
+      "step": 41310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.099522851350127e-05,
+      "loss": 2.9439,
+      "step": 41315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0985823041699376e-05,
+      "loss": 3.0134,
+      "step": 41320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.097641753500029e-05,
+      "loss": 2.9818,
+      "step": 41325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0967011993736945e-05,
+      "loss": 2.951,
+      "step": 41330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0957606418242275e-05,
+      "loss": 3.0182,
+      "step": 41335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.094820080884926e-05,
+      "loss": 2.9222,
+      "step": 41340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.093879516589083e-05,
+      "loss": 2.9674,
+      "step": 41345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.092938948969995e-05,
+      "loss": 2.8512,
+      "step": 41350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.091998378060956e-05,
+      "loss": 2.9721,
+      "step": 41355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0910578038952606e-05,
+      "loss": 2.9192,
+      "step": 41360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.090117226506206e-05,
+      "loss": 2.9295,
+      "step": 41365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.089176645927086e-05,
+      "loss": 2.9196,
+      "step": 41370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0882360621911996e-05,
+      "loss": 3.0086,
+      "step": 41375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.087295475331838e-05,
+      "loss": 2.9525,
+      "step": 41380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0863548853823e-05,
+      "loss": 3.0115,
+      "step": 41385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.08541429237588e-05,
+      "loss": 2.8975,
+      "step": 41390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.084473696345875e-05,
+      "loss": 2.9407,
+      "step": 41395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.083533097325583e-05,
+      "loss": 2.9065,
+      "step": 41400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0825924953482976e-05,
+      "loss": 2.914,
+      "step": 41405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0816518904473165e-05,
+      "loss": 2.8362,
+      "step": 41410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0807112826559344e-05,
+      "loss": 2.7742,
+      "step": 41415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.07977067200745e-05,
+      "loss": 3.0396,
+      "step": 41420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0788300585351615e-05,
+      "loss": 2.8498,
+      "step": 41425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.077889442272361e-05,
+      "loss": 2.907,
+      "step": 41430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.076948823252349e-05,
+      "loss": 2.8464,
+      "step": 41435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0760082015084206e-05,
+      "loss": 2.9244,
+      "step": 41440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.075067577073875e-05,
+      "loss": 2.891,
+      "step": 41445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.074126949982008e-05,
+      "loss": 2.8962,
+      "step": 41450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.073186320266118e-05,
+      "loss": 3.0067,
+      "step": 41455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0722456879595e-05,
+      "loss": 2.9449,
+      "step": 41460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.071305053095454e-05,
+      "loss": 2.8613,
+      "step": 41465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0703644157072774e-05,
+      "loss": 2.9358,
+      "step": 41470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.069423775828266e-05,
+      "loss": 2.9815,
+      "step": 41475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.068483133491719e-05,
+      "loss": 2.9841,
+      "step": 41480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.067542488730933e-05,
+      "loss": 2.8833,
+      "step": 41485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0666018415792084e-05,
+      "loss": 2.831,
+      "step": 41490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0656611920698416e-05,
+      "loss": 2.9688,
+      "step": 41495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0647205402361296e-05,
+      "loss": 2.8531,
+      "step": 41500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.063779886111374e-05,
+      "loss": 2.8753,
+      "step": 41505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0628392297288686e-05,
+      "loss": 2.7814,
+      "step": 41510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.061898571121916e-05,
+      "loss": 2.9172,
+      "step": 41515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0609579103238104e-05,
+      "loss": 2.9314,
+      "step": 41520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0600172473678545e-05,
+      "loss": 2.8877,
+      "step": 41525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.059076582287344e-05,
+      "loss": 2.9055,
+      "step": 41530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.058135915115577e-05,
+      "loss": 2.9909,
+      "step": 41535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.057195245885856e-05,
+      "loss": 2.9792,
+      "step": 41540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.056254574631476e-05,
+      "loss": 2.8078,
+      "step": 41545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.055313901385739e-05,
+      "loss": 2.9311,
+      "step": 41550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.054373226181941e-05,
+      "loss": 2.9218,
+      "step": 41555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.053432549053383e-05,
+      "loss": 2.9524,
+      "step": 41560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0524918700333635e-05,
+      "loss": 2.8233,
+      "step": 41565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0515511891551804e-05,
+      "loss": 2.9537,
+      "step": 41570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.050610506452136e-05,
+      "loss": 2.8801,
+      "step": 41575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0496698219575255e-05,
+      "loss": 2.8663,
+      "step": 41580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.048729135704651e-05,
+      "loss": 2.9171,
+      "step": 41585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.047788447726811e-05,
+      "loss": 2.8358,
+      "step": 41590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.046847758057307e-05,
+      "loss": 2.9535,
+      "step": 41595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.045907066729435e-05,
+      "loss": 2.8803,
+      "step": 41600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.044966373776496e-05,
+      "loss": 2.9857,
+      "step": 41605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.044025679231792e-05,
+      "loss": 2.9744,
+      "step": 41610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0430849831286184e-05,
+      "loss": 2.855,
+      "step": 41615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0421442855002795e-05,
+      "loss": 2.9106,
+      "step": 41620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.041203586380071e-05,
+      "loss": 2.876,
+      "step": 41625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.040262885801295e-05,
+      "loss": 2.9771,
+      "step": 41630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.03932218379725e-05,
+      "loss": 2.9014,
+      "step": 41635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.038381480401238e-05,
+      "loss": 2.8325,
+      "step": 41640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.037440775646559e-05,
+      "loss": 2.8483,
+      "step": 41645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.036500069566511e-05,
+      "loss": 2.907,
+      "step": 41650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.035559362194396e-05,
+      "loss": 2.9958,
+      "step": 41655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.034618653563512e-05,
+      "loss": 2.7899,
+      "step": 41660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0336779437071615e-05,
+      "loss": 2.8083,
+      "step": 41665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.032737232658643e-05,
+      "loss": 2.9086,
+      "step": 41670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0317965204512574e-05,
+      "loss": 2.8407,
+      "step": 41675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.030855807118306e-05,
+      "loss": 2.8692,
+      "step": 41680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.029915092693086e-05,
+      "loss": 2.901,
+      "step": 41685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.028974377208903e-05,
+      "loss": 2.8153,
+      "step": 41690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.028033660699053e-05,
+      "loss": 2.8221,
+      "step": 41695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.027092943196837e-05,
+      "loss": 2.8704,
+      "step": 41700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.026152224735558e-05,
+      "loss": 2.8883,
+      "step": 41705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.025211505348515e-05,
+      "loss": 2.8937,
+      "step": 41710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.024270785069009e-05,
+      "loss": 2.7984,
+      "step": 41715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0233300639303395e-05,
+      "loss": 2.9219,
+      "step": 41720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.022389341965809e-05,
+      "loss": 2.8422,
+      "step": 41725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.021448619208715e-05,
+      "loss": 2.9557,
+      "step": 41730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.02050789569236e-05,
+      "loss": 2.9417,
+      "step": 41735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.019567171450046e-05,
+      "loss": 2.945,
+      "step": 41740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0186264465150724e-05,
+      "loss": 2.9433,
+      "step": 41745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.017685720920741e-05,
+      "loss": 2.9682,
+      "step": 41750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.016744994700351e-05,
+      "loss": 2.9137,
+      "step": 41755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.015804267887205e-05,
+      "loss": 3.0105,
+      "step": 41760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0148635405146015e-05,
+      "loss": 2.9354,
+      "step": 41765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0139228126158435e-05,
+      "loss": 2.9147,
+      "step": 41770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0129820842242316e-05,
+      "loss": 2.9579,
+      "step": 41775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0120413553730636e-05,
+      "loss": 3.0003,
+      "step": 41780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.011100626095646e-05,
+      "loss": 2.8934,
+      "step": 41785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.010159896425276e-05,
+      "loss": 2.9254,
+      "step": 41790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.009219166395256e-05,
+      "loss": 2.8211,
+      "step": 41795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.008278436038885e-05,
+      "loss": 2.8713,
+      "step": 41800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.007337705389466e-05,
+      "loss": 2.8996,
+      "step": 41805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0063969744802985e-05,
+      "loss": 2.8635,
+      "step": 41810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.005456243344685e-05,
+      "loss": 2.9206,
+      "step": 41815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.004515512015926e-05,
+      "loss": 2.9601,
+      "step": 41820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.00357478052732e-05,
+      "loss": 3.0197,
+      "step": 41825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.002634048912171e-05,
+      "loss": 2.9467,
+      "step": 41830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0016933172037795e-05,
+      "loss": 2.8829,
+      "step": 41835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.000752585435448e-05,
+      "loss": 2.9359,
+      "step": 41840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.999811853640474e-05,
+      "loss": 2.9673,
+      "step": 41845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.99887112185216e-05,
+      "loss": 2.8821,
+      "step": 41850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.997930390103805e-05,
+      "loss": 2.8106,
+      "step": 41855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.996989658428715e-05,
+      "loss": 2.9116,
+      "step": 41860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9960489268601886e-05,
+      "loss": 2.9823,
+      "step": 41865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.995108195431525e-05,
+      "loss": 2.8884,
+      "step": 41870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9941674641760285e-05,
+      "loss": 2.8452,
+      "step": 41875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9932267331269964e-05,
+      "loss": 2.8961,
+      "step": 41880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.992286002317732e-05,
+      "loss": 2.9006,
+      "step": 41885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9913452717815365e-05,
+      "loss": 2.8479,
+      "step": 41890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.99040454155171e-05,
+      "loss": 2.8299,
+      "step": 41895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.989463811661554e-05,
+      "loss": 2.9297,
+      "step": 41900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.988523082144368e-05,
+      "loss": 2.8986,
+      "step": 41905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.987582353033454e-05,
+      "loss": 2.8759,
+      "step": 41910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.986641624362115e-05,
+      "loss": 2.8957,
+      "step": 41915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.985700896163647e-05,
+      "loss": 2.9528,
+      "step": 41920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.984760168471356e-05,
+      "loss": 2.9064,
+      "step": 41925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.983819441318539e-05,
+      "loss": 2.8224,
+      "step": 41930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9828787147384984e-05,
+      "loss": 2.8752,
+      "step": 41935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.981937988764537e-05,
+      "loss": 2.9506,
+      "step": 41940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.980997263429952e-05,
+      "loss": 2.8593,
+      "step": 41945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9800565387680464e-05,
+      "loss": 2.8744,
+      "step": 41950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.979115814812119e-05,
+      "loss": 2.9377,
+      "step": 41955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9781750915954725e-05,
+      "loss": 2.9245,
+      "step": 41960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.977234369151408e-05,
+      "loss": 2.8515,
+      "step": 41965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.976293647513225e-05,
+      "loss": 2.9396,
+      "step": 41970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.975352926714225e-05,
+      "loss": 2.8599,
+      "step": 41975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.974412206787708e-05,
+      "loss": 2.9012,
+      "step": 41980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9734714877669744e-05,
+      "loss": 2.8136,
+      "step": 41985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.972530769685324e-05,
+      "loss": 2.935,
+      "step": 41990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.97159005257606e-05,
+      "loss": 2.8891,
+      "step": 41995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.97064933647248e-05,
+      "loss": 2.9381,
+      "step": 42000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.969708621407887e-05,
+      "loss": 2.8008,
+      "step": 42005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.968767907415578e-05,
+      "loss": 2.868,
+      "step": 42010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.967827194528856e-05,
+      "loss": 2.9281,
+      "step": 42015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.966886482781022e-05,
+      "loss": 2.9555,
+      "step": 42020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.965945772205375e-05,
+      "loss": 2.9771,
+      "step": 42025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.965005062835216e-05,
+      "loss": 2.8174,
+      "step": 42030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9640643547038426e-05,
+      "loss": 2.9794,
+      "step": 42035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9631236478445576e-05,
+      "loss": 2.9281,
+      "step": 42040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9621829422906616e-05,
+      "loss": 2.8425,
+      "step": 42045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.961242238075452e-05,
+      "loss": 2.9581,
+      "step": 42050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9603015352322304e-05,
+      "loss": 2.8381,
+      "step": 42055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.959360833794298e-05,
+      "loss": 3.0273,
+      "step": 42060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.958420133794954e-05,
+      "loss": 2.8197,
+      "step": 42065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.957479435267496e-05,
+      "loss": 2.9964,
+      "step": 42070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9565387382452285e-05,
+      "loss": 2.9872,
+      "step": 42075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.955598042761446e-05,
+      "loss": 2.8531,
+      "step": 42080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9546573488494516e-05,
+      "loss": 2.8449,
+      "step": 42085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9537166565425444e-05,
+      "loss": 2.8992,
+      "step": 42090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.952775965874024e-05,
+      "loss": 2.8495,
+      "step": 42095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.951835276877189e-05,
+      "loss": 2.8897,
+      "step": 42100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.95089458958534e-05,
+      "loss": 2.9402,
+      "step": 42105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.949953904031776e-05,
+      "loss": 2.8328,
+      "step": 42110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9490132202497954e-05,
+      "loss": 2.9368,
+      "step": 42115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.948072538272699e-05,
+      "loss": 2.8434,
+      "step": 42120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9471318581337866e-05,
+      "loss": 2.8169,
+      "step": 42125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.946191179866355e-05,
+      "loss": 2.8424,
+      "step": 42130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.945250503503706e-05,
+      "loss": 2.8654,
+      "step": 42135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.944309829079136e-05,
+      "loss": 2.9656,
+      "step": 42140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.943369156625945e-05,
+      "loss": 2.9917,
+      "step": 42145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.942428486177434e-05,
+      "loss": 2.8314,
+      "step": 42150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.941487817766897e-05,
+      "loss": 2.8473,
+      "step": 42155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.940547151427637e-05,
+      "loss": 2.9291,
+      "step": 42160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9396064871929524e-05,
+      "loss": 2.9175,
+      "step": 42165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.938665825096141e-05,
+      "loss": 2.866,
+      "step": 42170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937725165170501e-05,
+      "loss": 2.9486,
+      "step": 42175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936784507449332e-05,
+      "loss": 2.9265,
+      "step": 42180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.93584385196593e-05,
+      "loss": 2.8777,
+      "step": 42185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9349031987535956e-05,
+      "loss": 2.9057,
+      "step": 42190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.933962547845627e-05,
+      "loss": 2.9555,
+      "step": 42195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.93302189927532e-05,
+      "loss": 2.9053,
+      "step": 42200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9320812530759766e-05,
+      "loss": 2.896,
+      "step": 42205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9311406092808905e-05,
+      "loss": 2.8992,
+      "step": 42210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.930199967923362e-05,
+      "loss": 2.8092,
+      "step": 42215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.929259329036689e-05,
+      "loss": 2.9638,
+      "step": 42220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.928318692654168e-05,
+      "loss": 2.8966,
+      "step": 42225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.927378058809099e-05,
+      "loss": 2.9118,
+      "step": 42230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9264374275347756e-05,
+      "loss": 2.9502,
+      "step": 42235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9254967988644996e-05,
+      "loss": 2.9228,
+      "step": 42240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.924556172831565e-05,
+      "loss": 2.9046,
+      "step": 42245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.923615549469269e-05,
+      "loss": 2.8974,
+      "step": 42250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.922674928810912e-05,
+      "loss": 2.8849,
+      "step": 42255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.92173431088979e-05,
+      "loss": 2.8482,
+      "step": 42260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.920793695739198e-05,
+      "loss": 2.9306,
+      "step": 42265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.919853083392434e-05,
+      "loss": 2.8831,
+      "step": 42270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918912473882796e-05,
+      "loss": 2.911,
+      "step": 42275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9179718672435796e-05,
+      "loss": 2.8715,
+      "step": 42280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.917031263508081e-05,
+      "loss": 2.8797,
+      "step": 42285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9160906627095966e-05,
+      "loss": 2.8701,
+      "step": 42290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915150064881424e-05,
+      "loss": 2.8416,
+      "step": 42295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9142094700568586e-05,
+      "loss": 2.9161,
+      "step": 42300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.913268878269196e-05,
+      "loss": 2.8616,
+      "step": 42305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.912328289551734e-05,
+      "loss": 2.8377,
+      "step": 42310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.911387703937766e-05,
+      "loss": 2.8841,
+      "step": 42315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.91044712146059e-05,
+      "loss": 2.9311,
+      "step": 42320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9095065421535014e-05,
+      "loss": 2.9403,
+      "step": 42325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9085659660497953e-05,
+      "loss": 2.969,
+      "step": 42330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9076253931827675e-05,
+      "loss": 2.91,
+      "step": 42335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9066848235857124e-05,
+      "loss": 2.8637,
+      "step": 42340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.905744257291927e-05,
+      "loss": 2.8777,
+      "step": 42345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.904803694334704e-05,
+      "loss": 3.007,
+      "step": 42350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.903863134747342e-05,
+      "loss": 2.892,
+      "step": 42355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.902922578563133e-05,
+      "loss": 3.0227,
+      "step": 42360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.901982025815374e-05,
+      "loss": 2.9264,
+      "step": 42365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.901041476537356e-05,
+      "loss": 2.9364,
+      "step": 42370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.900100930762377e-05,
+      "loss": 2.9619,
+      "step": 42375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.899160388523732e-05,
+      "loss": 2.8592,
+      "step": 42380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.898219849854712e-05,
+      "loss": 2.8884,
+      "step": 42385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.897279314788614e-05,
+      "loss": 2.9469,
+      "step": 42390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8963387833587306e-05,
+      "loss": 2.8527,
+      "step": 42395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.895398255598355e-05,
+      "loss": 2.9393,
+      "step": 42400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.894457731540784e-05,
+      "loss": 2.958,
+      "step": 42405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.893517211219308e-05,
+      "loss": 2.9713,
+      "step": 42410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.892576694667222e-05,
+      "loss": 2.896,
+      "step": 42415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.891636181917819e-05,
+      "loss": 2.9908,
+      "step": 42420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8906956730043925e-05,
+      "loss": 2.8603,
+      "step": 42425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.889755167960237e-05,
+      "loss": 2.9398,
+      "step": 42430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8888146668186416e-05,
+      "loss": 2.943,
+      "step": 42435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8878741696129036e-05,
+      "loss": 2.8671,
+      "step": 42440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.886933676376313e-05,
+      "loss": 2.9099,
+      "step": 42445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.885993187142162e-05,
+      "loss": 2.874,
+      "step": 42450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8850527019437444e-05,
+      "loss": 2.9534,
+      "step": 42455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.884112220814355e-05,
+      "loss": 2.9181,
+      "step": 42460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.88317174378728e-05,
+      "loss": 2.8305,
+      "step": 42465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.882231270895816e-05,
+      "loss": 2.9422,
+      "step": 42470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.881290802173254e-05,
+      "loss": 2.8714,
+      "step": 42475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.880350337652884e-05,
+      "loss": 2.7959,
+      "step": 42480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8794098773680005e-05,
+      "loss": 2.9847,
+      "step": 42485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.878469421351891e-05,
+      "loss": 2.8835,
+      "step": 42490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8775289696378505e-05,
+      "loss": 2.8258,
+      "step": 42495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.876588522259167e-05,
+      "loss": 2.9852,
+      "step": 42500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8756480792491335e-05,
+      "loss": 2.9664,
+      "step": 42505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.874707640641041e-05,
+      "loss": 2.7982,
+      "step": 42510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8737672064681776e-05,
+      "loss": 2.9799,
+      "step": 42515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8728267767638376e-05,
+      "loss": 2.9794,
+      "step": 42520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.871886351561307e-05,
+      "loss": 2.8782,
+      "step": 42525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.870945930893879e-05,
+      "loss": 2.863,
+      "step": 42530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8700055147948434e-05,
+      "loss": 2.9106,
+      "step": 42535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.869065103297489e-05,
+      "loss": 2.8969,
+      "step": 42540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8681246964351066e-05,
+      "loss": 2.9566,
+      "step": 42545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.867184294240981e-05,
+      "loss": 2.9531,
+      "step": 42550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8662438967484096e-05,
+      "loss": 2.8487,
+      "step": 42555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8653035039906764e-05,
+      "loss": 3.0496,
+      "step": 42560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8643631160010725e-05,
+      "loss": 2.7921,
+      "step": 42565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8634227328128844e-05,
+      "loss": 2.8979,
+      "step": 42570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8624823544594025e-05,
+      "loss": 3.0037,
+      "step": 42575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.861541980973916e-05,
+      "loss": 2.9023,
+      "step": 42580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.860601612389711e-05,
+      "loss": 2.8685,
+      "step": 42585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.859661248740078e-05,
+      "loss": 2.7932,
+      "step": 42590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8587208900583024e-05,
+      "loss": 2.9196,
+      "step": 42595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.857780536377674e-05,
+      "loss": 2.9647,
+      "step": 42600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8568401877314806e-05,
+      "loss": 3.0029,
+      "step": 42605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.855899844153008e-05,
+      "loss": 2.837,
+      "step": 42610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8549595056755456e-05,
+      "loss": 2.9453,
+      "step": 42615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.854019172332378e-05,
+      "loss": 2.9245,
+      "step": 42620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8530788441567945e-05,
+      "loss": 2.8637,
+      "step": 42625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8521385211820794e-05,
+      "loss": 2.8071,
+      "step": 42630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.851198203441521e-05,
+      "loss": 2.9946,
+      "step": 42635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.850257890968406e-05,
+      "loss": 2.9308,
+      "step": 42640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.849317583796018e-05,
+      "loss": 2.9134,
+      "step": 42645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.848377281957645e-05,
+      "loss": 2.9528,
+      "step": 42650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.847436985486573e-05,
+      "loss": 2.8827,
+      "step": 42655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8464966944160885e-05,
+      "loss": 2.9161,
+      "step": 42660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.845556408779474e-05,
+      "loss": 2.8445,
+      "step": 42665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.844616128610018e-05,
+      "loss": 2.8493,
+      "step": 42670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8436758539410024e-05,
+      "loss": 2.9208,
+      "step": 42675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.842735584805714e-05,
+      "loss": 2.9082,
+      "step": 42680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.841795321237437e-05,
+      "loss": 2.8631,
+      "step": 42685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.840855063269455e-05,
+      "loss": 2.8731,
+      "step": 42690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8399148109350544e-05,
+      "loss": 2.8203,
+      "step": 42695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.838974564267517e-05,
+      "loss": 2.9699,
+      "step": 42700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.838034323300126e-05,
+      "loss": 2.9191,
+      "step": 42705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8370940880661684e-05,
+      "loss": 2.9183,
+      "step": 42710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.836153858598925e-05,
+      "loss": 2.8539,
+      "step": 42715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.835213634931681e-05,
+      "loss": 2.8775,
+      "step": 42720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8342734170977165e-05,
+      "loss": 2.8791,
+      "step": 42725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.833333205130317e-05,
+      "loss": 2.918,
+      "step": 42730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.832392999062764e-05,
+      "loss": 3.0052,
+      "step": 42735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.831452798928339e-05,
+      "loss": 2.9502,
+      "step": 42740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.830512604760325e-05,
+      "loss": 2.8727,
+      "step": 42745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.829572416592006e-05,
+      "loss": 2.916,
+      "step": 42750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8286322344566605e-05,
+      "loss": 2.9364,
+      "step": 42755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8276920583875714e-05,
+      "loss": 2.8341,
+      "step": 42760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.826751888418022e-05,
+      "loss": 2.9292,
+      "step": 42765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.82581172458129e-05,
+      "loss": 2.9594,
+      "step": 42770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8248715669106594e-05,
+      "loss": 2.9356,
+      "step": 42775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8239314154394085e-05,
+      "loss": 2.9768,
+      "step": 42780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.822991270200818e-05,
+      "loss": 2.9665,
+      "step": 42785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.82205113122817e-05,
+      "loss": 2.9287,
+      "step": 42790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8211109985547426e-05,
+      "loss": 2.9005,
+      "step": 42795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8201708722138175e-05,
+      "loss": 2.9713,
+      "step": 42800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8192307522386715e-05,
+      "loss": 2.907,
+      "step": 42805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8182906386625857e-05,
+      "loss": 2.9091,
+      "step": 42810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.817350531518841e-05,
+      "loss": 2.9393,
+      "step": 42815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.816410430840713e-05,
+      "loss": 2.8847,
+      "step": 42820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.815470336661482e-05,
+      "loss": 2.961,
+      "step": 42825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.814530249014426e-05,
+      "loss": 2.8545,
+      "step": 42830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8135901679328234e-05,
+      "loss": 2.8913,
+      "step": 42835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8126500934499535e-05,
+      "loss": 2.9002,
+      "step": 42840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.81171002559909e-05,
+      "loss": 2.8944,
+      "step": 42845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.810769964413515e-05,
+      "loss": 2.8677,
+      "step": 42850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8098299099265035e-05,
+      "loss": 2.8724,
+      "step": 42855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.808889862171334e-05,
+      "loss": 2.8144,
+      "step": 42860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8079498211812813e-05,
+      "loss": 2.9015,
+      "step": 42865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.807009786989624e-05,
+      "loss": 2.8605,
+      "step": 42870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8060697596296366e-05,
+      "loss": 2.8306,
+      "step": 42875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.805129739134597e-05,
+      "loss": 2.933,
+      "step": 42880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.804189725537779e-05,
+      "loss": 3.0274,
+      "step": 42885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.803249718872459e-05,
+      "loss": 2.9481,
+      "step": 42890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8023097191719134e-05,
+      "loss": 2.976,
+      "step": 42895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.801369726469416e-05,
+      "loss": 2.9303,
+      "step": 42900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.800429740798243e-05,
+      "loss": 3.0019,
+      "step": 42905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.799489762191666e-05,
+      "loss": 2.7889,
+      "step": 42910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.798549790682962e-05,
+      "loss": 2.8226,
+      "step": 42915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.797609826305406e-05,
+      "loss": 2.825,
+      "step": 42920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.796669869092269e-05,
+      "loss": 2.9753,
+      "step": 42925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.795729919076827e-05,
+      "loss": 2.917,
+      "step": 42930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.794789976292352e-05,
+      "loss": 2.9584,
+      "step": 42935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.793850040772117e-05,
+      "loss": 2.871,
+      "step": 42940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.792910112549393e-05,
+      "loss": 2.9069,
+      "step": 42945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.791970191657458e-05,
+      "loss": 2.9889,
+      "step": 42950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7910302781295804e-05,
+      "loss": 2.8878,
+      "step": 42955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.790090371999032e-05,
+      "loss": 2.9317,
+      "step": 42960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.789150473299087e-05,
+      "loss": 2.8472,
+      "step": 42965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7882105820630145e-05,
+      "loss": 2.8341,
+      "step": 42970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.787270698324088e-05,
+      "loss": 2.9402,
+      "step": 42975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7863308221155764e-05,
+      "loss": 2.969,
+      "step": 42980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.78539095347075e-05,
+      "loss": 2.9746,
+      "step": 42985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.784451092422883e-05,
+      "loss": 2.9131,
+      "step": 42990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.783511239005241e-05,
+      "loss": 2.9419,
+      "step": 42995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.782571393251098e-05,
+      "loss": 2.8835,
+      "step": 43000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7816315551937205e-05,
+      "loss": 2.9506,
+      "step": 43005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.780691724866379e-05,
+      "loss": 2.9063,
+      "step": 43010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.779751902302342e-05,
+      "loss": 3.0282,
+      "step": 43015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.77881208753488e-05,
+      "loss": 2.8784,
+      "step": 43020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.777872280597261e-05,
+      "loss": 2.9606,
+      "step": 43025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.776932481522751e-05,
+      "loss": 2.8699,
+      "step": 43030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.775992690344622e-05,
+      "loss": 2.9613,
+      "step": 43035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.775052907096135e-05,
+      "loss": 2.9952,
+      "step": 43040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7741131318105664e-05,
+      "loss": 2.825,
+      "step": 43045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7731733645211775e-05,
+      "loss": 2.8093,
+      "step": 43050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7722336052612375e-05,
+      "loss": 2.881,
+      "step": 43055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.771293854064011e-05,
+      "loss": 2.8715,
+      "step": 43060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.770354110962765e-05,
+      "loss": 2.8958,
+      "step": 43065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.769414375990768e-05,
+      "loss": 2.8286,
+      "step": 43070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.768474649181282e-05,
+      "loss": 2.8603,
+      "step": 43075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.767534930567575e-05,
+      "loss": 2.8729,
+      "step": 43080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.76659522018291e-05,
+      "loss": 2.9196,
+      "step": 43085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7656555180605535e-05,
+      "loss": 2.9621,
+      "step": 43090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7647158242337704e-05,
+      "loss": 2.8346,
+      "step": 43095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.763776138735823e-05,
+      "loss": 2.9648,
+      "step": 43100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.762836461599978e-05,
+      "loss": 2.9319,
+      "step": 43105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.761896792859496e-05,
+      "loss": 2.9319,
+      "step": 43110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.760957132547642e-05,
+      "loss": 2.7752,
+      "step": 43115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.76001748069768e-05,
+      "loss": 2.8919,
+      "step": 43120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.759077837342871e-05,
+      "loss": 2.9066,
+      "step": 43125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7581382025164796e-05,
+      "loss": 2.9849,
+      "step": 43130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.757198576251765e-05,
+      "loss": 2.9518,
+      "step": 43135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.75625895858199e-05,
+      "loss": 2.8838,
+      "step": 43140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.755319349540418e-05,
+      "loss": 2.803,
+      "step": 43145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7543797491603105e-05,
+      "loss": 2.9143,
+      "step": 43150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.753440157474925e-05,
+      "loss": 2.9398,
+      "step": 43155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.752500574517527e-05,
+      "loss": 2.9295,
+      "step": 43160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7515610003213714e-05,
+      "loss": 2.9818,
+      "step": 43165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.750621434919723e-05,
+      "loss": 2.8677,
+      "step": 43170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.749681878345839e-05,
+      "loss": 2.8073,
+      "step": 43175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.748742330632979e-05,
+      "loss": 2.8979,
+      "step": 43180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.747802791814403e-05,
+      "loss": 2.9356,
+      "step": 43185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7468632619233684e-05,
+      "loss": 2.8315,
+      "step": 43190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.745923740993135e-05,
+      "loss": 2.9365,
+      "step": 43195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.74498422905696e-05,
+      "loss": 2.9608,
+      "step": 43200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.744044726148102e-05,
+      "loss": 3.0246,
+      "step": 43205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743105232299819e-05,
+      "loss": 2.8315,
+      "step": 43210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7421657475453664e-05,
+      "loss": 2.8921,
+      "step": 43215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7412262719180015e-05,
+      "loss": 2.975,
+      "step": 43220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.740286805450982e-05,
+      "loss": 2.9223,
+      "step": 43225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7393473481775636e-05,
+      "loss": 2.8817,
+      "step": 43230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.738407900131003e-05,
+      "loss": 2.9285,
+      "step": 43235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7374684613445515e-05,
+      "loss": 2.8751,
+      "step": 43240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.73652903185147e-05,
+      "loss": 2.9481,
+      "step": 43245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.735589611685012e-05,
+      "loss": 2.9344,
+      "step": 43250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.734650200878431e-05,
+      "loss": 2.8734,
+      "step": 43255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.73371079946498e-05,
+      "loss": 3.0235,
+      "step": 43260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.732771407477917e-05,
+      "loss": 2.9399,
+      "step": 43265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7318320249504903e-05,
+      "loss": 2.824,
+      "step": 43270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7308926519159566e-05,
+      "loss": 2.9082,
+      "step": 43275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7299532884075695e-05,
+      "loss": 2.9507,
+      "step": 43280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.729013934458578e-05,
+      "loss": 2.9087,
+      "step": 43285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7280745901022386e-05,
+      "loss": 2.8962,
+      "step": 43290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7271352553718e-05,
+      "loss": 2.9097,
+      "step": 43295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7261959303005136e-05,
+      "loss": 2.8723,
+      "step": 43300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7252566149216335e-05,
+      "loss": 2.9151,
+      "step": 43305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.724317309268408e-05,
+      "loss": 2.8644,
+      "step": 43310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.72337801337409e-05,
+      "loss": 2.9545,
+      "step": 43315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7224387272719265e-05,
+      "loss": 3.0377,
+      "step": 43320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.72149945099517e-05,
+      "loss": 2.9573,
+      "step": 43325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.720560184577069e-05,
+      "loss": 3.0573,
+      "step": 43330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.719620928050872e-05,
+      "loss": 2.917,
+      "step": 43335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.718681681449828e-05,
+      "loss": 2.8723,
+      "step": 43340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.717742444807187e-05,
+      "loss": 2.8547,
+      "step": 43345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.716803218156196e-05,
+      "loss": 2.947,
+      "step": 43350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.715864001530102e-05,
+      "loss": 3.0744,
+      "step": 43355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.714924794962155e-05,
+      "loss": 2.8656,
+      "step": 43360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7139855984855985e-05,
+      "loss": 2.9614,
+      "step": 43365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7130464121336806e-05,
+      "loss": 2.8231,
+      "step": 43370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7121072359396495e-05,
+      "loss": 2.8821,
+      "step": 43375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.711168069936748e-05,
+      "loss": 2.9829,
+      "step": 43380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.710228914158224e-05,
+      "loss": 2.9398,
+      "step": 43385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7092897686373216e-05,
+      "loss": 2.8951,
+      "step": 43390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.708350633407287e-05,
+      "loss": 2.9284,
+      "step": 43395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.707411508501362e-05,
+      "loss": 2.9694,
+      "step": 43400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.706472393952793e-05,
+      "loss": 2.9882,
+      "step": 43405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.705533289794824e-05,
+      "loss": 2.8918,
+      "step": 43410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7045941960606955e-05,
+      "loss": 2.8733,
+      "step": 43415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.703655112783654e-05,
+      "loss": 2.8998,
+      "step": 43420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7027160399969405e-05,
+      "loss": 2.988,
+      "step": 43425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.701776977733797e-05,
+      "loss": 2.8179,
+      "step": 43430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.700837926027467e-05,
+      "loss": 2.9333,
+      "step": 43435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.699898884911189e-05,
+      "loss": 2.8661,
+      "step": 43440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6989598544182066e-05,
+      "loss": 2.8859,
+      "step": 43445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.69802083458176e-05,
+      "loss": 2.9311,
+      "step": 43450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6970818254350926e-05,
+      "loss": 2.9304,
+      "step": 43455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6961428270114386e-05,
+      "loss": 3.012,
+      "step": 43460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6952038393440425e-05,
+      "loss": 2.9538,
+      "step": 43465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6942648624661406e-05,
+      "loss": 2.9218,
+      "step": 43470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6933258964109725e-05,
+      "loss": 2.9419,
+      "step": 43475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.692386941211779e-05,
+      "loss": 2.8696,
+      "step": 43480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6914479969017945e-05,
+      "loss": 2.8679,
+      "step": 43485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.69050906351426e-05,
+      "loss": 2.8585,
+      "step": 43490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.68957014108241e-05,
+      "loss": 2.98,
+      "step": 43495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.688631229639483e-05,
+      "loss": 2.9346,
+      "step": 43500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.687692329218717e-05,
+      "loss": 2.8714,
+      "step": 43505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6867534398533445e-05,
+      "loss": 2.8799,
+      "step": 43510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.685814561576606e-05,
+      "loss": 2.8271,
+      "step": 43515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.684875694421732e-05,
+      "loss": 2.9067,
+      "step": 43520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.68393683842196e-05,
+      "loss": 2.9303,
+      "step": 43525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6829979936105265e-05,
+      "loss": 2.921,
+      "step": 43530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6820591600206603e-05,
+      "loss": 2.8716,
+      "step": 43535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.681120337685601e-05,
+      "loss": 2.8622,
+      "step": 43540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.680181526638579e-05,
+      "loss": 2.9782,
+      "step": 43545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.679242726912828e-05,
+      "loss": 2.7777,
+      "step": 43550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.67830393854158e-05,
+      "loss": 2.9258,
+      "step": 43555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6773651615580695e-05,
+      "loss": 2.8793,
+      "step": 43560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6764263959955254e-05,
+      "loss": 2.8718,
+      "step": 43565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.675487641887181e-05,
+      "loss": 2.9616,
+      "step": 43570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.674548899266265e-05,
+      "loss": 2.8769,
+      "step": 43575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.67361016816601e-05,
+      "loss": 2.8026,
+      "step": 43580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.672671448619648e-05,
+      "loss": 2.9452,
+      "step": 43585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6717327406604046e-05,
+      "loss": 2.9575,
+      "step": 43590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.670794044321512e-05,
+      "loss": 2.7607,
+      "step": 43595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6698553596361975e-05,
+      "loss": 2.8861,
+      "step": 43600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6689166866376903e-05,
+      "loss": 2.9904,
+      "step": 43605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.66797802535922e-05,
+      "loss": 2.7183,
+      "step": 43610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.667039375834012e-05,
+      "loss": 2.9309,
+      "step": 43615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.666100738095296e-05,
+      "loss": 2.8882,
+      "step": 43620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6651621121762954e-05,
+      "loss": 2.9451,
+      "step": 43625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.66422349811024e-05,
+      "loss": 2.8785,
+      "step": 43630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.663284895930353e-05,
+      "loss": 2.877,
+      "step": 43635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6623463056698634e-05,
+      "loss": 2.9194,
+      "step": 43640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.661407727361994e-05,
+      "loss": 2.8546,
+      "step": 43645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.660469161039971e-05,
+      "loss": 2.9867,
+      "step": 43650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.659530606737017e-05,
+      "loss": 2.9301,
+      "step": 43655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6585920644863576e-05,
+      "loss": 2.9404,
+      "step": 43660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.657653534321216e-05,
+      "loss": 2.9351,
+      "step": 43665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.656715016274814e-05,
+      "loss": 2.8926,
+      "step": 43670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.655776510380376e-05,
+      "loss": 2.862,
+      "step": 43675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.654838016671123e-05,
+      "loss": 2.9149,
+      "step": 43680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6538995351802764e-05,
+      "loss": 2.8589,
+      "step": 43685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.65296106594106e-05,
+      "loss": 2.9521,
+      "step": 43690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.652022608986692e-05,
+      "loss": 2.9439,
+      "step": 43695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.651084164350394e-05,
+      "loss": 2.8656,
+      "step": 43700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.650145732065386e-05,
+      "loss": 2.8769,
+      "step": 43705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6492073121648874e-05,
+      "loss": 2.8504,
+      "step": 43710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.648268904682119e-05,
+      "loss": 2.8855,
+      "step": 43715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.647330509650296e-05,
+      "loss": 2.9433,
+      "step": 43720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.646392127102641e-05,
+      "loss": 2.8658,
+      "step": 43725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.645453757072368e-05,
+      "loss": 2.8996,
+      "step": 43730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.644515399592695e-05,
+      "loss": 2.8999,
+      "step": 43735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6435770546968413e-05,
+      "loss": 2.8182,
+      "step": 43740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.642638722418023e-05,
+      "loss": 2.8313,
+      "step": 43745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.641700402789455e-05,
+      "loss": 2.9728,
+      "step": 43750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6407620958443524e-05,
+      "loss": 2.9195,
+      "step": 43755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6398238016159325e-05,
+      "loss": 2.8449,
+      "step": 43760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.638885520137407e-05,
+      "loss": 2.9844,
+      "step": 43765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.637947251441994e-05,
+      "loss": 2.939,
+      "step": 43770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.637008995562904e-05,
+      "loss": 2.8532,
+      "step": 43775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6360707525333524e-05,
+      "loss": 2.9275,
+      "step": 43780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.63513252238655e-05,
+      "loss": 2.9599,
+      "step": 43785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6341943051557115e-05,
+      "loss": 2.9233,
+      "step": 43790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.633256100874049e-05,
+      "loss": 2.9712,
+      "step": 43795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6323179095747716e-05,
+      "loss": 2.9826,
+      "step": 43800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.631379731291093e-05,
+      "loss": 2.9321,
+      "step": 43805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.630441566056221e-05,
+      "loss": 2.983,
+      "step": 43810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6295034139033685e-05,
+      "loss": 2.9652,
+      "step": 43815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.628565274865744e-05,
+      "loss": 2.9352,
+      "step": 43820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.627627148976557e-05,
+      "loss": 2.9232,
+      "step": 43825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6266890362690165e-05,
+      "loss": 2.9214,
+      "step": 43830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.625750936776327e-05,
+      "loss": 2.9423,
+      "step": 43835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.624812850531703e-05,
+      "loss": 2.959,
+      "step": 43840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6238747775683476e-05,
+      "loss": 2.8807,
+      "step": 43845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6229367179194704e-05,
+      "loss": 2.9707,
+      "step": 43850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6219986716182745e-05,
+      "loss": 2.9425,
+      "step": 43855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.621060638697967e-05,
+      "loss": 2.9016,
+      "step": 43860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.620122619191756e-05,
+      "loss": 2.9023,
+      "step": 43865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6191846131328426e-05,
+      "loss": 2.8333,
+      "step": 43870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.618246620554434e-05,
+      "loss": 2.9553,
+      "step": 43875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6173086414897326e-05,
+      "loss": 2.949,
+      "step": 43880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6163706759719425e-05,
+      "loss": 2.9193,
+      "step": 43885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6154327240342685e-05,
+      "loss": 2.9385,
+      "step": 43890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.614494785709911e-05,
+      "loss": 2.9026,
+      "step": 43895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.613556861032073e-05,
+      "loss": 2.9277,
+      "step": 43900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.612618950033955e-05,
+      "loss": 2.8356,
+      "step": 43905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.611681052748759e-05,
+      "loss": 2.8807,
+      "step": 43910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.610743169209687e-05,
+      "loss": 2.9078,
+      "step": 43915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.609805299449937e-05,
+      "loss": 2.9875,
+      "step": 43920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.608867443502711e-05,
+      "loss": 3.0207,
+      "step": 43925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6079296014012046e-05,
+      "loss": 2.9336,
+      "step": 43930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.60699177317862e-05,
+      "loss": 2.9394,
+      "step": 43935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.606053958868154e-05,
+      "loss": 2.9272,
+      "step": 43940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.605116158503006e-05,
+      "loss": 2.8324,
+      "step": 43945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6041783721163706e-05,
+      "loss": 2.9437,
+      "step": 43950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.603240599741447e-05,
+      "loss": 2.9484,
+      "step": 43955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6023028414114286e-05,
+      "loss": 2.8851,
+      "step": 43960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.601365097159513e-05,
+      "loss": 2.9293,
+      "step": 43965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.600427367018897e-05,
+      "loss": 2.829,
+      "step": 43970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.599489651022772e-05,
+      "loss": 2.9342,
+      "step": 43975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.598551949204336e-05,
+      "loss": 2.8232,
+      "step": 43980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.597614261596778e-05,
+      "loss": 2.9774,
+      "step": 43985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.596676588233294e-05,
+      "loss": 3.057,
+      "step": 43990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5957389291470784e-05,
+      "loss": 2.9041,
+      "step": 43995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.594801284371321e-05,
+      "loss": 3.0497,
+      "step": 44000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5938636539392146e-05,
+      "loss": 2.9085,
+      "step": 44005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.592926037883949e-05,
+      "loss": 2.9056,
+      "step": 44010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.591988436238716e-05,
+      "loss": 2.8201,
+      "step": 44015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.591050849036707e-05,
+      "loss": 2.8265,
+      "step": 44020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.590113276311109e-05,
+      "loss": 2.8956,
+      "step": 44025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5891757180951116e-05,
+      "loss": 2.9861,
+      "step": 44030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.588238174421907e-05,
+      "loss": 2.966,
+      "step": 44035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5873006453246804e-05,
+      "loss": 2.8486,
+      "step": 44040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.586363130836619e-05,
+      "loss": 2.9629,
+      "step": 44045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.585425630990912e-05,
+      "loss": 2.8568,
+      "step": 44050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.584488145820744e-05,
+      "loss": 2.9276,
+      "step": 44055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.583550675359303e-05,
+      "loss": 2.9181,
+      "step": 44060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.582613219639773e-05,
+      "loss": 2.8823,
+      "step": 44065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.581675778695338e-05,
+      "loss": 2.8254,
+      "step": 44070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.580738352559187e-05,
+      "loss": 3.0095,
+      "step": 44075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.579800941264499e-05,
+      "loss": 2.988,
+      "step": 44080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5788635448444606e-05,
+      "loss": 2.8961,
+      "step": 44085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.577926163332253e-05,
+      "loss": 2.8591,
+      "step": 44090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.576988796761059e-05,
+      "loss": 2.9149,
+      "step": 44095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5760514451640625e-05,
+      "loss": 2.8419,
+      "step": 44100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5751141085744415e-05,
+      "loss": 2.8998,
+      "step": 44105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.574176787025379e-05,
+      "loss": 2.9229,
+      "step": 44110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.573239480550054e-05,
+      "loss": 2.9342,
+      "step": 44115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.572302189181648e-05,
+      "loss": 3.0193,
+      "step": 44120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.57136491295334e-05,
+      "loss": 2.9153,
+      "step": 44125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.570427651898305e-05,
+      "loss": 2.962,
+      "step": 44130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.569490406049726e-05,
+      "loss": 2.9374,
+      "step": 44135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.568553175440779e-05,
+      "loss": 2.8296,
+      "step": 44140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.567615960104641e-05,
+      "loss": 3.0175,
+      "step": 44145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5666787600744876e-05,
+      "loss": 3.0427,
+      "step": 44150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5657415753834964e-05,
+      "loss": 2.906,
+      "step": 44155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.564804406064841e-05,
+      "loss": 3.0017,
+      "step": 44160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.563867252151697e-05,
+      "loss": 2.8413,
+      "step": 44165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5629301136772405e-05,
+      "loss": 2.892,
+      "step": 44170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.561992990674643e-05,
+      "loss": 2.8715,
+      "step": 44175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5610558831770794e-05,
+      "loss": 2.9278,
+      "step": 44180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5601187912177205e-05,
+      "loss": 2.8789,
+      "step": 44185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.559181714829741e-05,
+      "loss": 2.8887,
+      "step": 44190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5582446540463106e-05,
+      "loss": 2.9944,
+      "step": 44195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.557307608900599e-05,
+      "loss": 2.9504,
+      "step": 44200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.556370579425781e-05,
+      "loss": 2.8941,
+      "step": 44205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5554335656550234e-05,
+      "loss": 2.9339,
+      "step": 44210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5544965676214967e-05,
+      "loss": 2.7734,
+      "step": 44215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.553559585358369e-05,
+      "loss": 2.855,
+      "step": 44220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.552622618898808e-05,
+      "loss": 2.9156,
+      "step": 44225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.551685668275982e-05,
+      "loss": 2.9799,
+      "step": 44230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.550748733523061e-05,
+      "loss": 2.9628,
+      "step": 44235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.549811814673208e-05,
+      "loss": 2.9297,
+      "step": 44240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.54887491175959e-05,
+      "loss": 2.9052,
+      "step": 44245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.547938024815373e-05,
+      "loss": 2.9849,
+      "step": 44250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5470011538737214e-05,
+      "loss": 2.9244,
+      "step": 44255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5460642989678e-05,
+      "loss": 2.8828,
+      "step": 44260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5451274601307726e-05,
+      "loss": 2.8709,
+      "step": 44265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.544190637395802e-05,
+      "loss": 2.9226,
+      "step": 44270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.543253830796051e-05,
+      "loss": 2.9081,
+      "step": 44275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.542317040364681e-05,
+      "loss": 2.8808,
+      "step": 44280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.541380266134855e-05,
+      "loss": 2.8187,
+      "step": 44285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.540443508139732e-05,
+      "loss": 2.9235,
+      "step": 44290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.539506766412473e-05,
+      "loss": 2.9426,
+      "step": 44295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5385700409862406e-05,
+      "loss": 2.9206,
+      "step": 44300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.537633331894189e-05,
+      "loss": 2.9578,
+      "step": 44305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5366966391694816e-05,
+      "loss": 3.0116,
+      "step": 44310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.535759962845273e-05,
+      "loss": 2.8446,
+      "step": 44315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5348233029547224e-05,
+      "loss": 2.9177,
+      "step": 44320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5338866595309833e-05,
+      "loss": 2.91,
+      "step": 44325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.532950032607218e-05,
+      "loss": 2.9717,
+      "step": 44330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.532013422216577e-05,
+      "loss": 2.8948,
+      "step": 44335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.53107682839222e-05,
+      "loss": 2.9116,
+      "step": 44340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.530140251167297e-05,
+      "loss": 2.863,
+      "step": 44345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5292036905749635e-05,
+      "loss": 2.8912,
+      "step": 44350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.528267146648375e-05,
+      "loss": 2.9406,
+      "step": 44355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5273306194206815e-05,
+      "loss": 2.9003,
+      "step": 44360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5263941089250375e-05,
+      "loss": 2.8943,
+      "step": 44365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.525457615194592e-05,
+      "loss": 2.9677,
+      "step": 44370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.524521138262498e-05,
+      "loss": 2.968,
+      "step": 44375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.523584678161904e-05,
+      "loss": 2.9805,
+      "step": 44380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5226482349259625e-05,
+      "loss": 2.8575,
+      "step": 44385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5217118085878204e-05,
+      "loss": 2.8777,
+      "step": 44390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5207753991806276e-05,
+      "loss": 2.9055,
+      "step": 44395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.519839006737531e-05,
+      "loss": 2.917,
+      "step": 44400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5189026312916795e-05,
+      "loss": 3.0174,
+      "step": 44405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.517966272876218e-05,
+      "loss": 2.9065,
+      "step": 44410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.517029931524295e-05,
+      "loss": 2.8974,
+      "step": 44415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.516093607269053e-05,
+      "loss": 2.9874,
+      "step": 44420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.515157300143639e-05,
+      "loss": 2.8611,
+      "step": 44425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5142210101811975e-05,
+      "loss": 2.9529,
+      "step": 44430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.513284737414872e-05,
+      "loss": 2.8991,
+      "step": 44435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.512348481877806e-05,
+      "loss": 2.9586,
+      "step": 44440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.511412243603143e-05,
+      "loss": 2.8818,
+      "step": 44445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5104760226240216e-05,
+      "loss": 2.9275,
+      "step": 44450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.509539818973586e-05,
+      "loss": 2.8818,
+      "step": 44455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.508603632684977e-05,
+      "loss": 2.9732,
+      "step": 44460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.507667463791332e-05,
+      "loss": 2.9147,
+      "step": 44465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.506731312325794e-05,
+      "loss": 2.9936,
+      "step": 44470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.505795178321499e-05,
+      "loss": 2.9108,
+      "step": 44475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.504859061811586e-05,
+      "loss": 3.0018,
+      "step": 44480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5039229628291954e-05,
+      "loss": 2.8991,
+      "step": 44485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.50298688140746e-05,
+      "loss": 2.8386,
+      "step": 44490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.502050817579519e-05,
+      "loss": 2.9816,
+      "step": 44495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5011147713785065e-05,
+      "loss": 2.8464,
+      "step": 44500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.500178742837558e-05,
+      "loss": 2.8541,
+      "step": 44505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.499242731989809e-05,
+      "loss": 2.8502,
+      "step": 44510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4983067388683934e-05,
+      "loss": 2.8511,
+      "step": 44515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.497370763506444e-05,
+      "loss": 2.7998,
+      "step": 44520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4964348059370894e-05,
+      "loss": 2.7973,
+      "step": 44525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4954988661934706e-05,
+      "loss": 2.8697,
+      "step": 44530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.494562944308712e-05,
+      "loss": 2.9155,
+      "step": 44535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.493627040315946e-05,
+      "loss": 3.003,
+      "step": 44540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.492691154248303e-05,
+      "loss": 2.9124,
+      "step": 44545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.491755286138913e-05,
+      "loss": 2.9246,
+      "step": 44550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.490819436020906e-05,
+      "loss": 2.8603,
+      "step": 44555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4898836039274054e-05,
+      "loss": 2.8843,
+      "step": 44560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4889477898915445e-05,
+      "loss": 2.9578,
+      "step": 44565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.488011993946446e-05,
+      "loss": 2.8702,
+      "step": 44570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4870762161252386e-05,
+      "loss": 2.9557,
+      "step": 44575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4861404564610466e-05,
+      "loss": 2.9838,
+      "step": 44580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.485204714986995e-05,
+      "loss": 2.8647,
+      "step": 44585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4842689917362106e-05,
+      "loss": 2.8439,
+      "step": 44590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4833332867418134e-05,
+      "loss": 2.9008,
+      "step": 44595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.48239760003693e-05,
+      "loss": 2.9269,
+      "step": 44600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4814619316546805e-05,
+      "loss": 2.7667,
+      "step": 44605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.480526281628186e-05,
+      "loss": 2.8787,
+      "step": 44610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4795906499905705e-05,
+      "loss": 2.9326,
+      "step": 44615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.478655036774952e-05,
+      "loss": 3.0095,
+      "step": 44620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.47771944201445e-05,
+      "loss": 2.8434,
+      "step": 44625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.476783865742186e-05,
+      "loss": 3.0054,
+      "step": 44630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.475848307991279e-05,
+      "loss": 2.9277,
+      "step": 44635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.474912768794844e-05,
+      "loss": 2.8668,
+      "step": 44640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.473977248186e-05,
+      "loss": 2.9239,
+      "step": 44645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.473041746197862e-05,
+      "loss": 2.8937,
+      "step": 44650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4721062628635474e-05,
+      "loss": 3.0579,
+      "step": 44655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.471170798216172e-05,
+      "loss": 2.9162,
+      "step": 44660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.470235352288848e-05,
+      "loss": 2.7974,
+      "step": 44665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.469299925114691e-05,
+      "loss": 2.8948,
+      "step": 44670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.468364516726814e-05,
+      "loss": 2.7967,
+      "step": 44675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.467429127158329e-05,
+      "loss": 2.9677,
+      "step": 44680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.466493756442349e-05,
+      "loss": 2.8908,
+      "step": 44685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4655584046119835e-05,
+      "loss": 2.8901,
+      "step": 44690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.464623071700345e-05,
+      "loss": 2.8011,
+      "step": 44695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4636877577405416e-05,
+      "loss": 2.8775,
+      "step": 44700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.462752462765685e-05,
+      "loss": 2.9391,
+      "step": 44705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.461817186808881e-05,
+      "loss": 2.8472,
+      "step": 44710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4608819299032376e-05,
+      "loss": 2.9716,
+      "step": 44715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.459946692081863e-05,
+      "loss": 2.8552,
+      "step": 44720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.459011473377866e-05,
+      "loss": 2.788,
+      "step": 44725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4580762738243486e-05,
+      "loss": 2.7556,
+      "step": 44730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.457141093454418e-05,
+      "loss": 2.9296,
+      "step": 44735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.45620593230118e-05,
+      "loss": 2.8234,
+      "step": 44740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.455270790397735e-05,
+      "loss": 2.825,
+      "step": 44745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.454335667777189e-05,
+      "loss": 2.9706,
+      "step": 44750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4534005644726424e-05,
+      "loss": 2.9238,
+      "step": 44755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.452465480517198e-05,
+      "loss": 2.8876,
+      "step": 44760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4515304159439575e-05,
+      "loss": 2.8997,
+      "step": 44765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.45059537078602e-05,
+      "loss": 2.9686,
+      "step": 44770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4496603450764865e-05,
+      "loss": 2.9031,
+      "step": 44775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.448725338848454e-05,
+      "loss": 2.896,
+      "step": 44780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.447790352135023e-05,
+      "loss": 2.9251,
+      "step": 44785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.446855384969291e-05,
+      "loss": 2.9041,
+      "step": 44790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4459204373843535e-05,
+      "loss": 2.7984,
+      "step": 44795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.444985509413308e-05,
+      "loss": 2.8837,
+      "step": 44800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.444050601089249e-05,
+      "loss": 2.8727,
+      "step": 44805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.443115712445271e-05,
+      "loss": 2.9026,
+      "step": 44810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4421808435144715e-05,
+      "loss": 2.9467,
+      "step": 44815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.441245994329939e-05,
+      "loss": 2.945,
+      "step": 44820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.44031116492477e-05,
+      "loss": 2.8869,
+      "step": 44825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.439376355332057e-05,
+      "loss": 2.8369,
+      "step": 44830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.438441565584888e-05,
+      "loss": 2.9719,
+      "step": 44835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.437506795716356e-05,
+      "loss": 3.0647,
+      "step": 44840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.436572045759551e-05,
+      "loss": 2.85,
+      "step": 44845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.435637315747561e-05,
+      "loss": 2.8595,
+      "step": 44850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.434702605713476e-05,
+      "loss": 3.0033,
+      "step": 44855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.433767915690383e-05,
+      "loss": 2.9919,
+      "step": 44860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4328332457113684e-05,
+      "loss": 2.9286,
+      "step": 44865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.431898595809521e-05,
+      "loss": 2.9363,
+      "step": 44870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.430963966017924e-05,
+      "loss": 2.9833,
+      "step": 44875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4300293563696646e-05,
+      "loss": 2.845,
+      "step": 44880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.429094766897824e-05,
+      "loss": 2.8593,
+      "step": 44885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.428160197635489e-05,
+      "loss": 2.965,
+      "step": 44890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.427225648615742e-05,
+      "loss": 2.8111,
+      "step": 44895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.426291119871663e-05,
+      "loss": 2.8135,
+      "step": 44900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.425356611436335e-05,
+      "loss": 2.9453,
+      "step": 44905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.424422123342839e-05,
+      "loss": 2.9655,
+      "step": 44910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.423487655624253e-05,
+      "loss": 2.7486,
+      "step": 44915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.422553208313658e-05,
+      "loss": 2.9257,
+      "step": 44920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.421618781444134e-05,
+      "loss": 2.9577,
+      "step": 44925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.420684375048756e-05,
+      "loss": 2.9647,
+      "step": 44930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.419749989160602e-05,
+      "loss": 2.8881,
+      "step": 44935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4188156238127495e-05,
+      "loss": 2.8861,
+      "step": 44940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.417881279038273e-05,
+      "loss": 2.889,
+      "step": 44945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.416946954870248e-05,
+      "loss": 2.8807,
+      "step": 44950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4160126513417476e-05,
+      "loss": 2.975,
+      "step": 44955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.415078368485847e-05,
+      "loss": 2.8477,
+      "step": 44960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.414144106335617e-05,
+      "loss": 2.9564,
+      "step": 44965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4132098649241306e-05,
+      "loss": 2.9149,
+      "step": 44970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.412275644284459e-05,
+      "loss": 2.9378,
+      "step": 44975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.411341444449673e-05,
+      "loss": 2.9147,
+      "step": 44980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.410407265452843e-05,
+      "loss": 2.8866,
+      "step": 44985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.409473107327036e-05,
+      "loss": 2.9111,
+      "step": 44990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4085389701053215e-05,
+      "loss": 2.9212,
+      "step": 44995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4076048538207676e-05,
+      "loss": 2.9136,
+      "step": 45000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.40667075850644e-05,
+      "loss": 2.9007,
+      "step": 45005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.405736684195406e-05,
+      "loss": 2.9616,
+      "step": 45010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4048026309207304e-05,
+      "loss": 2.9015,
+      "step": 45015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4038685987154756e-05,
+      "loss": 2.9921,
+      "step": 45020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4029345876127084e-05,
+      "loss": 3.0074,
+      "step": 45025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.402000597645493e-05,
+      "loss": 2.8787,
+      "step": 45030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.401066628846889e-05,
+      "loss": 2.9008,
+      "step": 45035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4001326812499576e-05,
+      "loss": 2.9426,
+      "step": 45040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3991987548877634e-05,
+      "loss": 2.9476,
+      "step": 45045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.398264849793362e-05,
+      "loss": 2.877,
+      "step": 45050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3973309659998165e-05,
+      "loss": 2.9289,
+      "step": 45055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.396397103540182e-05,
+      "loss": 2.9779,
+      "step": 45060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3954632624475184e-05,
+      "loss": 2.97,
+      "step": 45065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3945294427548836e-05,
+      "loss": 2.7401,
+      "step": 45070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.393595644495332e-05,
+      "loss": 2.804,
+      "step": 45075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.392661867701921e-05,
+      "loss": 2.8257,
+      "step": 45080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3917281124077034e-05,
+      "loss": 2.9057,
+      "step": 45085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.390794378645735e-05,
+      "loss": 2.8824,
+      "step": 45090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3898606664490676e-05,
+      "loss": 2.8999,
+      "step": 45095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3889269758507544e-05,
+      "loss": 2.963,
+      "step": 45100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3879933068838486e-05,
+      "loss": 2.9652,
+      "step": 45105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.387059659581398e-05,
+      "loss": 2.942,
+      "step": 45110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.386126033976457e-05,
+      "loss": 3.0323,
+      "step": 45115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.385192430102069e-05,
+      "loss": 2.8948,
+      "step": 45120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3842588479912896e-05,
+      "loss": 2.9057,
+      "step": 45125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3833252876771625e-05,
+      "loss": 2.921,
+      "step": 45130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3823917491927366e-05,
+      "loss": 2.9896,
+      "step": 45135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.381458232571057e-05,
+      "loss": 3.0639,
+      "step": 45140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3805247378451706e-05,
+      "loss": 2.9554,
+      "step": 45145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.379591265048122e-05,
+      "loss": 3.0371,
+      "step": 45150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.378657814212953e-05,
+      "loss": 2.952,
+      "step": 45155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3777243853727105e-05,
+      "loss": 2.9151,
+      "step": 45160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.376790978560434e-05,
+      "loss": 2.8824,
+      "step": 45165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.375857593809166e-05,
+      "loss": 2.7837,
+      "step": 45170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.374924231151949e-05,
+      "loss": 2.8867,
+      "step": 45175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3739908906218215e-05,
+      "loss": 2.9117,
+      "step": 45180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.373057572251824e-05,
+      "loss": 2.7457,
+      "step": 45185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3721242760749935e-05,
+      "loss": 2.9553,
+      "step": 45190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3711910021243686e-05,
+      "loss": 2.8098,
+      "step": 45195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3702577504329876e-05,
+      "loss": 2.8929,
+      "step": 45200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.369324521033884e-05,
+      "loss": 3.0292,
+      "step": 45205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.368391313960096e-05,
+      "loss": 2.9808,
+      "step": 45210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.367458129244655e-05,
+      "loss": 2.8936,
+      "step": 45215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.366524966920599e-05,
+      "loss": 2.9246,
+      "step": 45220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.365591827020958e-05,
+      "loss": 2.9093,
+      "step": 45225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3646587095787664e-05,
+      "loss": 2.8866,
+      "step": 45230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.363725614627053e-05,
+      "loss": 3.0084,
+      "step": 45235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.362792542198851e-05,
+      "loss": 2.9259,
+      "step": 45240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3618594923271896e-05,
+      "loss": 2.8957,
+      "step": 45245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3609264650450963e-05,
+      "loss": 2.9457,
+      "step": 45250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.359993460385602e-05,
+      "loss": 2.8333,
+      "step": 45255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.359060478381732e-05,
+      "loss": 2.8963,
+      "step": 45260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3581275190665145e-05,
+      "loss": 2.913,
+      "step": 45265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3571945824729735e-05,
+      "loss": 2.8571,
+      "step": 45270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.356261668634135e-05,
+      "loss": 2.9183,
+      "step": 45275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.355328777583025e-05,
+      "loss": 2.832,
+      "step": 45280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.354395909352665e-05,
+      "loss": 2.7973,
+      "step": 45285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3534630639760784e-05,
+      "loss": 2.8658,
+      "step": 45290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.352530241486285e-05,
+      "loss": 2.8837,
+      "step": 45295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3515974419163094e-05,
+      "loss": 2.8801,
+      "step": 45300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.35066466529917e-05,
+      "loss": 3.0072,
+      "step": 45305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3497319116678856e-05,
+      "loss": 2.9726,
+      "step": 45310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.348799181055474e-05,
+      "loss": 2.9227,
+      "step": 45315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3478664734949545e-05,
+      "loss": 2.7852,
+      "step": 45320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3469337890193466e-05,
+      "loss": 2.9259,
+      "step": 45325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.346001127661662e-05,
+      "loss": 2.9885,
+      "step": 45330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.345068489454919e-05,
+      "loss": 2.9137,
+      "step": 45335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.34413587443213e-05,
+      "loss": 2.9102,
+      "step": 45340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3432032826263104e-05,
+      "loss": 3.0478,
+      "step": 45345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.342270714070471e-05,
+      "loss": 2.8214,
+      "step": 45350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3413381687976255e-05,
+      "loss": 2.885,
+      "step": 45355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3404056468407864e-05,
+      "loss": 2.8876,
+      "step": 45360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.33947314823296e-05,
+      "loss": 2.9079,
+      "step": 45365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.338540673007161e-05,
+      "loss": 2.9814,
+      "step": 45370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3376082211963934e-05,
+      "loss": 2.9264,
+      "step": 45375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3366757928336674e-05,
+      "loss": 2.8783,
+      "step": 45380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.33574338795199e-05,
+      "loss": 2.7975,
+      "step": 45385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.334811006584368e-05,
+      "loss": 2.9522,
+      "step": 45390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.333878648763806e-05,
+      "loss": 2.8718,
+      "step": 45395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.332946314523308e-05,
+      "loss": 2.8647,
+      "step": 45400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.332014003895878e-05,
+      "loss": 2.9111,
+      "step": 45405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.331081716914521e-05,
+      "loss": 2.7772,
+      "step": 45410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.330149453612235e-05,
+      "loss": 2.9712,
+      "step": 45415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.329217214022024e-05,
+      "loss": 3.0201,
+      "step": 45420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.328284998176889e-05,
+      "loss": 2.9051,
+      "step": 45425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.32735280610983e-05,
+      "loss": 2.8638,
+      "step": 45430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.326420637853843e-05,
+      "loss": 2.8894,
+      "step": 45435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3254884934419284e-05,
+      "loss": 2.9156,
+      "step": 45440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3245563729070806e-05,
+      "loss": 2.9476,
+      "step": 45445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.323624276282297e-05,
+      "loss": 2.9228,
+      "step": 45450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.322692203600575e-05,
+      "loss": 2.9257,
+      "step": 45455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.321760154894905e-05,
+      "loss": 2.8964,
+      "step": 45460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3208281301982844e-05,
+      "loss": 2.8253,
+      "step": 45465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.319896129543704e-05,
+      "loss": 2.8739,
+      "step": 45470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.318964152964157e-05,
+      "loss": 2.9474,
+      "step": 45475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.318032200492633e-05,
+      "loss": 2.9348,
+      "step": 45480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3171002721621225e-05,
+      "loss": 2.8569,
+      "step": 45485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.316168368005617e-05,
+      "loss": 2.8269,
+      "step": 45490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.315236488056101e-05,
+      "loss": 2.8937,
+      "step": 45495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.314304632346566e-05,
+      "loss": 2.9039,
+      "step": 45500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.313372800909997e-05,
+      "loss": 2.9148,
+      "step": 45505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3124409937793796e-05,
+      "loss": 2.9068,
+      "step": 45510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.311509210987699e-05,
+      "loss": 2.9914,
+      "step": 45515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.310577452567942e-05,
+      "loss": 2.8051,
+      "step": 45520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3096457185530884e-05,
+      "loss": 3.0092,
+      "step": 45525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.308714008976123e-05,
+      "loss": 2.9111,
+      "step": 45530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3077823238700274e-05,
+      "loss": 2.7696,
+      "step": 45535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.306850663267781e-05,
+      "loss": 2.8873,
+      "step": 45540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.305919027202365e-05,
+      "loss": 3.0249,
+      "step": 45545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.304987415706757e-05,
+      "loss": 2.8828,
+      "step": 45550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.304055828813935e-05,
+      "loss": 2.9336,
+      "step": 45555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3031242665568793e-05,
+      "loss": 2.8643,
+      "step": 45560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.302192728968564e-05,
+      "loss": 2.8128,
+      "step": 45565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.301261216081964e-05,
+      "loss": 2.998,
+      "step": 45570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.300329727930055e-05,
+      "loss": 2.9039,
+      "step": 45575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2993982645458106e-05,
+      "loss": 2.98,
+      "step": 45580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.298466825962205e-05,
+      "loss": 2.8775,
+      "step": 45585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.297535412212208e-05,
+      "loss": 2.8555,
+      "step": 45590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.296604023328793e-05,
+      "loss": 3.0177,
+      "step": 45595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.295672659344928e-05,
+      "loss": 2.8098,
+      "step": 45600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2947413202935844e-05,
+      "loss": 3.0137,
+      "step": 45605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2938100062077264e-05,
+      "loss": 2.9257,
+      "step": 45610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.292878717120329e-05,
+      "loss": 2.9056,
+      "step": 45615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.291947453064353e-05,
+      "loss": 2.9611,
+      "step": 45620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2910162140727684e-05,
+      "loss": 2.9225,
+      "step": 45625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.290085000178536e-05,
+      "loss": 2.9549,
+      "step": 45630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2891538114146226e-05,
+      "loss": 2.8355,
+      "step": 45635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.288222647813991e-05,
+      "loss": 2.9304,
+      "step": 45640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.287291509409603e-05,
+      "loss": 2.9253,
+      "step": 45645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.286360396234422e-05,
+      "loss": 2.8714,
+      "step": 45650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.285429308321405e-05,
+      "loss": 2.8374,
+      "step": 45655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.284498245703513e-05,
+      "loss": 2.9479,
+      "step": 45660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.283567208413707e-05,
+      "loss": 2.9023,
+      "step": 45665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.282636196484942e-05,
+      "loss": 2.8369,
+      "step": 45670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2817052099501756e-05,
+      "loss": 2.8007,
+      "step": 45675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.280774248842364e-05,
+      "loss": 2.91,
+      "step": 45680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.279843313194463e-05,
+      "loss": 2.8815,
+      "step": 45685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.278912403039427e-05,
+      "loss": 2.9225,
+      "step": 45690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.277981518410208e-05,
+      "loss": 2.9783,
+      "step": 45695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2770506593397594e-05,
+      "loss": 2.9191,
+      "step": 45700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.276119825861033e-05,
+      "loss": 2.8715,
+      "step": 45705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2751890180069764e-05,
+      "loss": 2.9565,
+      "step": 45710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.274258235810543e-05,
+      "loss": 2.9665,
+      "step": 45715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2733274793046816e-05,
+      "loss": 2.7919,
+      "step": 45720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.272396748522338e-05,
+      "loss": 2.8476,
+      "step": 45725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2714660434964615e-05,
+      "loss": 2.9584,
+      "step": 45730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.270535364259996e-05,
+      "loss": 2.8906,
+      "step": 45735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.269604710845886e-05,
+      "loss": 2.9343,
+      "step": 45740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2686740832870795e-05,
+      "loss": 2.9993,
+      "step": 45745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.267743481616516e-05,
+      "loss": 2.8052,
+      "step": 45750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.266812905867141e-05,
+      "loss": 2.9281,
+      "step": 45755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.265882356071893e-05,
+      "loss": 2.8661,
+      "step": 45760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.264951832263714e-05,
+      "loss": 2.9273,
+      "step": 45765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.264021334475545e-05,
+      "loss": 2.8677,
+      "step": 45770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.263090862740322e-05,
+      "loss": 2.8087,
+      "step": 45775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.262160417090986e-05,
+      "loss": 2.9239,
+      "step": 45780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.26122999756047e-05,
+      "loss": 2.9746,
+      "step": 45785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.260299604181713e-05,
+      "loss": 2.9493,
+      "step": 45790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2593692369876496e-05,
+      "loss": 2.8968,
+      "step": 45795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.258438896011213e-05,
+      "loss": 2.9128,
+      "step": 45800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2575085812853376e-05,
+      "loss": 3.044,
+      "step": 45805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2565782928429516e-05,
+      "loss": 2.8204,
+      "step": 45810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.255648030716993e-05,
+      "loss": 2.9147,
+      "step": 45815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2547177949403874e-05,
+      "loss": 2.8672,
+      "step": 45820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.253787585546067e-05,
+      "loss": 2.9211,
+      "step": 45825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2528574025669584e-05,
+      "loss": 2.9177,
+      "step": 45830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.25192724603599e-05,
+      "loss": 2.7487,
+      "step": 45835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.250997115986089e-05,
+      "loss": 2.8132,
+      "step": 45840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.250067012450179e-05,
+      "loss": 2.8542,
+      "step": 45845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2491369354611874e-05,
+      "loss": 2.9606,
+      "step": 45850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.248206885052036e-05,
+      "loss": 2.8317,
+      "step": 45855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2472768612556504e-05,
+      "loss": 2.8621,
+      "step": 45860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2463468641049486e-05,
+      "loss": 2.8639,
+      "step": 45865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.245416893632854e-05,
+      "loss": 2.8613,
+      "step": 45870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2444869498722875e-05,
+      "loss": 2.9872,
+      "step": 45875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2435570328561656e-05,
+      "loss": 2.9561,
+      "step": 45880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.24262714261741e-05,
+      "loss": 2.8905,
+      "step": 45885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.241697279188934e-05,
+      "loss": 2.9595,
+      "step": 45890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2407674426036565e-05,
+      "loss": 2.9239,
+      "step": 45895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.239837632894493e-05,
+      "loss": 2.8988,
+      "step": 45900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.238907850094356e-05,
+      "loss": 2.9393,
+      "step": 45905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.237978094236159e-05,
+      "loss": 2.985,
+      "step": 45910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2370483653528166e-05,
+      "loss": 2.9244,
+      "step": 45915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.23611866347724e-05,
+      "loss": 2.8644,
+      "step": 45920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.235188988642338e-05,
+      "loss": 2.8847,
+      "step": 45925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2342593408810224e-05,
+      "loss": 2.9363,
+      "step": 45930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.233329720226199e-05,
+      "loss": 2.7798,
+      "step": 45935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2324001267107774e-05,
+      "loss": 2.9514,
+      "step": 45940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.231470560367665e-05,
+      "loss": 2.8838,
+      "step": 45945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.230541021229766e-05,
+      "loss": 2.8373,
+      "step": 45950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.229611509329986e-05,
+      "loss": 2.8381,
+      "step": 45955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.228682024701228e-05,
+      "loss": 2.9839,
+      "step": 45960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.227752567376396e-05,
+      "loss": 2.8927,
+      "step": 45965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.226823137388392e-05,
+      "loss": 2.8626,
+      "step": 45970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2258937347701154e-05,
+      "loss": 2.8932,
+      "step": 45975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2249643595544685e-05,
+      "loss": 2.8299,
+      "step": 45980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2240350117743475e-05,
+      "loss": 2.8789,
+      "step": 45985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.223105691462652e-05,
+      "loss": 2.9513,
+      "step": 45990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2221763986522795e-05,
+      "loss": 2.9326,
+      "step": 45995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.221247133376125e-05,
+      "loss": 2.9212,
+      "step": 46000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.220317895667083e-05,
+      "loss": 2.8815,
+      "step": 46005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2193886855580503e-05,
+      "loss": 2.872,
+      "step": 46010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.218459503081919e-05,
+      "loss": 2.8351,
+      "step": 46015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.217530348271579e-05,
+      "loss": 2.9016,
+      "step": 46020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.216601221159925e-05,
+      "loss": 2.9412,
+      "step": 46025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.215672121779844e-05,
+      "loss": 2.8886,
+      "step": 46030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.214743050164229e-05,
+      "loss": 2.8877,
+      "step": 46035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.213814006345964e-05,
+      "loss": 2.9126,
+      "step": 46040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.212884990357939e-05,
+      "loss": 2.8604,
+      "step": 46045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2119560022330404e-05,
+      "loss": 2.795,
+      "step": 46050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.211027042004151e-05,
+      "loss": 2.9019,
+      "step": 46055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.210098109704159e-05,
+      "loss": 3.0329,
+      "step": 46060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.209169205365943e-05,
+      "loss": 3.0237,
+      "step": 46065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.208240329022389e-05,
+      "loss": 2.9359,
+      "step": 46070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2073114807063774e-05,
+      "loss": 2.8973,
+      "step": 46075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.206382660450788e-05,
+      "loss": 2.8184,
+      "step": 46080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.205453868288501e-05,
+      "loss": 2.8597,
+      "step": 46085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.204525104252393e-05,
+      "loss": 3.0093,
+      "step": 46090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.203596368375343e-05,
+      "loss": 2.8749,
+      "step": 46095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.202667660690228e-05,
+      "loss": 2.9225,
+      "step": 46100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.20173898122992e-05,
+      "loss": 2.7877,
+      "step": 46105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.200810330027297e-05,
+      "loss": 2.8578,
+      "step": 46110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.199881707115232e-05,
+      "loss": 2.9252,
+      "step": 46115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1989531125265954e-05,
+      "loss": 2.9572,
+      "step": 46120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.19802454629426e-05,
+      "loss": 2.8205,
+      "step": 46125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.197096008451097e-05,
+      "loss": 2.838,
+      "step": 46130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.196167499029973e-05,
+      "loss": 2.8229,
+      "step": 46135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.195239018063759e-05,
+      "loss": 2.9878,
+      "step": 46140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.194310565585321e-05,
+      "loss": 2.9164,
+      "step": 46145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.193382141627524e-05,
+      "loss": 2.9025,
+      "step": 46150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.192453746223238e-05,
+      "loss": 2.8686,
+      "step": 46155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.191525379405322e-05,
+      "loss": 2.8818,
+      "step": 46160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1905970412066424e-05,
+      "loss": 2.9291,
+      "step": 46165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.18966873166006e-05,
+      "loss": 2.9996,
+      "step": 46170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.188740450798436e-05,
+      "loss": 2.8877,
+      "step": 46175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.187812198654634e-05,
+      "loss": 3.0035,
+      "step": 46180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.186883975261509e-05,
+      "loss": 2.8529,
+      "step": 46185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.185955780651921e-05,
+      "loss": 2.9055,
+      "step": 46190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.185027614858726e-05,
+      "loss": 2.8951,
+      "step": 46195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.184099477914782e-05,
+      "loss": 2.9374,
+      "step": 46200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.183171369852942e-05,
+      "loss": 2.8652,
+      "step": 46205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.182243290706064e-05,
+      "loss": 2.9683,
+      "step": 46210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1813152405069977e-05,
+      "loss": 2.8662,
+      "step": 46215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.180387219288596e-05,
+      "loss": 2.9091,
+      "step": 46220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.179459227083712e-05,
+      "loss": 2.8944,
+      "step": 46225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.178531263925192e-05,
+      "loss": 2.8811,
+      "step": 46230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.177603329845888e-05,
+      "loss": 2.8309,
+      "step": 46235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.176675424878646e-05,
+      "loss": 2.9003,
+      "step": 46240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.175747549056315e-05,
+      "loss": 2.8819,
+      "step": 46245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1748197024117386e-05,
+      "loss": 2.8812,
+      "step": 46250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.173891884977763e-05,
+      "loss": 2.9532,
+      "step": 46255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.172964096787233e-05,
+      "loss": 2.8743,
+      "step": 46260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1720363378729884e-05,
+      "loss": 2.9055,
+      "step": 46265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.171108608267875e-05,
+      "loss": 2.9445,
+      "step": 46270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.17018090800473e-05,
+      "loss": 2.9106,
+      "step": 46275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.169253237116396e-05,
+      "loss": 2.847,
+      "step": 46280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1683255956357096e-05,
+      "loss": 2.8926,
+      "step": 46285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.167397983595509e-05,
+      "loss": 2.8943,
+      "step": 46290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.166470401028631e-05,
+      "loss": 2.961,
+      "step": 46295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.165542847967911e-05,
+      "loss": 2.9044,
+      "step": 46300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1646153244461826e-05,
+      "loss": 2.8442,
+      "step": 46305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.163687830496281e-05,
+      "loss": 2.9043,
+      "step": 46310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1627603661510386e-05,
+      "loss": 2.9731,
+      "step": 46315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.161832931443285e-05,
+      "loss": 2.8924,
+      "step": 46320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1609055264058515e-05,
+      "loss": 2.9293,
+      "step": 46325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1599781510715686e-05,
+      "loss": 2.9455,
+      "step": 46330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1590508054732616e-05,
+      "loss": 2.7547,
+      "step": 46335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.15812348964376e-05,
+      "loss": 2.9157,
+      "step": 46340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.15719620361589e-05,
+      "loss": 2.8783,
+      "step": 46345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.156268947422474e-05,
+      "loss": 2.9189,
+      "step": 46350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1553417210963395e-05,
+      "loss": 2.9555,
+      "step": 46355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.154414524670307e-05,
+      "loss": 3.0112,
+      "step": 46360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1534873581772004e-05,
+      "loss": 2.8767,
+      "step": 46365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1525602216498375e-05,
+      "loss": 2.808,
+      "step": 46370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.15163311512104e-05,
+      "loss": 2.8771,
+      "step": 46375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.150706038623628e-05,
+      "loss": 2.8555,
+      "step": 46380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.149778992190416e-05,
+      "loss": 2.9256,
+      "step": 46385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1488519758542236e-05,
+      "loss": 2.8909,
+      "step": 46390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.147924989647863e-05,
+      "loss": 2.8629,
+      "step": 46395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.14699803360415e-05,
+      "loss": 2.8153,
+      "step": 46400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.146071107755899e-05,
+      "loss": 2.8975,
+      "step": 46405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.145144212135923e-05,
+      "loss": 2.8647,
+      "step": 46410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1442173467770314e-05,
+      "loss": 2.9434,
+      "step": 46415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.143290511712035e-05,
+      "loss": 2.9089,
+      "step": 46420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.142363706973742e-05,
+      "loss": 2.8595,
+      "step": 46425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.141436932594962e-05,
+      "loss": 2.8656,
+      "step": 46430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1405101886085005e-05,
+      "loss": 2.8925,
+      "step": 46435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1395834750471646e-05,
+      "loss": 2.8886,
+      "step": 46440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1386567919437585e-05,
+      "loss": 2.9035,
+      "step": 46445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.137730139331085e-05,
+      "loss": 2.8529,
+      "step": 46450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1368035172419476e-05,
+      "loss": 2.8763,
+      "step": 46455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.135876925709149e-05,
+      "loss": 2.9854,
+      "step": 46460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.134950364765487e-05,
+      "loss": 2.8961,
+      "step": 46465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1340238344437645e-05,
+      "loss": 2.9256,
+      "step": 46470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.133097334776777e-05,
+      "loss": 2.9651,
+      "step": 46475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1321708657973204e-05,
+      "loss": 2.876,
+      "step": 46480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.131244427538196e-05,
+      "loss": 2.8265,
+      "step": 46485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1303180200321937e-05,
+      "loss": 2.8828,
+      "step": 46490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.129391643312111e-05,
+      "loss": 2.8942,
+      "step": 46495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.128465297410736e-05,
+      "loss": 2.9878,
+      "step": 46500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1275389823608655e-05,
+      "loss": 2.9373,
+      "step": 46505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.126612698195289e-05,
+      "loss": 2.8588,
+      "step": 46510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.125686444946796e-05,
+      "loss": 3.0465,
+      "step": 46515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.124760222648173e-05,
+      "loss": 2.8462,
+      "step": 46520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.12383403133221e-05,
+      "loss": 2.9066,
+      "step": 46525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.122907871031692e-05,
+      "loss": 2.8722,
+      "step": 46530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1219817417794035e-05,
+      "loss": 2.8896,
+      "step": 46535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.121055643608131e-05,
+      "loss": 2.9865,
+      "step": 46540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1201295765506535e-05,
+      "loss": 2.76,
+      "step": 46545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1192035406397576e-05,
+      "loss": 2.8682,
+      "step": 46550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.11827753590822e-05,
+      "loss": 2.8983,
+      "step": 46555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.117351562388824e-05,
+      "loss": 2.916,
+      "step": 46560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.116425620114346e-05,
+      "loss": 2.9135,
+      "step": 46565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.115499709117563e-05,
+      "loss": 2.8688,
+      "step": 46570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.114573829431253e-05,
+      "loss": 2.7582,
+      "step": 46575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1136479810881904e-05,
+      "loss": 2.9261,
+      "step": 46580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.112722164121148e-05,
+      "loss": 2.9466,
+      "step": 46585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.111796378562903e-05,
+      "loss": 2.8713,
+      "step": 46590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1108706244462224e-05,
+      "loss": 2.8752,
+      "step": 46595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.109944901803879e-05,
+      "loss": 2.8668,
+      "step": 46600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.109019210668643e-05,
+      "loss": 2.896,
+      "step": 46605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1080935510732845e-05,
+      "loss": 2.9954,
+      "step": 46610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1071679230505676e-05,
+      "loss": 2.808,
+      "step": 46615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1062423266332615e-05,
+      "loss": 3.0288,
+      "step": 46620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.10531676185413e-05,
+      "loss": 2.9004,
+      "step": 46625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.104391228745936e-05,
+      "loss": 2.9292,
+      "step": 46630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.103465727341446e-05,
+      "loss": 2.9372,
+      "step": 46635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1025402576734184e-05,
+      "loss": 3.0251,
+      "step": 46640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.101614819774617e-05,
+      "loss": 2.9851,
+      "step": 46645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.100689413677798e-05,
+      "loss": 3.0039,
+      "step": 46650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0997640394157225e-05,
+      "loss": 2.7701,
+      "step": 46655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.098838697021147e-05,
+      "loss": 2.8131,
+      "step": 46660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.097913386526827e-05,
+      "loss": 2.9037,
+      "step": 46665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.09698810796552e-05,
+      "loss": 2.8973,
+      "step": 46670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.096062861369977e-05,
+      "loss": 2.9522,
+      "step": 46675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.095137646772953e-05,
+      "loss": 2.8817,
+      "step": 46680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.094212464207198e-05,
+      "loss": 2.9178,
+      "step": 46685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.093287313705464e-05,
+      "loss": 2.9489,
+      "step": 46690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.092362195300501e-05,
+      "loss": 2.9444,
+      "step": 46695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.091437109025054e-05,
+      "loss": 2.8386,
+      "step": 46700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.090512054911873e-05,
+      "loss": 3.0204,
+      "step": 46705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.089587032993704e-05,
+      "loss": 2.8354,
+      "step": 46710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.088662043303293e-05,
+      "loss": 2.9653,
+      "step": 46715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.087737085873381e-05,
+      "loss": 2.8975,
+      "step": 46720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.086812160736712e-05,
+      "loss": 2.9286,
+      "step": 46725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.085887267926027e-05,
+      "loss": 2.9006,
+      "step": 46730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0849624074740664e-05,
+      "loss": 2.9301,
+      "step": 46735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.084037579413571e-05,
+      "loss": 2.831,
+      "step": 46740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.083112783777275e-05,
+      "loss": 2.7731,
+      "step": 46745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0821880205979204e-05,
+      "loss": 2.8841,
+      "step": 46750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.081263289908239e-05,
+      "loss": 2.8119,
+      "step": 46755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.080338591740966e-05,
+      "loss": 2.8395,
+      "step": 46760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.079413926128837e-05,
+      "loss": 2.9517,
+      "step": 46765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0784892931045817e-05,
+      "loss": 2.8429,
+      "step": 46770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0775646927009326e-05,
+      "loss": 2.9135,
+      "step": 46775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076640124950619e-05,
+      "loss": 2.8931,
+      "step": 46780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075715589886371e-05,
+      "loss": 2.9191,
+      "step": 46785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0747910875409145e-05,
+      "loss": 2.8451,
+      "step": 46790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0738666179469765e-05,
+      "loss": 2.8268,
+      "step": 46795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072942181137282e-05,
+      "loss": 2.9936,
+      "step": 46800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072017777144558e-05,
+      "loss": 2.8985,
+      "step": 46805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071093406001524e-05,
+      "loss": 2.9151,
+      "step": 46810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070169067740904e-05,
+      "loss": 2.9259,
+      "step": 46815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069244762395419e-05,
+      "loss": 3.0211,
+      "step": 46820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0683204899977854e-05,
+      "loss": 2.923,
+      "step": 46825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.067396250580726e-05,
+      "loss": 2.9365,
+      "step": 46830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0664720441769544e-05,
+      "loss": 2.8815,
+      "step": 46835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.065547870819188e-05,
+      "loss": 2.9395,
+      "step": 46840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064623730540142e-05,
+      "loss": 2.9099,
+      "step": 46845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063699623372529e-05,
+      "loss": 2.9113,
+      "step": 46850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062775549349063e-05,
+      "loss": 2.961,
+      "step": 46855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0618515085024536e-05,
+      "loss": 2.9347,
+      "step": 46860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0609275008654125e-05,
+      "loss": 2.9163,
+      "step": 46865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.060003526470649e-05,
+      "loss": 2.9602,
+      "step": 46870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.059079585350868e-05,
+      "loss": 2.8486,
+      "step": 46875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.05815567753878e-05,
+      "loss": 2.9528,
+      "step": 46880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0572318030670874e-05,
+      "loss": 2.9597,
+      "step": 46885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.056307961968496e-05,
+      "loss": 2.9143,
+      "step": 46890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.055384154275708e-05,
+      "loss": 2.9014,
+      "step": 46895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.054460380021427e-05,
+      "loss": 2.8904,
+      "step": 46900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0535366392383525e-05,
+      "loss": 2.9183,
+      "step": 46905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0526129319591844e-05,
+      "loss": 2.8029,
+      "step": 46910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.05168925821662e-05,
+      "loss": 3.019,
+      "step": 46915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.050765618043358e-05,
+      "loss": 2.9063,
+      "step": 46920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.049842011472094e-05,
+      "loss": 2.8586,
+      "step": 46925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0489184385355215e-05,
+      "loss": 2.891,
+      "step": 46930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0479948992663366e-05,
+      "loss": 2.8918,
+      "step": 46935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0470713936972296e-05,
+      "loss": 2.9762,
+      "step": 46940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.046147921860892e-05,
+      "loss": 2.8747,
+      "step": 46945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.045224483790015e-05,
+      "loss": 2.8804,
+      "step": 46950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.044301079517286e-05,
+      "loss": 2.8562,
+      "step": 46955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.043377709075394e-05,
+      "loss": 2.8776,
+      "step": 46960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.042454372497024e-05,
+      "loss": 2.9053,
+      "step": 46965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.041531069814861e-05,
+      "loss": 2.8825,
+      "step": 46970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0406078010615924e-05,
+      "loss": 2.8669,
+      "step": 46975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.039684566269897e-05,
+      "loss": 2.8216,
+      "step": 46980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.038761365472459e-05,
+      "loss": 2.8387,
+      "step": 46985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.037838198701956e-05,
+      "loss": 2.8807,
+      "step": 46990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0369150659910694e-05,
+      "loss": 2.9208,
+      "step": 46995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.035991967372478e-05,
+      "loss": 2.9135,
+      "step": 47000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.035068902878858e-05,
+      "loss": 2.7999,
+      "step": 47005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.034145872542884e-05,
+      "loss": 2.9015,
+      "step": 47010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.03322287639723e-05,
+      "loss": 2.9711,
+      "step": 47015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0322999144745724e-05,
+      "loss": 2.9344,
+      "step": 47020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0313769868075796e-05,
+      "loss": 2.9357,
+      "step": 47025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.030454093428925e-05,
+      "loss": 2.8744,
+      "step": 47030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.029531234371275e-05,
+      "loss": 2.8439,
+      "step": 47035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.028608409667302e-05,
+      "loss": 2.9063,
+      "step": 47040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.027685619349668e-05,
+      "loss": 2.9963,
+      "step": 47045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.026762863451043e-05,
+      "loss": 2.8812,
+      "step": 47050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.025840142004092e-05,
+      "loss": 2.8168,
+      "step": 47055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.024917455041475e-05,
+      "loss": 2.925,
+      "step": 47060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.023994802595858e-05,
+      "loss": 2.8959,
+      "step": 47065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.023072184699899e-05,
+      "loss": 2.9393,
+      "step": 47070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.022149601386259e-05,
+      "loss": 2.8875,
+      "step": 47075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.021227052687597e-05,
+      "loss": 2.9127,
+      "step": 47080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.02030453863657e-05,
+      "loss": 2.972,
+      "step": 47085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.019382059265835e-05,
+      "loss": 2.9442,
+      "step": 47090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0184596146080424e-05,
+      "loss": 2.9107,
+      "step": 47095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.017537204695853e-05,
+      "loss": 2.9184,
+      "step": 47100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.016614829561914e-05,
+      "loss": 2.8514,
+      "step": 47105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.015692489238879e-05,
+      "loss": 2.7951,
+      "step": 47110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0147701837593976e-05,
+      "loss": 2.8519,
+      "step": 47115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.013847913156117e-05,
+      "loss": 2.905,
+      "step": 47120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.012925677461688e-05,
+      "loss": 2.8208,
+      "step": 47125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0120034767087526e-05,
+      "loss": 2.9018,
+      "step": 47130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.01108131092996e-05,
+      "loss": 2.9018,
+      "step": 47135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0101591801579504e-05,
+      "loss": 2.9377,
+      "step": 47140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.009237084425368e-05,
+      "loss": 2.8651,
+      "step": 47145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.008315023764855e-05,
+      "loss": 2.8963,
+      "step": 47150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.00739299820905e-05,
+      "loss": 2.8775,
+      "step": 47155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.006471007790594e-05,
+      "loss": 2.9241,
+      "step": 47160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.005549052542121e-05,
+      "loss": 2.9118,
+      "step": 47165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.00462713249627e-05,
+      "loss": 2.9593,
+      "step": 47170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.003705247685675e-05,
+      "loss": 2.8568,
+      "step": 47175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.00278339814297e-05,
+      "loss": 2.9078,
+      "step": 47180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.001861583900789e-05,
+      "loss": 2.8712,
+      "step": 47185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0009398049917615e-05,
+      "loss": 2.9577,
+      "step": 47190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.000018061448516e-05,
+      "loss": 2.8068,
+      "step": 47195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.999096353303686e-05,
+      "loss": 2.9098,
+      "step": 47200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.998174680589898e-05,
+      "loss": 2.8668,
+      "step": 47205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9972530433397756e-05,
+      "loss": 2.9187,
+      "step": 47210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.996331441585946e-05,
+      "loss": 2.8828,
+      "step": 47215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.995409875361032e-05,
+      "loss": 2.8497,
+      "step": 47220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.994488344697656e-05,
+      "loss": 2.9433,
+      "step": 47225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9935668496284414e-05,
+      "loss": 2.864,
+      "step": 47230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992645390186006e-05,
+      "loss": 2.8741,
+      "step": 47235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9917239664029695e-05,
+      "loss": 2.8827,
+      "step": 47240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9908025783119486e-05,
+      "loss": 2.8828,
+      "step": 47245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9898812259455595e-05,
+      "loss": 2.974,
+      "step": 47250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.98895990933642e-05,
+      "loss": 2.952,
+      "step": 47255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.988038628517141e-05,
+      "loss": 2.9087,
+      "step": 47260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.987117383520336e-05,
+      "loss": 2.9613,
+      "step": 47265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9861961743786155e-05,
+      "loss": 2.8681,
+      "step": 47270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.985275001124589e-05,
+      "loss": 2.9049,
+      "step": 47275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9843538637908674e-05,
+      "loss": 2.7775,
+      "step": 47280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9834327624100557e-05,
+      "loss": 3.0006,
+      "step": 47285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9825116970147604e-05,
+      "loss": 2.8467,
+      "step": 47290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.981590667637589e-05,
+      "loss": 2.9152,
+      "step": 47295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.980669674311142e-05,
+      "loss": 2.9904,
+      "step": 47300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.979748717068022e-05,
+      "loss": 2.876,
+      "step": 47305
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9788277959408324e-05,
+      "loss": 2.9096,
+      "step": 47310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.97790691096217e-05,
+      "loss": 2.9134,
+      "step": 47315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9769860621646354e-05,
+      "loss": 2.9127,
+      "step": 47320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9760652495808244e-05,
+      "loss": 2.8876,
+      "step": 47325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.975144473243333e-05,
+      "loss": 2.891,
+      "step": 47330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9742237331847575e-05,
+      "loss": 2.7757,
+      "step": 47335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.973303029437689e-05,
+      "loss": 2.9254,
+      "step": 47340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.972382362034721e-05,
+      "loss": 3.0371,
+      "step": 47345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.971461731008443e-05,
+      "loss": 3.0135,
+      "step": 47350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.970541136391445e-05,
+      "loss": 2.861,
+      "step": 47355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.969620578216317e-05,
+      "loss": 2.8666,
+      "step": 47360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.968700056515643e-05,
+      "loss": 2.8419,
+      "step": 47365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.96777957132201e-05,
+      "loss": 2.7882,
+      "step": 47370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.966859122668003e-05,
+      "loss": 2.9215,
+      "step": 47375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.965938710586202e-05,
+      "loss": 2.9293,
+      "step": 47380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9650183351091927e-05,
+      "loss": 2.9132,
+      "step": 47385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9640979962695525e-05,
+      "loss": 2.7945,
+      "step": 47390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.963177694099862e-05,
+      "loss": 2.951,
+      "step": 47395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.962257428632699e-05,
+      "loss": 2.888,
+      "step": 47400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.961337199900641e-05,
+      "loss": 2.9584,
+      "step": 47405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9604170079362604e-05,
+      "loss": 2.8921,
+      "step": 47410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.959496852772134e-05,
+      "loss": 2.948,
+      "step": 47415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.958576734440832e-05,
+      "loss": 2.7739,
+      "step": 47420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.957656652974928e-05,
+      "loss": 2.9554,
+      "step": 47425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.95673660840699e-05,
+      "loss": 2.9182,
+      "step": 47430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.955816600769588e-05,
+      "loss": 3.0563,
+      "step": 47435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.95489663009529e-05,
+      "loss": 2.9016,
+      "step": 47440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.953976696416659e-05,
+      "loss": 2.981,
+      "step": 47445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.953056799766264e-05,
+      "loss": 2.9309,
+      "step": 47450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9521369401766645e-05,
+      "loss": 2.8518,
+      "step": 47455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.951217117680425e-05,
+      "loss": 2.8675,
+      "step": 47460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.950297332310106e-05,
+      "loss": 2.9341,
+      "step": 47465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.949377584098267e-05,
+      "loss": 2.9461,
+      "step": 47470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9484578730774664e-05,
+      "loss": 2.9444,
+      "step": 47475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9475381992802594e-05,
+      "loss": 2.8834,
+      "step": 47480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9466185627392036e-05,
+      "loss": 2.8655,
+      "step": 47485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9456989634868516e-05,
+      "loss": 2.887,
+      "step": 47490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.944779401555759e-05,
+      "loss": 3.0046,
+      "step": 47495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.943859876978475e-05,
+      "loss": 2.9412,
+      "step": 47500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9429403897875504e-05,
+      "loss": 2.9352,
+      "step": 47505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.942020940015536e-05,
+      "loss": 2.9832,
+      "step": 47510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.941101527694977e-05,
+      "loss": 2.8766,
+      "step": 47515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.940182152858421e-05,
+      "loss": 2.9766,
+      "step": 47520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.939262815538412e-05,
+      "loss": 2.9218,
+      "step": 47525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.938343515767494e-05,
+      "loss": 2.9232,
+      "step": 47530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9374242535782113e-05,
+      "loss": 2.8589,
+      "step": 47535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.936505029003102e-05,
+      "loss": 2.9191,
+      "step": 47540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9355858420747075e-05,
+      "loss": 2.9161,
+      "step": 47545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.934666692825565e-05,
+      "loss": 2.8668,
+      "step": 47550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.933747581288213e-05,
+      "loss": 2.9219,
+      "step": 47555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.932828507495185e-05,
+      "loss": 2.8077,
+      "step": 47560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.931909471479018e-05,
+      "loss": 2.8329,
+      "step": 47565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.930990473272244e-05,
+      "loss": 2.8854,
+      "step": 47570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.930071512907393e-05,
+      "loss": 2.8931,
+      "step": 47575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9291525904169985e-05,
+      "loss": 2.9376,
+      "step": 47580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.928233705833586e-05,
+      "loss": 2.8085,
+      "step": 47585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.927314859189684e-05,
+      "loss": 2.9867,
+      "step": 47590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.926396050517821e-05,
+      "loss": 2.8699,
+      "step": 47595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9254772798505216e-05,
+      "loss": 2.9526,
+      "step": 47600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.924558547220307e-05,
+      "loss": 2.8566,
+      "step": 47605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9236398526597006e-05,
+      "loss": 2.9072,
+      "step": 47610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.922721196201225e-05,
+      "loss": 2.9946,
+      "step": 47615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.921802577877397e-05,
+      "loss": 2.8434,
+      "step": 47620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.920883997720738e-05,
+      "loss": 2.867,
+      "step": 47625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.919965455763761e-05,
+      "loss": 2.8458,
+      "step": 47630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.919046952038984e-05,
+      "loss": 2.8327,
+      "step": 47635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.918128486578922e-05,
+      "loss": 2.9118,
+      "step": 47640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.917210059416085e-05,
+      "loss": 2.9403,
+      "step": 47645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.916291670582988e-05,
+      "loss": 2.8619,
+      "step": 47650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.915373320112137e-05,
+      "loss": 2.9422,
+      "step": 47655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.914455008036043e-05,
+      "loss": 2.9673,
+      "step": 47660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.913536734387214e-05,
+      "loss": 2.8214,
+      "step": 47665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.912618499198154e-05,
+      "loss": 2.8908,
+      "step": 47670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.911700302501371e-05,
+      "loss": 2.9251,
+      "step": 47675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9107821443293645e-05,
+      "loss": 2.8867,
+      "step": 47680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.909864024714637e-05,
+      "loss": 2.8728,
+      "step": 47685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.908945943689691e-05,
+      "loss": 2.8925,
+      "step": 47690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.908027901287026e-05,
+      "loss": 2.9217,
+      "step": 47695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.907109897539138e-05,
+      "loss": 2.9156,
+      "step": 47700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.906191932478525e-05,
+      "loss": 2.8542,
+      "step": 47705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9052740061376796e-05,
+      "loss": 2.7797,
+      "step": 47710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.904356118549098e-05,
+      "loss": 2.9808,
+      "step": 47715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.903438269745272e-05,
+      "loss": 2.8401,
+      "step": 47720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.902520459758692e-05,
+      "loss": 2.8259,
+      "step": 47725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.901602688621848e-05,
+      "loss": 2.84,
+      "step": 47730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.900684956367228e-05,
+      "loss": 2.9451,
+      "step": 47735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.899767263027318e-05,
+      "loss": 2.8464,
+      "step": 47740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.898849608634606e-05,
+      "loss": 2.9228,
+      "step": 47745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.897931993221574e-05,
+      "loss": 2.8041,
+      "step": 47750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.897014416820706e-05,
+      "loss": 2.9335,
+      "step": 47755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.896096879464481e-05,
+      "loss": 2.9454,
+      "step": 47760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.89517938118538e-05,
+      "loss": 2.8469,
+      "step": 47765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8942619220158846e-05,
+      "loss": 2.8748,
+      "step": 47770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.893344501988468e-05,
+      "loss": 2.9036,
+      "step": 47775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.892427121135608e-05,
+      "loss": 2.8908,
+      "step": 47780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.891509779489776e-05,
+      "loss": 2.8807,
+      "step": 47785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.89059247708345e-05,
+      "loss": 2.9371,
+      "step": 47790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.889675213949099e-05,
+      "loss": 2.8182,
+      "step": 47795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.888757990119194e-05,
+      "loss": 2.8198,
+      "step": 47800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8878408056262014e-05,
+      "loss": 2.9618,
+      "step": 47805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8869236605025926e-05,
+      "loss": 2.9422,
+      "step": 47810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.886006554780831e-05,
+      "loss": 2.9643,
+      "step": 47815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.885089488493381e-05,
+      "loss": 2.9366,
+      "step": 47820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8841724616727084e-05,
+      "loss": 2.9277,
+      "step": 47825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.883255474351272e-05,
+      "loss": 2.8826,
+      "step": 47830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.882338526561534e-05,
+      "loss": 2.8456,
+      "step": 47835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.881421618335953e-05,
+      "loss": 2.9391,
+      "step": 47840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.880504749706987e-05,
+      "loss": 2.8583,
+      "step": 47845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.879587920707093e-05,
+      "loss": 2.8946,
+      "step": 47850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.878671131368724e-05,
+      "loss": 2.8808,
+      "step": 47855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.877754381724335e-05,
+      "loss": 2.9189,
+      "step": 47860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.876837671806377e-05,
+      "loss": 2.9125,
+      "step": 47865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8759210016473016e-05,
+      "loss": 2.9662,
+      "step": 47870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8750043712795583e-05,
+      "loss": 2.8462,
+      "step": 47875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8740877807355936e-05,
+      "loss": 2.8777,
+      "step": 47880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.873171230047853e-05,
+      "loss": 2.8408,
+      "step": 47885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.872254719248785e-05,
+      "loss": 3.0047,
+      "step": 47890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.871338248370832e-05,
+      "loss": 2.816,
+      "step": 47895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.870421817446435e-05,
+      "loss": 2.8636,
+      "step": 47900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.869505426508036e-05,
+      "loss": 2.8712,
+      "step": 47905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8685890755880724e-05,
+      "loss": 2.9863,
+      "step": 47910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.867672764718985e-05,
+      "loss": 2.824,
+      "step": 47915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8667564939332085e-05,
+      "loss": 2.8438,
+      "step": 47920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.865840263263177e-05,
+      "loss": 2.9068,
+      "step": 47925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8649240727413286e-05,
+      "loss": 3.0574,
+      "step": 47930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.86400792240009e-05,
+      "loss": 2.8091,
+      "step": 47935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.863091812271896e-05,
+      "loss": 2.8368,
+      "step": 47940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8621757423891735e-05,
+      "loss": 2.8722,
+      "step": 47945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.861259712784352e-05,
+      "loss": 2.8465,
+      "step": 47950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.860343723489858e-05,
+      "loss": 2.9623,
+      "step": 47955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8594277745381155e-05,
+      "loss": 2.979,
+      "step": 47960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.858511865961551e-05,
+      "loss": 2.961,
+      "step": 47965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8575959977925835e-05,
+      "loss": 2.9576,
+      "step": 47970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8566801700636354e-05,
+      "loss": 2.8747,
+      "step": 47975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8557643828071264e-05,
+      "loss": 2.9505,
+      "step": 47980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8548486360554726e-05,
+      "loss": 2.9375,
+      "step": 47985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8539329298410925e-05,
+      "loss": 2.9338,
+      "step": 47990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.853017264196401e-05,
+      "loss": 2.9505,
+      "step": 47995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.852101639153814e-05,
+      "loss": 2.9607,
+      "step": 48000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.851186054745738e-05,
+      "loss": 2.8576,
+      "step": 48005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.85027051100459e-05,
+      "loss": 3.0187,
+      "step": 48010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.849355007962774e-05,
+      "loss": 2.9327,
+      "step": 48015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8484395456527015e-05,
+      "loss": 2.9188,
+      "step": 48020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.847524124106779e-05,
+      "loss": 2.86,
+      "step": 48025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.84660874335741e-05,
+      "loss": 2.9125,
+      "step": 48030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.845693403436999e-05,
+      "loss": 2.9415,
+      "step": 48035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.844778104377947e-05,
+      "loss": 2.9056,
+      "step": 48040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.843862846212655e-05,
+      "loss": 2.9567,
+      "step": 48045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8429476289735246e-05,
+      "loss": 2.8412,
+      "step": 48050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.842032452692951e-05,
+      "loss": 2.8611,
+      "step": 48055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8411173174033316e-05,
+      "loss": 2.8843,
+      "step": 48060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.84020222313706e-05,
+      "loss": 2.9259,
+      "step": 48065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.839287169926533e-05,
+      "loss": 2.8577,
+      "step": 48070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8383721578041386e-05,
+      "loss": 2.7741,
+      "step": 48075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8374571868022685e-05,
+      "loss": 2.8483,
+      "step": 48080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.836542256953313e-05,
+      "loss": 2.9058,
+      "step": 48085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.835627368289661e-05,
+      "loss": 2.8819,
+      "step": 48090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.834712520843695e-05,
+      "loss": 2.9778,
+      "step": 48095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8337977146478026e-05,
+      "loss": 2.9398,
+      "step": 48100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8328829497343666e-05,
+      "loss": 2.9269,
+      "step": 48105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8319682261357677e-05,
+      "loss": 2.7758,
+      "step": 48110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8310535438843875e-05,
+      "loss": 2.8069,
+      "step": 48115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.830138903012604e-05,
+      "loss": 2.8387,
+      "step": 48120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.829224303552794e-05,
+      "loss": 2.8966,
+      "step": 48125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.828309745537336e-05,
+      "loss": 2.8065,
+      "step": 48130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8273952289986014e-05,
+      "loss": 2.9595,
+      "step": 48135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8264807539689664e-05,
+      "loss": 2.9614,
+      "step": 48140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.825566320480799e-05,
+      "loss": 2.8909,
+      "step": 48145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.824651928566472e-05,
+      "loss": 2.9893,
+      "step": 48150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.823737578258353e-05,
+      "loss": 2.9216,
+      "step": 48155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.822823269588809e-05,
+      "loss": 2.9665,
+      "step": 48160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8219090025902065e-05,
+      "loss": 2.9953,
+      "step": 48165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8209947772949084e-05,
+      "loss": 2.8871,
+      "step": 48170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.820080593735278e-05,
+      "loss": 2.8856,
+      "step": 48175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.819166451943675e-05,
+      "loss": 2.814,
+      "step": 48180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8182523519524635e-05,
+      "loss": 2.9595,
+      "step": 48185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.817338293793998e-05,
+      "loss": 2.9174,
+      "step": 48190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.816424277500637e-05,
+      "loss": 3.0236,
+      "step": 48195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.815510303104735e-05,
+      "loss": 2.7529,
+      "step": 48200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.814596370638645e-05,
+      "loss": 2.9175,
+      "step": 48205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.813682480134722e-05,
+      "loss": 2.8647,
+      "step": 48210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.812768631625313e-05,
+      "loss": 2.8905,
+      "step": 48215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.811854825142772e-05,
+      "loss": 2.8677,
+      "step": 48220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.810941060719443e-05,
+      "loss": 3.0172,
+      "step": 48225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.810027338387674e-05,
+      "loss": 2.8289,
+      "step": 48230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.80911365817981e-05,
+      "loss": 2.9624,
+      "step": 48235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8082000201281936e-05,
+      "loss": 2.7746,
+      "step": 48240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8072864242651684e-05,
+      "loss": 2.8172,
+      "step": 48245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.806372870623073e-05,
+      "loss": 2.8942,
+      "step": 48250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.805459359234247e-05,
+      "loss": 2.9034,
+      "step": 48255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8045458901310275e-05,
+      "loss": 2.9592,
+      "step": 48260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.803632463345751e-05,
+      "loss": 2.9327,
+      "step": 48265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.802719078910753e-05,
+      "loss": 2.8712,
+      "step": 48270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.801805736858364e-05,
+      "loss": 2.9077,
+      "step": 48275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.800892437220915e-05,
+      "loss": 2.9915,
+      "step": 48280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.799979180030739e-05,
+      "loss": 2.8929,
+      "step": 48285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7990659653201644e-05,
+      "loss": 2.8936,
+      "step": 48290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.798152793121515e-05,
+      "loss": 2.9275,
+      "step": 48295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.797239663467118e-05,
+      "loss": 2.92,
+      "step": 48300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.796326576389299e-05,
+      "loss": 2.9091,
+      "step": 48305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7954135319203775e-05,
+      "loss": 2.9722,
+      "step": 48310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.794500530092677e-05,
+      "loss": 2.9169,
+      "step": 48315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7935875709385135e-05,
+      "loss": 2.8924,
+      "step": 48320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.792674654490209e-05,
+      "loss": 2.9191,
+      "step": 48325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7917617807800764e-05,
+      "loss": 2.9367,
+      "step": 48330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.790848949840432e-05,
+      "loss": 2.8349,
+      "step": 48335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.78993616170359e-05,
+      "loss": 2.8805,
+      "step": 48340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7890234164018604e-05,
+      "loss": 2.9629,
+      "step": 48345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7881107139675555e-05,
+      "loss": 2.9863,
+      "step": 48350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7871980544329824e-05,
+      "loss": 2.9692,
+      "step": 48355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7862854378304485e-05,
+      "loss": 3.0062,
+      "step": 48360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7853728641922614e-05,
+      "loss": 2.9182,
+      "step": 48365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.784460333550723e-05,
+      "loss": 2.9383,
+      "step": 48370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.783547845938138e-05,
+      "loss": 2.8813,
+      "step": 48375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.782635401386804e-05,
+      "loss": 2.8133,
+      "step": 48380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.781722999929026e-05,
+      "loss": 2.9173,
+      "step": 48385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.780810641597099e-05,
+      "loss": 2.9106,
+      "step": 48390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7798983264233204e-05,
+      "loss": 2.7891,
+      "step": 48395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7789860544399845e-05,
+      "loss": 2.9146,
+      "step": 48400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.778073825679386e-05,
+      "loss": 2.9019,
+      "step": 48405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7771616401738165e-05,
+      "loss": 2.9138,
+      "step": 48410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.776249497955565e-05,
+      "loss": 2.8921,
+      "step": 48415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7753373990569236e-05,
+      "loss": 3.0236,
+      "step": 48420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.774425343510176e-05,
+      "loss": 2.9319,
+      "step": 48425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.773513331347611e-05,
+      "loss": 2.9439,
+      "step": 48430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.772601362601512e-05,
+      "loss": 2.8415,
+      "step": 48435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.771689437304162e-05,
+      "loss": 2.8698,
+      "step": 48440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.770777555487842e-05,
+      "loss": 2.9346,
+      "step": 48445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7698657171848316e-05,
+      "loss": 2.8633,
+      "step": 48450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7689539224274096e-05,
+      "loss": 2.9813,
+      "step": 48455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.768042171247853e-05,
+      "loss": 2.9688,
+      "step": 48460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7671304636784344e-05,
+      "loss": 2.8692,
+      "step": 48465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.766218799751431e-05,
+      "loss": 2.957,
+      "step": 48470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.765307179499112e-05,
+      "loss": 2.8645,
+      "step": 48475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.764395602953748e-05,
+      "loss": 2.9209,
+      "step": 48480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.76348407014761e-05,
+      "loss": 2.8727,
+      "step": 48485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.762572581112965e-05,
+      "loss": 2.8281,
+      "step": 48490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.761661135882078e-05,
+      "loss": 2.8714,
+      "step": 48495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.760749734487214e-05,
+      "loss": 2.9104,
+      "step": 48500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.759838376960634e-05,
+      "loss": 2.8229,
+      "step": 48505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.758927063334601e-05,
+      "loss": 2.914,
+      "step": 48510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7580157936413746e-05,
+      "loss": 2.8605,
+      "step": 48515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.757104567913211e-05,
+      "loss": 2.9233,
+      "step": 48520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.75619338618237e-05,
+      "loss": 2.8629,
+      "step": 48525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.755282248481103e-05,
+      "loss": 2.7977,
+      "step": 48530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7543711548416646e-05,
+      "loss": 2.903,
+      "step": 48535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7534601052963094e-05,
+      "loss": 2.8671,
+      "step": 48540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.752549099877284e-05,
+      "loss": 2.9764,
+      "step": 48545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7516381386168386e-05,
+      "loss": 2.8763,
+      "step": 48550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.75072722154722e-05,
+      "loss": 2.8577,
+      "step": 48555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.749816348700674e-05,
+      "loss": 2.9459,
+      "step": 48560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.748905520109446e-05,
+      "loss": 2.8926,
+      "step": 48565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.747994735805776e-05,
+      "loss": 2.9773,
+      "step": 48570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.747083995821905e-05,
+      "loss": 2.868,
+      "step": 48575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.746173300190075e-05,
+      "loss": 2.9763,
+      "step": 48580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7452626489425224e-05,
+      "loss": 2.8758,
+      "step": 48585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.744352042111482e-05,
+      "loss": 3.0054,
+      "step": 48590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7434414797291913e-05,
+      "loss": 2.9132,
+      "step": 48595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7425309618278805e-05,
+      "loss": 2.9997,
+      "step": 48600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7416204884397834e-05,
+      "loss": 2.9118,
+      "step": 48605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.740710059597127e-05,
+      "loss": 2.902,
+      "step": 48610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7397996753321416e-05,
+      "loss": 2.9642,
+      "step": 48615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7388893356770546e-05,
+      "loss": 2.9379,
+      "step": 48620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.73797904066409e-05,
+      "loss": 2.8313,
+      "step": 48625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.737068790325472e-05,
+      "loss": 2.9639,
+      "step": 48630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.736158584693421e-05,
+      "loss": 3.0014,
+      "step": 48635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7352484238001596e-05,
+      "loss": 2.9109,
+      "step": 48640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.734338307677906e-05,
+      "loss": 2.8968,
+      "step": 48645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.733428236358876e-05,
+      "loss": 2.8546,
+      "step": 48650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.732518209875288e-05,
+      "loss": 2.9277,
+      "step": 48655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7316082282593526e-05,
+      "loss": 2.8942,
+      "step": 48660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.730698291543284e-05,
+      "loss": 2.9309,
+      "step": 48665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7297883997592946e-05,
+      "loss": 2.8228,
+      "step": 48670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.728878552939591e-05,
+      "loss": 2.8916,
+      "step": 48675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.727968751116382e-05,
+      "loss": 2.9722,
+      "step": 48680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.727058994321875e-05,
+      "loss": 2.9328,
+      "step": 48685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.726149282588274e-05,
+      "loss": 2.8925,
+      "step": 48690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.725239615947781e-05,
+      "loss": 2.9602,
+      "step": 48695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7243299944325984e-05,
+      "loss": 2.8013,
+      "step": 48700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.723420418074924e-05,
+      "loss": 2.8224,
+      "step": 48705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.722510886906959e-05,
+      "loss": 2.9739,
+      "step": 48710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.721601400960897e-05,
+      "loss": 2.9141,
+      "step": 48715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7206919602689336e-05,
+      "loss": 2.8596,
+      "step": 48720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.719782564863264e-05,
+      "loss": 2.883,
+      "step": 48725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.718873214776078e-05,
+      "loss": 2.8827,
+      "step": 48730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.717963910039567e-05,
+      "loss": 2.8308,
+      "step": 48735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7170546506859174e-05,
+      "loss": 2.9242,
+      "step": 48740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.716145436747318e-05,
+      "loss": 2.8897,
+      "step": 48745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.715236268255955e-05,
+      "loss": 2.8971,
+      "step": 48750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7143271452440095e-05,
+      "loss": 2.9043,
+      "step": 48755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7134180677436656e-05,
+      "loss": 2.8103,
+      "step": 48760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.712509035787103e-05,
+      "loss": 2.9491,
+      "step": 48765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7116000494065e-05,
+      "loss": 2.9986,
+      "step": 48770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.710691108634034e-05,
+      "loss": 2.9472,
+      "step": 48775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.709782213501882e-05,
+      "loss": 2.9606,
+      "step": 48780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.708873364042218e-05,
+      "loss": 2.891,
+      "step": 48785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.707964560287212e-05,
+      "loss": 2.8307,
+      "step": 48790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.707055802269038e-05,
+      "loss": 2.9449,
+      "step": 48795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7061470900198626e-05,
+      "loss": 2.8906,
+      "step": 48800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.705238423571855e-05,
+      "loss": 2.8041,
+      "step": 48805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.70432980295718e-05,
+      "loss": 2.9584,
+      "step": 48810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7034212282080015e-05,
+      "loss": 2.8211,
+      "step": 48815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.702512699356485e-05,
+      "loss": 2.9108,
+      "step": 48820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.701604216434788e-05,
+      "loss": 2.9545,
+      "step": 48825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.700695779475073e-05,
+      "loss": 2.9304,
+      "step": 48830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.699787388509495e-05,
+      "loss": 2.8371,
+      "step": 48835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.698879043570211e-05,
+      "loss": 3.0229,
+      "step": 48840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.697970744689378e-05,
+      "loss": 2.9218,
+      "step": 48845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.697062491899145e-05,
+      "loss": 2.904,
+      "step": 48850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.6961542852316674e-05,
+      "loss": 2.9405,
+      "step": 48855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.6952461247190915e-05,
+      "loss": 2.9303,
+      "step": 48860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.694338010393567e-05,
+      "loss": 2.7948,
+      "step": 48865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.693429942287239e-05,
+      "loss": 2.9561,
+      "step": 48870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.692521920432252e-05,
+      "loss": 2.9002,
+      "step": 48875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.691613944860751e-05,
+      "loss": 2.9315,
+      "step": 48880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.690706015604879e-05,
+      "loss": 2.8629,
+      "step": 48885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.6897981326967714e-05,
+      "loss": 2.9275,
+      "step": 48890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.688890296168568e-05,
+      "loss": 2.8694,
+      "step": 48895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6879825060524076e-05,
+      "loss": 2.9704,
+      "step": 48900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6870747623804225e-05,
+      "loss": 2.893,
+      "step": 48905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.686167065184748e-05,
+      "loss": 2.856,
+      "step": 48910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.685259414497513e-05,
+      "loss": 2.8669,
+      "step": 48915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6843518103508486e-05,
+      "loss": 2.8622,
+      "step": 48920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.683444252776885e-05,
+      "loss": 2.855,
+      "step": 48925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.682536741807747e-05,
+      "loss": 2.9137,
+      "step": 48930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6816292774755606e-05,
+      "loss": 2.9132,
+      "step": 48935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.680721859812448e-05,
+      "loss": 2.9709,
+      "step": 48940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.679814488850531e-05,
+      "loss": 2.9749,
+      "step": 48945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.678907164621932e-05,
+      "loss": 2.9606,
+      "step": 48950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.677999887158766e-05,
+      "loss": 2.8376,
+      "step": 48955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6770926564931526e-05,
+      "loss": 2.9032,
+      "step": 48960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.676185472657204e-05,
+      "loss": 2.8845,
+      "step": 48965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6752783356830354e-05,
+      "loss": 2.9383,
+      "step": 48970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.674371245602759e-05,
+      "loss": 2.9259,
+      "step": 48975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6734642024484863e-05,
+      "loss": 2.8478,
+      "step": 48980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6725572062523225e-05,
+      "loss": 2.7759,
+      "step": 48985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.671650257046376e-05,
+      "loss": 2.842,
+      "step": 48990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6707433548627526e-05,
+      "loss": 2.8927,
+      "step": 48995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.669836499733553e-05,
+      "loss": 2.9744,
+      "step": 49000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.668929691690883e-05,
+      "loss": 2.8655,
+      "step": 49005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.66802293076684e-05,
+      "loss": 2.8677,
+      "step": 49010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6671162169935245e-05,
+      "loss": 2.9014,
+      "step": 49015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6662095504030305e-05,
+      "loss": 2.9393,
+      "step": 49020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.665302931027455e-05,
+      "loss": 2.8877,
+      "step": 49025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.664396358898891e-05,
+      "loss": 2.897,
+      "step": 49030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.663489834049431e-05,
+      "loss": 2.8611,
+      "step": 49035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6625833565111645e-05,
+      "loss": 2.9342,
+      "step": 49040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6616769263161786e-05,
+      "loss": 2.8854,
+      "step": 49045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.660770543496562e-05,
+      "loss": 2.8525,
+      "step": 49050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.659864208084401e-05,
+      "loss": 2.8949,
+      "step": 49055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.658957920111775e-05,
+      "loss": 2.8803,
+      "step": 49060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.65805167961077e-05,
+      "loss": 2.947,
+      "step": 49065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6571454866134605e-05,
+      "loss": 2.8333,
+      "step": 49070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.656239341151932e-05,
+      "loss": 2.8775,
+      "step": 49075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.655333243258257e-05,
+      "loss": 2.8715,
+      "step": 49080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.654427192964511e-05,
+      "loss": 2.9054,
+      "step": 49085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6535211903027665e-05,
+      "loss": 2.8768,
+      "step": 49090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.652615235305097e-05,
+      "loss": 2.8595,
+      "step": 49095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.651709328003572e-05,
+      "loss": 2.9802,
+      "step": 49100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6508034684302587e-05,
+      "loss": 2.9048,
+      "step": 49105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.649897656617224e-05,
+      "loss": 2.9519,
+      "step": 49110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.648991892596533e-05,
+      "loss": 2.8505,
+      "step": 49115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.648086176400251e-05,
+      "loss": 2.9025,
+      "step": 49120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6471805080604345e-05,
+      "loss": 2.9404,
+      "step": 49125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.646274887609147e-05,
+      "loss": 2.8537,
+      "step": 49130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6453693150784476e-05,
+      "loss": 2.9159,
+      "step": 49135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6444637905003883e-05,
+      "loss": 2.886,
+      "step": 49140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6435583139070295e-05,
+      "loss": 2.8481,
+      "step": 49145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.642652885330419e-05,
+      "loss": 2.8969,
+      "step": 49150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6417475048026104e-05,
+      "loss": 2.9117,
+      "step": 49155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.640842172355654e-05,
+      "loss": 2.8943,
+      "step": 49160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6399368880215966e-05,
+      "loss": 2.8972,
+      "step": 49165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.639031651832483e-05,
+      "loss": 2.9753,
+      "step": 49170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6381264638203606e-05,
+      "loss": 2.9286,
+      "step": 49175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.637221324017273e-05,
+      "loss": 2.8801,
+      "step": 49180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6363162324552567e-05,
+      "loss": 2.9162,
+      "step": 49185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6354111891663555e-05,
+      "loss": 2.8548,
+      "step": 49190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.634506194182604e-05,
+      "loss": 2.8518,
+      "step": 49195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6336012475360404e-05,
+      "loss": 3.0207,
+      "step": 49200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.632696349258698e-05,
+      "loss": 2.7986,
+      "step": 49205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6317914993826086e-05,
+      "loss": 2.9907,
+      "step": 49210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.630886697939805e-05,
+      "loss": 2.9752,
+      "step": 49215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.629981944962314e-05,
+      "loss": 2.8634,
+      "step": 49220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6290772404821635e-05,
+      "loss": 2.8259,
+      "step": 49225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.628172584531382e-05,
+      "loss": 2.9336,
+      "step": 49230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.627267977141989e-05,
+      "loss": 2.8559,
+      "step": 49235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.62636341834601e-05,
+      "loss": 2.9081,
+      "step": 49240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6254589081754644e-05,
+      "loss": 2.937,
+      "step": 49245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.624554446662371e-05,
+      "loss": 2.952,
+      "step": 49250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.623650033838747e-05,
+      "loss": 2.8711,
+      "step": 49255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.622745669736607e-05,
+      "loss": 2.8836,
+      "step": 49260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6218413543879673e-05,
+      "loss": 2.7847,
+      "step": 49265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.620937087824835e-05,
+      "loss": 2.9955,
+      "step": 49270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.620032870079224e-05,
+      "loss": 2.8989,
+      "step": 49275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.619128701183143e-05,
+      "loss": 2.882,
+      "step": 49280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6182245811685975e-05,
+      "loss": 2.9945,
+      "step": 49285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6173205100675925e-05,
+      "loss": 2.8449,
+      "step": 49290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6164164879121317e-05,
+      "loss": 2.8826,
+      "step": 49295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.615512514734215e-05,
+      "loss": 2.968,
+      "step": 49300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6146085905658444e-05,
+      "loss": 2.8794,
+      "step": 49305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.613704715439018e-05,
+      "loss": 2.8473,
+      "step": 49310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.612800889385729e-05,
+      "loss": 2.9379,
+      "step": 49315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6118971124379766e-05,
+      "loss": 2.8964,
+      "step": 49320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.61099338462775e-05,
+      "loss": 2.9852,
+      "step": 49325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.610089705987041e-05,
+      "loss": 2.9906,
+      "step": 49330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.60918607654784e-05,
+      "loss": 2.9417,
+      "step": 49335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6082824963421344e-05,
+      "loss": 2.854,
+      "step": 49340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.60737896540191e-05,
+      "loss": 2.9319,
+      "step": 49345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.60647548375915e-05,
+      "loss": 2.8872,
+      "step": 49350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.605572051445837e-05,
+      "loss": 3.0319,
+      "step": 49355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6046686684939535e-05,
+      "loss": 2.8669,
+      "step": 49360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6037653349354755e-05,
+      "loss": 2.7754,
+      "step": 49365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.602862050802382e-05,
+      "loss": 2.9358,
+      "step": 49370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.60195881612665e-05,
+      "loss": 2.9358,
+      "step": 49375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.60105563094025e-05,
+      "loss": 2.8608,
+      "step": 49380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6001524952751546e-05,
+      "loss": 2.8326,
+      "step": 49385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.599249409163337e-05,
+      "loss": 2.929,
+      "step": 49390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5983463726367606e-05,
+      "loss": 2.901,
+      "step": 49395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.597443385727396e-05,
+      "loss": 2.916,
+      "step": 49400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.596540448467205e-05,
+      "loss": 2.8717,
+      "step": 49405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5956375608881535e-05,
+      "loss": 2.9276,
+      "step": 49410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.594734723022202e-05,
+      "loss": 2.9518,
+      "step": 49415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5938319349013095e-05,
+      "loss": 2.8386,
+      "step": 49420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.592929196557435e-05,
+      "loss": 2.9316,
+      "step": 49425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.592026508022532e-05,
+      "loss": 2.9839,
+      "step": 49430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.591123869328558e-05,
+      "loss": 2.8599,
+      "step": 49435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5902212805074644e-05,
+      "loss": 2.9559,
+      "step": 49440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5893187415911996e-05,
+      "loss": 2.8878,
+      "step": 49445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.588416252611717e-05,
+      "loss": 2.9037,
+      "step": 49450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5875138136009614e-05,
+      "loss": 2.8806,
+      "step": 49455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5866114245908766e-05,
+      "loss": 2.8739,
+      "step": 49460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.585709085613408e-05,
+      "loss": 2.8636,
+      "step": 49465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5848067967005e-05,
+      "loss": 2.9325,
+      "step": 49470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.583904557884089e-05,
+      "loss": 2.8228,
+      "step": 49475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5830023691961154e-05,
+      "loss": 2.7728,
+      "step": 49480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.582100230668517e-05,
+      "loss": 2.8455,
+      "step": 49485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.581198142333225e-05,
+      "loss": 2.951,
+      "step": 49490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5802961042221756e-05,
+      "loss": 3.027,
+      "step": 49495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5793941163672985e-05,
+      "loss": 2.9868,
+      "step": 49500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5784921788005235e-05,
+      "loss": 2.88,
+      "step": 49505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.577590291553778e-05,
+      "loss": 2.9179,
+      "step": 49510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5766884546589894e-05,
+      "loss": 2.775,
+      "step": 49515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.575786668148081e-05,
+      "loss": 2.9502,
+      "step": 49520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.574884932052974e-05,
+      "loss": 2.9972,
+      "step": 49525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5739832464055925e-05,
+      "loss": 2.9152,
+      "step": 49530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.573081611237851e-05,
+      "loss": 2.864,
+      "step": 49535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.572180026581668e-05,
+      "loss": 2.7963,
+      "step": 49540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.571278492468961e-05,
+      "loss": 2.9011,
+      "step": 49545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.57037700893164e-05,
+      "loss": 2.8435,
+      "step": 49550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5694755760016195e-05,
+      "loss": 2.8197,
+      "step": 49555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.568574193710807e-05,
+      "loss": 2.9483,
+      "step": 49560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.567672862091111e-05,
+      "loss": 2.8984,
+      "step": 49565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.566771581174441e-05,
+      "loss": 2.8521,
+      "step": 49570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.565870350992698e-05,
+      "loss": 2.8569,
+      "step": 49575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.564969171577786e-05,
+      "loss": 2.9021,
+      "step": 49580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.564068042961605e-05,
+      "loss": 2.8138,
+      "step": 49585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.563166965176056e-05,
+      "loss": 2.8816,
+      "step": 49590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5622659382530347e-05,
+      "loss": 2.7989,
+      "step": 49595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.561364962224437e-05,
+      "loss": 2.8261,
+      "step": 49600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.560464037122155e-05,
+      "loss": 2.9039,
+      "step": 49605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5595631629780834e-05,
+      "loss": 2.9047,
+      "step": 49610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.558662339824112e-05,
+      "loss": 2.8932,
+      "step": 49615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.557761567692127e-05,
+      "loss": 2.9595,
+      "step": 49620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5568608466140166e-05,
+      "loss": 2.8806,
+      "step": 49625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5559601766216646e-05,
+      "loss": 2.8761,
+      "step": 49630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.555059557746954e-05,
+      "loss": 2.9075,
+      "step": 49635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.554158990021765e-05,
+      "loss": 2.8564,
+      "step": 49640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.553258473477979e-05,
+      "loss": 2.9474,
+      "step": 49645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.552358008147473e-05,
+      "loss": 2.8259,
+      "step": 49650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.551457594062121e-05,
+      "loss": 2.8283,
+      "step": 49655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5505572312537984e-05,
+      "loss": 2.8201,
+      "step": 49660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5496569197543726e-05,
+      "loss": 2.8528,
+      "step": 49665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.548756659595722e-05,
+      "loss": 2.9518,
+      "step": 49670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.547856450809709e-05,
+      "loss": 2.8184,
+      "step": 49675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.546956293428202e-05,
+      "loss": 2.909,
+      "step": 49680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5460561874830655e-05,
+      "loss": 2.936,
+      "step": 49685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5451561330061626e-05,
+      "loss": 2.8514,
+      "step": 49690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.544256130029355e-05,
+      "loss": 3.0247,
+      "step": 49695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5433561785844994e-05,
+      "loss": 2.9352,
+      "step": 49700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5424562787034563e-05,
+      "loss": 2.9041,
+      "step": 49705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5415564304180795e-05,
+      "loss": 2.9124,
+      "step": 49710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.540656633760223e-05,
+      "loss": 2.8741,
+      "step": 49715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.539756888761739e-05,
+      "loss": 2.9083,
+      "step": 49720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5388571954544775e-05,
+      "loss": 2.8459,
+      "step": 49725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.537957553870288e-05,
+      "loss": 2.9388,
+      "step": 49730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.537057964041014e-05,
+      "loss": 2.9441,
+      "step": 49735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.536158425998503e-05,
+      "loss": 2.9542,
+      "step": 49740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5352589397745965e-05,
+      "loss": 2.9403,
+      "step": 49745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.534359505401136e-05,
+      "loss": 2.8725,
+      "step": 49750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5334601229099605e-05,
+      "loss": 2.8109,
+      "step": 49755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.532560792332905e-05,
+      "loss": 2.8542,
+      "step": 49760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5316615137018083e-05,
+      "loss": 2.9148,
+      "step": 49765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5307622870485034e-05,
+      "loss": 2.8948,
+      "step": 49770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.529863112404822e-05,
+      "loss": 2.8531,
+      "step": 49775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5289639898025925e-05,
+      "loss": 2.9092,
+      "step": 49780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.528064919273645e-05,
+      "loss": 2.9154,
+      "step": 49785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5271659008498035e-05,
+      "loss": 2.9335,
+      "step": 49790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5262669345628933e-05,
+      "loss": 2.8888,
+      "step": 49795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.525368020444739e-05,
+      "loss": 2.9249,
+      "step": 49800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5244691585271584e-05,
+      "loss": 2.8588,
+      "step": 49805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.523570348841972e-05,
+      "loss": 2.982,
+      "step": 49810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5226715914209965e-05,
+      "loss": 2.9228,
+      "step": 49815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.521772886296046e-05,
+      "loss": 2.8323,
+      "step": 49820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.520874233498936e-05,
+      "loss": 2.9869,
+      "step": 49825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.519975633061476e-05,
+      "loss": 2.8121,
+      "step": 49830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.519077085015477e-05,
+      "loss": 2.9474,
+      "step": 49835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.518178589392745e-05,
+      "loss": 2.8658,
+      "step": 49840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.517280146225087e-05,
+      "loss": 2.9511,
+      "step": 49845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.516381755544308e-05,
+      "loss": 2.8982,
+      "step": 49850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.515483417382208e-05,
+      "loss": 2.9254,
+      "step": 49855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.514585131770587e-05,
+      "loss": 2.866,
+      "step": 49860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.513686898741246e-05,
+      "loss": 2.9062,
+      "step": 49865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.512788718325983e-05,
+      "loss": 2.7968,
+      "step": 49870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.511890590556588e-05,
+      "loss": 2.8616,
+      "step": 49875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.510992515464857e-05,
+      "loss": 2.8244,
+      "step": 49880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.51009449308258e-05,
+      "loss": 2.9386,
+      "step": 49885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.509196523441547e-05,
+      "loss": 2.9487,
+      "step": 49890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.508298606573543e-05,
+      "loss": 2.9378,
+      "step": 49895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.507400742510356e-05,
+      "loss": 2.7278,
+      "step": 49900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5065029312837684e-05,
+      "loss": 2.8715,
+      "step": 49905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5056051729255615e-05,
+      "loss": 2.9553,
+      "step": 49910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.504707467467516e-05,
+      "loss": 2.8923,
+      "step": 49915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.503809814941409e-05,
+      "loss": 2.964,
+      "step": 49920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5029122153790175e-05,
+      "loss": 2.9258,
+      "step": 49925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.502014668812115e-05,
+      "loss": 2.9911,
+      "step": 49930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.501117175272473e-05,
+      "loss": 3.001,
+      "step": 49935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.500219734791864e-05,
+      "loss": 2.8578,
+      "step": 49940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.499322347402055e-05,
+      "loss": 2.9161,
+      "step": 49945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4984250131348125e-05,
+      "loss": 2.9797,
+      "step": 49950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4975277320219024e-05,
+      "loss": 2.9108,
+      "step": 49955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4966305040950856e-05,
+      "loss": 2.9012,
+      "step": 49960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.495733329386126e-05,
+      "loss": 2.7867,
+      "step": 49965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4948362079267805e-05,
+      "loss": 2.9478,
+      "step": 49970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.493939139748808e-05,
+      "loss": 2.8494,
+      "step": 49975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4930421248839616e-05,
+      "loss": 2.8634,
+      "step": 49980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4921451633639985e-05,
+      "loss": 2.9413,
+      "step": 49985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.491248255220666e-05,
+      "loss": 2.91,
+      "step": 49990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4903514004857165e-05,
+      "loss": 2.7384,
+      "step": 49995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4894545991908974e-05,
+      "loss": 2.957,
+      "step": 50000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 83588,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 6.7306902257664e+18,
+  "trial_name": null,
+  "trial_params": null
+}