diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,57619 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.6229377354266636,
+  "eval_steps": 500,
+  "global_step": 48000,
+  "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": 10528.3602,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 6687.8523,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 4057.8527,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 2440.0055,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 918.6812,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 438.8141,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 218.0395,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 117.5601,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 65.6845,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 41.1982,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 26.9585,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 23.7878,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 11.6092,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-05,
+      "loss": 9.5522,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 9.2246,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 19.3979,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.5e-05,
+      "loss": 20.5651,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-05,
+      "loss": 11.3577,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-05,
+      "loss": 10.4228,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 8.7842,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99999989583607e-05,
+      "loss": 8.8357,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99999958334428e-05,
+      "loss": 24.8971,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999062524643e-05,
+      "loss": 15.4338,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999998333377183e-05,
+      "loss": 10.207,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99999739590193e-05,
+      "loss": 8.7549,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999996250098922e-05,
+      "loss": 8.6327,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999994895968208e-05,
+      "loss": 69.2748,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999993333509843e-05,
+      "loss": 35.2525,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999991562723893e-05,
+      "loss": 20.4729,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999989583610432e-05,
+      "loss": 15.1346,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999987396169542e-05,
+      "loss": 10.4522,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999985000401313e-05,
+      "loss": 8.9556,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999982396305847e-05,
+      "loss": 11.546,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999979583883253e-05,
+      "loss": 11.8412,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999976563133644e-05,
+      "loss": 8.7958,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999973334057148e-05,
+      "loss": 9.8873,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999969896653903e-05,
+      "loss": 9.4615,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999966250924046e-05,
+      "loss": 8.4886,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999962396867735e-05,
+      "loss": 8.4348,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999958334485127e-05,
+      "loss": 8.2923,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999954063776392e-05,
+      "loss": 11.411,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999949584741708e-05,
+      "loss": 8.8399,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999944897381261e-05,
+      "loss": 8.2989,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999940001695247e-05,
+      "loss": 11.1191,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999934897683871e-05,
+      "loss": 8.6937,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999929585347342e-05,
+      "loss": 8.4847,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999924064685887e-05,
+      "loss": 8.1527,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999918335699731e-05,
+      "loss": 9.7911,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999912398389115e-05,
+      "loss": 8.2426,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999906252754288e-05,
+      "loss": 8.1047,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999899898795501e-05,
+      "loss": 8.0198,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999893336513022e-05,
+      "loss": 8.0427,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999886565907127e-05,
+      "loss": 8.0683,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999879586978093e-05,
+      "loss": 8.0244,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999872399726214e-05,
+      "loss": 7.8759,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999865004151787e-05,
+      "loss": 7.9009,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999857400255123e-05,
+      "loss": 7.9521,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999849588036539e-05,
+      "loss": 8.1759,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999841567496355e-05,
+      "loss": 7.8597,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999833338634911e-05,
+      "loss": 7.95,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99982490145255e-05,
+      "loss": 7.7471,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99981625594962e-05,
+      "loss": 7.6981,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999807402126484e-05,
+      "loss": 7.735,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999798339983509e-05,
+      "loss": 7.7098,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999789069521073e-05,
+      "loss": 7.8198,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999779590739562e-05,
+      "loss": 7.6386,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999769903639372e-05,
+      "loss": 7.5256,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999760008220906e-05,
+      "loss": 8.2843,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999749904484577e-05,
+      "loss": 7.5201,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999739592430805e-05,
+      "loss": 7.3527,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99972907206002e-05,
+      "loss": 7.5818,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999718343372662e-05,
+      "loss": 7.4863,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999707406369174e-05,
+      "loss": 7.3259,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999696261050015e-05,
+      "loss": 7.3973,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999684907415648e-05,
+      "loss": 7.2858,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999673345466547e-05,
+      "loss": 7.4155,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999661575203195e-05,
+      "loss": 7.3212,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999649596626077e-05,
+      "loss": 7.3454,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999637409735698e-05,
+      "loss": 7.316,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999625014532563e-05,
+      "loss": 7.4996,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99961241101719e-05,
+      "loss": 7.1404,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999599599190102e-05,
+      "loss": 7.1452,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999586579051835e-05,
+      "loss": 7.1595,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999573350602929e-05,
+      "loss": 7.095,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999559913843939e-05,
+      "loss": 7.1892,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999546268775419e-05,
+      "loss": 7.1067,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999532415397944e-05,
+      "loss": 7.0592,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999518353712088e-05,
+      "loss": 6.9847,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999504083718435e-05,
+      "loss": 6.8458,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999489605417582e-05,
+      "loss": 6.8357,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999474918810135e-05,
+      "loss": 6.8389,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9994600238967e-05,
+      "loss": 6.888,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999444920677901e-05,
+      "loss": 6.6489,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999429609154366e-05,
+      "loss": 6.9422,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999414089326735e-05,
+      "loss": 6.7878,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999398361195652e-05,
+      "loss": 6.6583,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999382424761774e-05,
+      "loss": 6.7453,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999366280025766e-05,
+      "loss": 6.6946,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999349926988297e-05,
+      "loss": 6.6918,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999333365650053e-05,
+      "loss": 6.672,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99931659601172e-05,
+      "loss": 6.3523,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999299618074e-05,
+      "loss": 6.5328,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999282431837598e-05,
+      "loss": 6.3866,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999265037303231e-05,
+      "loss": 6.3992,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999247434471623e-05,
+      "loss": 6.3259,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99922962334351e-05,
+      "loss": 6.2238,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999211603919629e-05,
+      "loss": 6.2489,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999193376200737e-05,
+      "loss": 6.1845,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99917494018759e-05,
+      "loss": 6.4484,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999156295880956e-05,
+      "loss": 6.2503,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999137443281612e-05,
+      "loss": 6.1151,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999118382390344e-05,
+      "loss": 6.1102,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999099113207945e-05,
+      "loss": 6.024,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999079635735222e-05,
+      "loss": 6.0134,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999059949972982e-05,
+      "loss": 6.1742,
+      "step": 575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999040055922046e-05,
+      "loss": 6.1613,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999019953583245e-05,
+      "loss": 5.9236,
+      "step": 585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998999642957414e-05,
+      "loss": 5.9576,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998979124045401e-05,
+      "loss": 5.9045,
+      "step": 595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998958396848061e-05,
+      "loss": 5.8175,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998937461366255e-05,
+      "loss": 5.767,
+      "step": 605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998916317600859e-05,
+      "loss": 5.7914,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998894965552752e-05,
+      "loss": 5.8786,
+      "step": 615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998873405222824e-05,
+      "loss": 5.6447,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998851636611975e-05,
+      "loss": 5.7074,
+      "step": 625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998829659721107e-05,
+      "loss": 5.6804,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998807474551141e-05,
+      "loss": 5.5762,
+      "step": 635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998785081102998e-05,
+      "loss": 5.543,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998762479377613e-05,
+      "loss": 5.9243,
+      "step": 645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998739669375927e-05,
+      "loss": 5.6351,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998716651098889e-05,
+      "loss": 5.7794,
+      "step": 655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998693424547462e-05,
+      "loss": 5.6046,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99866998972261e-05,
+      "loss": 5.7066,
+      "step": 665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998646346625312e-05,
+      "loss": 5.5379,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998622495256549e-05,
+      "loss": 5.2487,
+      "step": 675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998598435617319e-05,
+      "loss": 5.2542,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998574167708623e-05,
+      "loss": 5.4895,
+      "step": 685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998549691531472e-05,
+      "loss": 5.363,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998525007086886e-05,
+      "loss": 5.3214,
+      "step": 695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998500114375894e-05,
+      "loss": 5.3859,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998475013399532e-05,
+      "loss": 5.4646,
+      "step": 705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998449704158847e-05,
+      "loss": 5.3215,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998424186654892e-05,
+      "loss": 5.3789,
+      "step": 715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998398460888731e-05,
+      "loss": 5.2912,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998372526861438e-05,
+      "loss": 5.3282,
+      "step": 725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99834638457409e-05,
+      "loss": 5.2784,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99832003402778e-05,
+      "loss": 5.2221,
+      "step": 735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998293475223602e-05,
+      "loss": 5.3546,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998266708162662e-05,
+      "loss": 5.1042,
+      "step": 745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998239732846082e-05,
+      "loss": 5.3971,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998212549274978e-05,
+      "loss": 5.0517,
+      "step": 755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998185157450489e-05,
+      "loss": 5.0409,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998157557373752e-05,
+      "loss": 5.1364,
+      "step": 765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998129749045919e-05,
+      "loss": 5.0723,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998101732468147e-05,
+      "loss": 5.2825,
+      "step": 775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998073507641605e-05,
+      "loss": 5.0896,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998045074567469e-05,
+      "loss": 5.108,
+      "step": 785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998016433246922e-05,
+      "loss": 5.2063,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99798758368116e-05,
+      "loss": 5.1849,
+      "step": 795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997958525871382e-05,
+      "loss": 5.0194,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997929259818801e-05,
+      "loss": 4.9961,
+      "step": 805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997899785524636e-05,
+      "loss": 5.0527,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997870102990115e-05,
+      "loss": 5.0261,
+      "step": 815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997840212216473e-05,
+      "loss": 4.8737,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997810113204958e-05,
+      "loss": 5.2025,
+      "step": 825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997779805956821e-05,
+      "loss": 4.9258,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997749290473329e-05,
+      "loss": 4.9611,
+      "step": 835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99771856675575e-05,
+      "loss": 4.9593,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997687634805365e-05,
+      "loss": 5.0952,
+      "step": 845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997656494623464e-05,
+      "loss": 4.8342,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997625146211343e-05,
+      "loss": 4.9877,
+      "step": 855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997593589570308e-05,
+      "loss": 5.0479,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997561824701675e-05,
+      "loss": 4.9491,
+      "step": 865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997529851606768e-05,
+      "loss": 4.8176,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997497670286916e-05,
+      "loss": 4.8742,
+      "step": 875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997465280743463e-05,
+      "loss": 4.6354,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997432682977759e-05,
+      "loss": 4.9222,
+      "step": 885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997399876991158e-05,
+      "loss": 4.7676,
+      "step": 890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997366862785031e-05,
+      "loss": 4.7997,
+      "step": 895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997333640360753e-05,
+      "loss": 4.9123,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997300209719706e-05,
+      "loss": 4.5341,
+      "step": 905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997266570863286e-05,
+      "loss": 4.8668,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997232723792892e-05,
+      "loss": 4.8284,
+      "step": 915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997198668509936e-05,
+      "loss": 4.8636,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997164405015835e-05,
+      "loss": 4.6223,
+      "step": 925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997129933312018e-05,
+      "loss": 4.6779,
+      "step": 930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99709525339992e-05,
+      "loss": 4.8246,
+      "step": 935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997060365280987e-05,
+      "loss": 5.0387,
+      "step": 940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997025268956675e-05,
+      "loss": 4.8227,
+      "step": 945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996989964428442e-05,
+      "loss": 4.8571,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996954451697761e-05,
+      "loss": 4.7381,
+      "step": 955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996918730766111e-05,
+      "loss": 4.6939,
+      "step": 960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996882801634984e-05,
+      "loss": 4.8104,
+      "step": 965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99684666430587e-05,
+      "loss": 4.8539,
+      "step": 970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996810318780281e-05,
+      "loss": 4.7034,
+      "step": 975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996773765059728e-05,
+      "loss": 4.6509,
+      "step": 980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996737003145734e-05,
+      "loss": 4.6037,
+      "step": 985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996700033039833e-05,
+      "loss": 4.7647,
+      "step": 990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996662854743564e-05,
+      "loss": 4.6092,
+      "step": 995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996625468258475e-05,
+      "loss": 4.4709,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996587873586125e-05,
+      "loss": 4.3718,
+      "step": 1005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996550070728081e-05,
+      "loss": 4.5629,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996512059685917e-05,
+      "loss": 4.5885,
+      "step": 1015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996473840461217e-05,
+      "loss": 4.507,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996435413055574e-05,
+      "loss": 4.5448,
+      "step": 1025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996396777470589e-05,
+      "loss": 4.3898,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99635793370787e-05,
+      "loss": 4.7504,
+      "step": 1035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996318881769038e-05,
+      "loss": 4.608,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996279621655718e-05,
+      "loss": 4.6096,
+      "step": 1045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996240153369547e-05,
+      "loss": 4.6146,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996200476912169e-05,
+      "loss": 4.4774,
+      "step": 1055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996160592285238e-05,
+      "loss": 4.6986,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996120499490414e-05,
+      "loss": 4.5067,
+      "step": 1065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99608019852937e-05,
+      "loss": 4.4878,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996039689403782e-05,
+      "loss": 4.562,
+      "step": 1075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99599897211534e-05,
+      "loss": 4.5123,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995958046665742e-05,
+      "loss": 4.5702,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995916913056689e-05,
+      "loss": 4.4982,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995875571289899e-05,
+      "loss": 4.5956,
+      "step": 1095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995834021367092e-05,
+      "loss": 4.4616,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995792263289999e-05,
+      "loss": 4.44,
+      "step": 1105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995750297060362e-05,
+      "loss": 4.516,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995708122679927e-05,
+      "loss": 4.5102,
+      "step": 1115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995665740150455e-05,
+      "loss": 4.5239,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995623149473708e-05,
+      "loss": 4.526,
+      "step": 1125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995580350651461e-05,
+      "loss": 4.4464,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9955373436855e-05,
+      "loss": 4.3762,
+      "step": 1135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995494128577614e-05,
+      "loss": 4.5048,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995450705329605e-05,
+      "loss": 4.438,
+      "step": 1145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995407073943281e-05,
+      "loss": 4.4053,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995363234420462e-05,
+      "loss": 4.403,
+      "step": 1155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995319186762975e-05,
+      "loss": 4.4729,
+      "step": 1160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995274930972651e-05,
+      "loss": 4.306,
+      "step": 1165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995230467051338e-05,
+      "loss": 4.2232,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995185795000888e-05,
+      "loss": 4.4612,
+      "step": 1175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995140914823162e-05,
+      "loss": 4.4677,
+      "step": 1180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995095826520028e-05,
+      "loss": 4.5076,
+      "step": 1185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995050530093367e-05,
+      "loss": 4.3424,
+      "step": 1190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995005025545064e-05,
+      "loss": 4.4137,
+      "step": 1195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99495931287702e-05,
+      "loss": 4.5057,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994913392091132e-05,
+      "loss": 4.5174,
+      "step": 1205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99486726318932e-05,
+      "loss": 4.4418,
+      "step": 1210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994820926173503e-05,
+      "loss": 4.4457,
+      "step": 1215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994774381045611e-05,
+      "loss": 4.3618,
+      "step": 1220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994727627807585e-05,
+      "loss": 4.3269,
+      "step": 1225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994680666461374e-05,
+      "loss": 4.2469,
+      "step": 1230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994633497008931e-05,
+      "loss": 4.2927,
+      "step": 1235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994586119452225e-05,
+      "loss": 4.378,
+      "step": 1240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994538533793228e-05,
+      "loss": 4.4607,
+      "step": 1245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994490740033921e-05,
+      "loss": 4.2645,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9944427381763e-05,
+      "loss": 4.3857,
+      "step": 1255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994394528222363e-05,
+      "loss": 4.3105,
+      "step": 1260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994346110174117e-05,
+      "loss": 4.2787,
+      "step": 1265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99429748403358e-05,
+      "loss": 4.2893,
+      "step": 1270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99424864980278e-05,
+      "loss": 4.4233,
+      "step": 1275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994199607483749e-05,
+      "loss": 4.3662,
+      "step": 1280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994150357078533e-05,
+      "loss": 4.3181,
+      "step": 1285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994100898589182e-05,
+      "loss": 4.2899,
+      "step": 1290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994051232017757e-05,
+      "loss": 4.4271,
+      "step": 1295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994001357366328e-05,
+      "loss": 4.3051,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993951274636973e-05,
+      "loss": 4.3965,
+      "step": 1305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993900983831779e-05,
+      "loss": 4.1912,
+      "step": 1310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993850484952841e-05,
+      "loss": 4.3821,
+      "step": 1315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993799778002263e-05,
+      "loss": 4.3911,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993748862982157e-05,
+      "loss": 4.3327,
+      "step": 1325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993697739894647e-05,
+      "loss": 4.3175,
+      "step": 1330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99364640874186e-05,
+      "loss": 4.2368,
+      "step": 1335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993594869525938e-05,
+      "loss": 4.3687,
+      "step": 1340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993543122249022e-05,
+      "loss": 4.2933,
+      "step": 1345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993491166913277e-05,
+      "loss": 4.2921,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993439003520861e-05,
+      "loss": 4.2929,
+      "step": 1355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993386632073952e-05,
+      "loss": 4.3897,
+      "step": 1360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993334052574727e-05,
+      "loss": 4.1101,
+      "step": 1365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99328126502538e-05,
+      "loss": 4.3542,
+      "step": 1370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993228269428112e-05,
+      "loss": 4.2218,
+      "step": 1375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993175065785127e-05,
+      "loss": 4.2695,
+      "step": 1380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993121654098644e-05,
+      "loss": 4.2662,
+      "step": 1385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993068034370889e-05,
+      "loss": 4.4571,
+      "step": 1390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993014206604094e-05,
+      "loss": 4.3126,
+      "step": 1395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992960170800505e-05,
+      "loss": 4.2749,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99290592696237e-05,
+      "loss": 4.2117,
+      "step": 1405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992851475091951e-05,
+      "loss": 4.3038,
+      "step": 1410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992796815191516e-05,
+      "loss": 4.3913,
+      "step": 1415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992741947263345e-05,
+      "loss": 4.385,
+      "step": 1420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99268687130972e-05,
+      "loss": 4.3114,
+      "step": 1425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99263158733294e-05,
+      "loss": 4.3224,
+      "step": 1430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992576095335303e-05,
+      "loss": 4.3217,
+      "step": 1435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992520395319125e-05,
+      "loss": 4.214,
+      "step": 1440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992464487286726e-05,
+      "loss": 4.2451,
+      "step": 1445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992408371240437e-05,
+      "loss": 4.2021,
+      "step": 1450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992352047182594e-05,
+      "loss": 4.2699,
+      "step": 1455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992295515115544e-05,
+      "loss": 4.0462,
+      "step": 1460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992238775041642e-05,
+      "loss": 4.1375,
+      "step": 1465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992181826963253e-05,
+      "loss": 4.3295,
+      "step": 1470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992124670882751e-05,
+      "loss": 4.2765,
+      "step": 1475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992067306802516e-05,
+      "loss": 4.0998,
+      "step": 1480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992009734724938e-05,
+      "loss": 4.3352,
+      "step": 1485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991951954652416e-05,
+      "loss": 4.1988,
+      "step": 1490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991893966587357e-05,
+      "loss": 4.0411,
+      "step": 1495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991835770532177e-05,
+      "loss": 4.2398,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991777366489303e-05,
+      "loss": 4.0982,
+      "step": 1505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991718754461166e-05,
+      "loss": 4.0629,
+      "step": 1510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991659934450208e-05,
+      "loss": 4.2935,
+      "step": 1515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99160090645888e-05,
+      "loss": 4.2682,
+      "step": 1520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991541670489644e-05,
+      "loss": 4.2641,
+      "step": 1525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991482226544966e-05,
+      "loss": 4.2645,
+      "step": 1530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991422574627323e-05,
+      "loss": 4.191,
+      "step": 1535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991362714739199e-05,
+      "loss": 4.0572,
+      "step": 1540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99130264688309e-05,
+      "loss": 4.1518,
+      "step": 1545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991242371061498e-05,
+      "loss": 3.965,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991181887276934e-05,
+      "loss": 4.1602,
+      "step": 1555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991121195531921e-05,
+      "loss": 4.0165,
+      "step": 1560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991060295828985e-05,
+      "loss": 4.4698,
+      "step": 1565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990999188170662e-05,
+      "loss": 4.1153,
+      "step": 1570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990937872559503e-05,
+      "loss": 4.1537,
+      "step": 1575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990876348998057e-05,
+      "loss": 4.0296,
+      "step": 1580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99081461748889e-05,
+      "loss": 4.0899,
+      "step": 1585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990752678034576e-05,
+      "loss": 4.4666,
+      "step": 1590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990690530637694e-05,
+      "loss": 4.0777,
+      "step": 1595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990628175300832e-05,
+      "loss": 4.1831,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99056561202659e-05,
+      "loss": 4.2038,
+      "step": 1605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990502840817575e-05,
+      "loss": 4.0024,
+      "step": 1610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990439861676401e-05,
+      "loss": 4.1512,
+      "step": 1615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990376674605693e-05,
+      "loss": 4.1716,
+      "step": 1620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990313279608084e-05,
+      "loss": 4.0762,
+      "step": 1625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990249676686212e-05,
+      "loss": 4.0593,
+      "step": 1630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990185865842731e-05,
+      "loss": 4.167,
+      "step": 1635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9901218470803e-05,
+      "loss": 4.1156,
+      "step": 1640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990057620401585e-05,
+      "loss": 4.1577,
+      "step": 1645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989993185809262e-05,
+      "loss": 4.1516,
+      "step": 1650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989928543306014e-05,
+      "loss": 3.9278,
+      "step": 1655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989863692894536e-05,
+      "loss": 4.1181,
+      "step": 1660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98979863457753e-05,
+      "loss": 3.9141,
+      "step": 1665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989733368357708e-05,
+      "loss": 4.0798,
+      "step": 1670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989667894237788e-05,
+      "loss": 4.2422,
+      "step": 1675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989602212220497e-05,
+      "loss": 4.0751,
+      "step": 1680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989536322308574e-05,
+      "loss": 4.1773,
+      "step": 1685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989470224504762e-05,
+      "loss": 4.1891,
+      "step": 1690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989403918811817e-05,
+      "loss": 4.1227,
+      "step": 1695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989337405232502e-05,
+      "loss": 4.2275,
+      "step": 1700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989270683769585e-05,
+      "loss": 4.5025,
+      "step": 1705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989203754425849e-05,
+      "loss": 4.1113,
+      "step": 1710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989136617204082e-05,
+      "loss": 4.0134,
+      "step": 1715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989069272107081e-05,
+      "loss": 4.0166,
+      "step": 1720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989001719137652e-05,
+      "loss": 4.0493,
+      "step": 1725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98893395829861e-05,
+      "loss": 3.9417,
+      "step": 1730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988865989592777e-05,
+      "loss": 4.0295,
+      "step": 1735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988797813022985e-05,
+      "loss": 4.1445,
+      "step": 1740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988729428592079e-05,
+      "loss": 3.9497,
+      "step": 1745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9886608363029e-05,
+      "loss": 3.934,
+      "step": 1750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988592036158315e-05,
+      "loss": 3.9706,
+      "step": 1755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988523028161185e-05,
+      "loss": 4.1756,
+      "step": 1760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988453812314387e-05,
+      "loss": 4.13,
+      "step": 1765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988384388620804e-05,
+      "loss": 4.1659,
+      "step": 1770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988314757083329e-05,
+      "loss": 3.9587,
+      "step": 1775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988244917704863e-05,
+      "loss": 3.9467,
+      "step": 1780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988174870488318e-05,
+      "loss": 3.9251,
+      "step": 1785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988104615436608e-05,
+      "loss": 3.9825,
+      "step": 1790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988034152552665e-05,
+      "loss": 4.1152,
+      "step": 1795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987963481839425e-05,
+      "loss": 4.1015,
+      "step": 1800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987892603299827e-05,
+      "loss": 4.0109,
+      "step": 1805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987821516936828e-05,
+      "loss": 4.0797,
+      "step": 1810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987750222753391e-05,
+      "loss": 3.8929,
+      "step": 1815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987678720752484e-05,
+      "loss": 4.0281,
+      "step": 1820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987607010937088e-05,
+      "loss": 4.0368,
+      "step": 1825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987535093310189e-05,
+      "loss": 3.9764,
+      "step": 1830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987462967874785e-05,
+      "loss": 3.9384,
+      "step": 1835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98739063463388e-05,
+      "loss": 4.0935,
+      "step": 1840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987318093590489e-05,
+      "loss": 4.0399,
+      "step": 1845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987245344747633e-05,
+      "loss": 3.9343,
+      "step": 1850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987172388108346e-05,
+      "loss": 4.1169,
+      "step": 1855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987099223675664e-05,
+      "loss": 4.1881,
+      "step": 1860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987025851452639e-05,
+      "loss": 3.9644,
+      "step": 1865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986952271442325e-05,
+      "loss": 3.9282,
+      "step": 1870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986878483647788e-05,
+      "loss": 4.0864,
+      "step": 1875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986804488072106e-05,
+      "loss": 4.0871,
+      "step": 1880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986730284718358e-05,
+      "loss": 4.0207,
+      "step": 1885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986655873589639e-05,
+      "loss": 4.0282,
+      "step": 1890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986581254689044e-05,
+      "loss": 4.1313,
+      "step": 1895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98650642801969e-05,
+      "loss": 3.9277,
+      "step": 1900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986431393584687e-05,
+      "loss": 4.0817,
+      "step": 1905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986356151387168e-05,
+      "loss": 3.9686,
+      "step": 1910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986280701430263e-05,
+      "loss": 3.9252,
+      "step": 1915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986205043717118e-05,
+      "loss": 4.2058,
+      "step": 1920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986129178250884e-05,
+      "loss": 3.8852,
+      "step": 1925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986053105034723e-05,
+      "loss": 3.882,
+      "step": 1930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985976824071804e-05,
+      "loss": 3.8405,
+      "step": 1935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985900335365307e-05,
+      "loss": 3.8391,
+      "step": 1940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985823638918417e-05,
+      "loss": 3.9754,
+      "step": 1945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98574673473433e-05,
+      "loss": 4.0084,
+      "step": 1950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98566962281625e-05,
+      "loss": 3.8991,
+      "step": 1955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98559230316739e-05,
+      "loss": 4.0776,
+      "step": 1960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985514775790973e-05,
+      "loss": 4.0117,
+      "step": 1965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985437040690229e-05,
+      "loss": 3.8785,
+      "step": 1970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985359097868395e-05,
+      "loss": 3.9752,
+      "step": 1975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98528094732872e-05,
+      "loss": 3.9614,
+      "step": 1980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985202589074459e-05,
+      "loss": 3.8681,
+      "step": 1985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985124023108878e-05,
+      "loss": 3.96,
+      "step": 1990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985045249435251e-05,
+      "loss": 3.5778,
+      "step": 1995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984966268056858e-05,
+      "loss": 3.9381,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984887078976992e-05,
+      "loss": 3.7649,
+      "step": 2005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984807682198951e-05,
+      "loss": 3.8185,
+      "step": 2010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984728077726043e-05,
+      "loss": 3.9538,
+      "step": 2015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984648265561587e-05,
+      "loss": 4.0259,
+      "step": 2020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984568245708905e-05,
+      "loss": 3.9496,
+      "step": 2025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984488018171334e-05,
+      "loss": 3.7867,
+      "step": 2030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984407582952216e-05,
+      "loss": 3.8714,
+      "step": 2035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984326940054901e-05,
+      "loss": 4.109,
+      "step": 2040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984246089482752e-05,
+      "loss": 3.969,
+      "step": 2045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984165031239133e-05,
+      "loss": 3.9571,
+      "step": 2050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984083765327425e-05,
+      "loss": 3.9045,
+      "step": 2055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984002291751014e-05,
+      "loss": 3.9319,
+      "step": 2060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983920610513294e-05,
+      "loss": 4.0575,
+      "step": 2065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983838721617666e-05,
+      "loss": 3.9408,
+      "step": 2070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983756625067544e-05,
+      "loss": 3.9511,
+      "step": 2075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98367432086635e-05,
+      "loss": 3.8585,
+      "step": 2080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983591809017511e-05,
+      "loss": 3.9117,
+      "step": 2085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983509089524467e-05,
+      "loss": 3.9002,
+      "step": 2090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983426162390661e-05,
+      "loss": 3.9478,
+      "step": 2095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983343027619551e-05,
+      "loss": 4.007,
+      "step": 2100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983259685214602e-05,
+      "loss": 4.0119,
+      "step": 2105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983176135179284e-05,
+      "loss": 3.8091,
+      "step": 2110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983092377517078e-05,
+      "loss": 3.855,
+      "step": 2115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983008412231476e-05,
+      "loss": 4.0124,
+      "step": 2120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982924239325974e-05,
+      "loss": 3.8999,
+      "step": 2125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982839858804082e-05,
+      "loss": 4.0585,
+      "step": 2130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982755270669314e-05,
+      "loss": 4.1439,
+      "step": 2135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982670474925193e-05,
+      "loss": 3.7766,
+      "step": 2140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982585471575255e-05,
+      "loss": 3.8133,
+      "step": 2145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98250026062304e-05,
+      "loss": 3.8988,
+      "step": 2150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982414842072098e-05,
+      "loss": 3.8527,
+      "step": 2155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98232921592599e-05,
+      "loss": 3.7237,
+      "step": 2160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982243382188282e-05,
+      "loss": 3.7703,
+      "step": 2165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98215734086255e-05,
+      "loss": 3.9084,
+      "step": 2170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982071091952379e-05,
+      "loss": 3.7053,
+      "step": 2175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981984635461365e-05,
+      "loss": 3.8052,
+      "step": 2180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981897971393106e-05,
+      "loss": 3.7833,
+      "step": 2185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981811099751217e-05,
+      "loss": 3.8606,
+      "step": 2190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981724020539316e-05,
+      "loss": 3.7918,
+      "step": 2195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981636733761032e-05,
+      "loss": 3.8142,
+      "step": 2200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98154923942e-05,
+      "loss": 4.0098,
+      "step": 2205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981461537519868e-05,
+      "loss": 3.818,
+      "step": 2210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981373628064286e-05,
+      "loss": 3.9823,
+      "step": 2215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981285511056921e-05,
+      "loss": 3.946,
+      "step": 2220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981197186501443e-05,
+      "loss": 3.8575,
+      "step": 2225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98110865440153e-05,
+      "loss": 3.7523,
+      "step": 2230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981019914760876e-05,
+      "loss": 3.9216,
+      "step": 2235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980930967583174e-05,
+      "loss": 4.0703,
+      "step": 2240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980841812872131e-05,
+      "loss": 3.8232,
+      "step": 2245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98075245063146e-05,
+      "loss": 3.8986,
+      "step": 2250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980662880864888e-05,
+      "loss": 3.7734,
+      "step": 2255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980573103576145e-05,
+      "loss": 3.8168,
+      "step": 2260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980483118768973e-05,
+      "loss": 3.8561,
+      "step": 2265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980392926447118e-05,
+      "loss": 3.6764,
+      "step": 2270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98030252661434e-05,
+      "loss": 3.8736,
+      "step": 2275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980211919274407e-05,
+      "loss": 3.7941,
+      "step": 2280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980121104431092e-05,
+      "loss": 3.8161,
+      "step": 2285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98003008208818e-05,
+      "loss": 3.8802,
+      "step": 2290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979938852249462e-05,
+      "loss": 3.6575,
+      "step": 2295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979847414918741e-05,
+      "loss": 3.8705,
+      "step": 2300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979755770099825e-05,
+      "loss": 3.8766,
+      "step": 2305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979663917796536e-05,
+      "loss": 3.8988,
+      "step": 2310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979571858012698e-05,
+      "loss": 3.9053,
+      "step": 2315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979479590752145e-05,
+      "loss": 3.8319,
+      "step": 2320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979387116018725e-05,
+      "loss": 3.7358,
+      "step": 2325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979294433816288e-05,
+      "loss": 3.8135,
+      "step": 2330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979201544148699e-05,
+      "loss": 3.6866,
+      "step": 2335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979108447019827e-05,
+      "loss": 3.7815,
+      "step": 2340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97901514243355e-05,
+      "loss": 3.7651,
+      "step": 2345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978921630393756e-05,
+      "loss": 3.7935,
+      "step": 2350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978827910904341e-05,
+      "loss": 3.701,
+      "step": 2355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978733983969211e-05,
+      "loss": 3.9672,
+      "step": 2360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978639849592277e-05,
+      "loss": 3.8439,
+      "step": 2365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978545507777465e-05,
+      "loss": 3.5983,
+      "step": 2370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978450958528703e-05,
+      "loss": 3.7194,
+      "step": 2375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978356201849932e-05,
+      "loss": 3.7725,
+      "step": 2380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978261237745098e-05,
+      "loss": 3.6088,
+      "step": 2385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97816606621816e-05,
+      "loss": 3.8247,
+      "step": 2390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978070687273081e-05,
+      "loss": 3.6705,
+      "step": 2395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977975100913836e-05,
+      "loss": 3.8596,
+      "step": 2400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977879307144411e-05,
+      "loss": 3.7436,
+      "step": 2405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977783305968791e-05,
+      "loss": 3.8896,
+      "step": 2410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977687097390982e-05,
+      "loss": 3.7224,
+      "step": 2415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977590681414988e-05,
+      "loss": 3.8332,
+      "step": 2420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97749405804483e-05,
+      "loss": 3.6033,
+      "step": 2425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97739722728453e-05,
+      "loss": 3.661,
+      "step": 2430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977300189138125e-05,
+      "loss": 3.5483,
+      "step": 2435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977202943609658e-05,
+      "loss": 3.6442,
+      "step": 2440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97710549070318e-05,
+      "loss": 3.8234,
+      "step": 2445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977007830422754e-05,
+      "loss": 3.7409,
+      "step": 2450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976909962772445e-05,
+      "loss": 3.8008,
+      "step": 2455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976811887756332e-05,
+      "loss": 3.6454,
+      "step": 2460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976713605378503e-05,
+      "loss": 3.6874,
+      "step": 2465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97661511564305e-05,
+      "loss": 3.8255,
+      "step": 2470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976516418554083e-05,
+      "loss": 3.8976,
+      "step": 2475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976417514115706e-05,
+      "loss": 3.8052,
+      "step": 2480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976318402332045e-05,
+      "loss": 3.7647,
+      "step": 2485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976219083207228e-05,
+      "loss": 3.7029,
+      "step": 2490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976119556745395e-05,
+      "loss": 3.8499,
+      "step": 2495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976019822950691e-05,
+      "loss": 3.6965,
+      "step": 2500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975919881827271e-05,
+      "loss": 3.6632,
+      "step": 2505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975819733379299e-05,
+      "loss": 3.6977,
+      "step": 2510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97571937761095e-05,
+      "loss": 3.792,
+      "step": 2515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975618814526404e-05,
+      "loss": 3.7527,
+      "step": 2520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97551804412985e-05,
+      "loss": 3.7873,
+      "step": 2525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97541706642549e-05,
+      "loss": 3.8223,
+      "step": 2530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975315881417528e-05,
+      "loss": 3.7001,
+      "step": 2535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97521448911018e-05,
+      "loss": 3.614,
+      "step": 2540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975112889507671e-05,
+      "loss": 3.804,
+      "step": 2545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975011082614236e-05,
+      "loss": 3.6633,
+      "step": 2550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974909068434116e-05,
+      "loss": 3.766,
+      "step": 2555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974806846971561e-05,
+      "loss": 3.7277,
+      "step": 2560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974704418230829e-05,
+      "loss": 3.7706,
+      "step": 2565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97460178221619e-05,
+      "loss": 3.9363,
+      "step": 2570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974498938931918e-05,
+      "loss": 3.7452,
+      "step": 2575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974395888382301e-05,
+      "loss": 3.8096,
+      "step": 2580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97429263057163e-05,
+      "loss": 3.5793,
+      "step": 2585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974189165504208e-05,
+      "loss": 3.7751,
+      "step": 2590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974085493184348e-05,
+      "loss": 3.4817,
+      "step": 2595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973981613616366e-05,
+      "loss": 3.8654,
+      "step": 2600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973877526804591e-05,
+      "loss": 3.7166,
+      "step": 2605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973773232753363e-05,
+      "loss": 3.589,
+      "step": 2610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973668731467025e-05,
+      "loss": 3.7605,
+      "step": 2615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97356402294993e-05,
+      "loss": 3.8379,
+      "step": 2620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973459107206444e-05,
+      "loss": 3.6393,
+      "step": 2625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973353984240936e-05,
+      "loss": 3.7516,
+      "step": 2630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973248654057785e-05,
+      "loss": 3.7324,
+      "step": 2635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973143116661383e-05,
+      "loss": 3.8134,
+      "step": 2640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973037372056125e-05,
+      "loss": 3.6336,
+      "step": 2645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972931420246417e-05,
+      "loss": 3.7175,
+      "step": 2650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972825261236674e-05,
+      "loss": 3.6829,
+      "step": 2655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972718895031319e-05,
+      "loss": 3.7483,
+      "step": 2660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972612321634784e-05,
+      "loss": 3.8346,
+      "step": 2665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97250554105151e-05,
+      "loss": 3.6759,
+      "step": 2670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972398553285944e-05,
+      "loss": 3.7844,
+      "step": 2675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972291358342545e-05,
+      "loss": 3.7251,
+      "step": 2680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972183956225781e-05,
+      "loss": 3.6734,
+      "step": 2685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972076346940125e-05,
+      "loss": 3.6143,
+      "step": 2690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97196853049006e-05,
+      "loss": 3.7423,
+      "step": 2695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971860506880079e-05,
+      "loss": 3.5916,
+      "step": 2700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971752276114685e-05,
+      "loss": 3.7098,
+      "step": 2705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971643838198385e-05,
+      "loss": 3.7401,
+      "step": 2710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971535193135697e-05,
+      "loss": 3.6206,
+      "step": 2715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97142634093115e-05,
+      "loss": 3.7094,
+      "step": 2720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971317281589277e-05,
+      "loss": 3.6989,
+      "step": 2725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971208015114621e-05,
+      "loss": 3.7372,
+      "step": 2730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971098541511739e-05,
+      "loss": 3.7684,
+      "step": 2735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970988860785188e-05,
+      "loss": 3.7175,
+      "step": 2740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970878972939542e-05,
+      "loss": 3.5757,
+      "step": 2745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970768877979375e-05,
+      "loss": 3.6578,
+      "step": 2750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970658575909277e-05,
+      "loss": 3.8053,
+      "step": 2755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970548066733845e-05,
+      "loss": 3.8084,
+      "step": 2760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97043735045768e-05,
+      "loss": 3.5823,
+      "step": 2765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970326427085397e-05,
+      "loss": 3.5621,
+      "step": 2770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970215296621618e-05,
+      "loss": 3.6453,
+      "step": 2775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970103959070973e-05,
+      "loss": 3.6324,
+      "step": 2780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969992414438098e-05,
+      "loss": 3.6421,
+      "step": 2785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969880662727647e-05,
+      "loss": 3.6979,
+      "step": 2790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96976870394427e-05,
+      "loss": 3.7717,
+      "step": 2795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969656538092635e-05,
+      "loss": 3.5921,
+      "step": 2800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969544165177415e-05,
+      "loss": 3.6523,
+      "step": 2805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969431585203292e-05,
+      "loss": 3.5973,
+      "step": 2810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969318798174956e-05,
+      "loss": 3.6409,
+      "step": 2815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969205804097106e-05,
+      "loss": 3.8719,
+      "step": 2820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969092602974452e-05,
+      "loss": 3.6199,
+      "step": 2825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96897919481171e-05,
+      "loss": 3.581,
+      "step": 2830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968865579613604e-05,
+      "loss": 3.498,
+      "step": 2835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968751757384867e-05,
+      "loss": 3.7733,
+      "step": 2840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968637728130245e-05,
+      "loss": 3.6642,
+      "step": 2845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968523491854487e-05,
+      "loss": 3.6408,
+      "step": 2850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96840904856235e-05,
+      "loss": 3.645,
+      "step": 2855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968294398258608e-05,
+      "loss": 3.5725,
+      "step": 2860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968179540948036e-05,
+      "loss": 3.499,
+      "step": 2865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968064476635415e-05,
+      "loss": 3.6428,
+      "step": 2870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967949205325545e-05,
+      "loss": 3.5377,
+      "step": 2875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967833727023227e-05,
+      "loss": 3.6677,
+      "step": 2880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967718041733271e-05,
+      "loss": 3.7848,
+      "step": 2885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9676021494605e-05,
+      "loss": 3.7136,
+      "step": 2890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96748605020974e-05,
+      "loss": 3.7574,
+      "step": 2895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967369743985829e-05,
+      "loss": 3.6463,
+      "step": 2900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967253230793614e-05,
+      "loss": 3.5646,
+      "step": 2905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967136510637947e-05,
+      "loss": 3.6975,
+      "step": 2910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967019583523696e-05,
+      "loss": 3.5592,
+      "step": 2915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966902449455729e-05,
+      "loss": 3.6276,
+      "step": 2920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966785108438926e-05,
+      "loss": 3.4425,
+      "step": 2925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966667560478179e-05,
+      "loss": 3.6963,
+      "step": 2930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966549805578383e-05,
+      "loss": 3.5922,
+      "step": 2935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966431843744448e-05,
+      "loss": 3.5844,
+      "step": 2940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966313674981283e-05,
+      "loss": 3.6152,
+      "step": 2945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966195299293816e-05,
+      "loss": 3.6211,
+      "step": 2950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96607671668698e-05,
+      "loss": 3.751,
+      "step": 2955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965957927165715e-05,
+      "loss": 3.6826,
+      "step": 2960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965838930734966e-05,
+      "loss": 3.6035,
+      "step": 2965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965719727399695e-05,
+      "loss": 3.6311,
+      "step": 2970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96560031716487e-05,
+      "loss": 3.4974,
+      "step": 2975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965480700035465e-05,
+      "loss": 3.589,
+      "step": 2980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965360876016461e-05,
+      "loss": 3.8072,
+      "step": 2985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965240845112855e-05,
+      "loss": 3.6455,
+      "step": 2990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965120607329646e-05,
+      "loss": 3.5673,
+      "step": 2995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965000162671842e-05,
+      "loss": 3.6921,
+      "step": 3000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964879511144465e-05,
+      "loss": 3.6252,
+      "step": 3005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964758652752539e-05,
+      "loss": 3.6994,
+      "step": 3010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964637587501102e-05,
+      "loss": 3.5901,
+      "step": 3015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964516315395198e-05,
+      "loss": 3.608,
+      "step": 3020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964394836439877e-05,
+      "loss": 3.6343,
+      "step": 3025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964273150640203e-05,
+      "loss": 3.633,
+      "step": 3030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964151258001247e-05,
+      "loss": 3.5528,
+      "step": 3035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964029158528085e-05,
+      "loss": 3.6649,
+      "step": 3040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963906852225806e-05,
+      "loss": 3.7181,
+      "step": 3045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963784339099506e-05,
+      "loss": 3.5542,
+      "step": 3050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963661619154289e-05,
+      "loss": 3.6142,
+      "step": 3055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96353869239527e-05,
+      "loss": 3.6908,
+      "step": 3060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963415558827566e-05,
+      "loss": 3.6327,
+      "step": 3065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963292218456313e-05,
+      "loss": 4.1142,
+      "step": 3070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963168671286647e-05,
+      "loss": 3.7395,
+      "step": 3075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963044917323717e-05,
+      "loss": 3.5108,
+      "step": 3080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962920956572678e-05,
+      "loss": 3.613,
+      "step": 3085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962796789038698e-05,
+      "loss": 3.5607,
+      "step": 3090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962672414726946e-05,
+      "loss": 3.6774,
+      "step": 3095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962547833642606e-05,
+      "loss": 3.6508,
+      "step": 3100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962423045790869e-05,
+      "loss": 3.8347,
+      "step": 3105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962298051176935e-05,
+      "loss": 3.5638,
+      "step": 3110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962172849806011e-05,
+      "loss": 3.5978,
+      "step": 3115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962047441683313e-05,
+      "loss": 3.5277,
+      "step": 3120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961921826814069e-05,
+      "loss": 3.5847,
+      "step": 3125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96179600520351e-05,
+      "loss": 3.6733,
+      "step": 3130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961669976856879e-05,
+      "loss": 3.7302,
+      "step": 3135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961543741779427e-05,
+      "loss": 3.7044,
+      "step": 3140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961417299976416e-05,
+      "loss": 3.5079,
+      "step": 3145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96129065145311e-05,
+      "loss": 3.5669,
+      "step": 3150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961163796214789e-05,
+      "loss": 3.4367,
+      "step": 3155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961036734266738e-05,
+      "loss": 3.5754,
+      "step": 3160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96090946561425e-05,
+      "loss": 3.535,
+      "step": 3165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960781990262631e-05,
+      "loss": 3.6315,
+      "step": 3170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960654308217188e-05,
+      "loss": 3.5252,
+      "step": 3175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960526419483243e-05,
+      "loss": 3.4845,
+      "step": 3180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960398324066124e-05,
+      "loss": 3.6144,
+      "step": 3185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960270021971168e-05,
+      "loss": 3.4807,
+      "step": 3190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960141513203722e-05,
+      "loss": 3.5003,
+      "step": 3195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96001279776914e-05,
+      "loss": 3.6925,
+      "step": 3200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959883875672784e-05,
+      "loss": 3.4753,
+      "step": 3205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959754746920027e-05,
+      "loss": 3.5775,
+      "step": 3210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959625411516249e-05,
+      "loss": 3.5095,
+      "step": 3215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959495869466837e-05,
+      "loss": 3.566,
+      "step": 3220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95936612077719e-05,
+      "loss": 3.506,
+      "step": 3225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959236165452714e-05,
+      "loss": 3.4896,
+      "step": 3230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959106003498824e-05,
+      "loss": 3.6424,
+      "step": 3235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958975634920943e-05,
+      "loss": 3.6044,
+      "step": 3240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958845059724502e-05,
+      "loss": 3.4675,
+      "step": 3245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958714277914942e-05,
+      "loss": 3.5242,
+      "step": 3250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958583289497714e-05,
+      "loss": 3.4823,
+      "step": 3255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958452094478273e-05,
+      "loss": 3.4791,
+      "step": 3260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958320692862086e-05,
+      "loss": 3.4719,
+      "step": 3265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958189084654628e-05,
+      "loss": 3.6643,
+      "step": 3270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958057269861383e-05,
+      "loss": 3.5136,
+      "step": 3275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957925248487843e-05,
+      "loss": 3.6911,
+      "step": 3280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957793020539508e-05,
+      "loss": 3.5952,
+      "step": 3285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957660586021888e-05,
+      "loss": 3.5041,
+      "step": 3290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957527944940502e-05,
+      "loss": 3.4999,
+      "step": 3295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957395097300874e-05,
+      "loss": 3.7052,
+      "step": 3300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957262043108542e-05,
+      "loss": 3.6078,
+      "step": 3305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957128782369049e-05,
+      "loss": 3.6518,
+      "step": 3310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956995315087945e-05,
+      "loss": 3.5553,
+      "step": 3315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956861641270794e-05,
+      "loss": 3.5532,
+      "step": 3320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956727760923164e-05,
+      "loss": 3.5346,
+      "step": 3325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956593674050634e-05,
+      "loss": 3.3782,
+      "step": 3330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956459380658791e-05,
+      "loss": 3.4515,
+      "step": 3335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95632488075323e-05,
+      "loss": 3.421,
+      "step": 3340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956190174339552e-05,
+      "loss": 3.6851,
+      "step": 3345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956055261423375e-05,
+      "loss": 3.5862,
+      "step": 3350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955920142010317e-05,
+      "loss": 3.5198,
+      "step": 3355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95578481610601e-05,
+      "loss": 3.5807,
+      "step": 3360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955649283716089e-05,
+      "loss": 3.5529,
+      "step": 3365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955513544846205e-05,
+      "loss": 3.493,
+      "step": 3370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955377599502009e-05,
+      "loss": 3.544,
+      "step": 3375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95524144768917e-05,
+      "loss": 3.6384,
+      "step": 3380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955105089413358e-05,
+      "loss": 3.5704,
+      "step": 3385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954968524680254e-05,
+      "loss": 3.4637,
+      "step": 3390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954831753495551e-05,
+      "loss": 3.4671,
+      "step": 3395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954694775864946e-05,
+      "loss": 3.5318,
+      "step": 3400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954557591794145e-05,
+      "loss": 3.5891,
+      "step": 3405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954420201288866e-05,
+      "loss": 3.6734,
+      "step": 3410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954282604354832e-05,
+      "loss": 3.5979,
+      "step": 3415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954144800997775e-05,
+      "loss": 3.5402,
+      "step": 3420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95400679122344e-05,
+      "loss": 3.481,
+      "step": 3425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953868575037575e-05,
+      "loss": 3.6445,
+      "step": 3430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953730152445939e-05,
+      "loss": 3.4723,
+      "step": 3435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9535915234543e-05,
+      "loss": 3.3773,
+      "step": 3440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953452688068434e-05,
+      "loss": 3.5953,
+      "step": 3445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953313646294125e-05,
+      "loss": 3.473,
+      "step": 3450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953174398137166e-05,
+      "loss": 3.4873,
+      "step": 3455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953034943603361e-05,
+      "loss": 3.4852,
+      "step": 3460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952895282698517e-05,
+      "loss": 3.4974,
+      "step": 3465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952755415428456e-05,
+      "loss": 3.5111,
+      "step": 3470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952615341799006e-05,
+      "loss": 3.5446,
+      "step": 3475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952475061816e-05,
+      "loss": 3.3728,
+      "step": 3480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952334575485287e-05,
+      "loss": 3.4672,
+      "step": 3485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952193882812718e-05,
+      "loss": 3.4318,
+      "step": 3490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952052983804152e-05,
+      "loss": 3.4436,
+      "step": 3495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951911878465465e-05,
+      "loss": 3.6293,
+      "step": 3500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951770566802535e-05,
+      "loss": 3.5053,
+      "step": 3505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951629048821248e-05,
+      "loss": 3.5206,
+      "step": 3510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951487324527502e-05,
+      "loss": 3.6428,
+      "step": 3515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9513453939272e-05,
+      "loss": 3.5191,
+      "step": 3520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951203257026259e-05,
+      "loss": 3.4813,
+      "step": 3525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951060913830597e-05,
+      "loss": 3.5473,
+      "step": 3530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95091836434615e-05,
+      "loss": 3.5187,
+      "step": 3535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950775608578852e-05,
+      "loss": 3.4161,
+      "step": 3540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950632646534654e-05,
+      "loss": 3.4291,
+      "step": 3545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950489478219513e-05,
+      "loss": 3.6152,
+      "step": 3550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950346103639392e-05,
+      "loss": 3.4532,
+      "step": 3555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950202522800267e-05,
+      "loss": 3.4047,
+      "step": 3560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950058735708119e-05,
+      "loss": 3.5802,
+      "step": 3565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94991474236894e-05,
+      "loss": 3.6083,
+      "step": 3570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949770542788727e-05,
+      "loss": 3.5919,
+      "step": 3575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949626136973492e-05,
+      "loss": 3.5198,
+      "step": 3580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949481524929249e-05,
+      "loss": 3.4496,
+      "step": 3585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949336706662024e-05,
+      "loss": 3.5822,
+      "step": 3590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949191682177853e-05,
+      "loss": 3.4473,
+      "step": 3595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949046451482775e-05,
+      "loss": 3.5359,
+      "step": 3600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948901014582843e-05,
+      "loss": 3.5929,
+      "step": 3605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948755371484118e-05,
+      "loss": 3.4388,
+      "step": 3610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948609522192665e-05,
+      "loss": 3.5946,
+      "step": 3615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948463466714563e-05,
+      "loss": 3.4513,
+      "step": 3620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948317205055898e-05,
+      "loss": 3.4974,
+      "step": 3625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948170737222762e-05,
+      "loss": 3.3208,
+      "step": 3630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948024063221259e-05,
+      "loss": 3.4279,
+      "step": 3635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947877183057501e-05,
+      "loss": 3.3797,
+      "step": 3640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947730096737609e-05,
+      "loss": 3.4428,
+      "step": 3645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947582804267706e-05,
+      "loss": 3.6542,
+      "step": 3650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947435305653934e-05,
+      "loss": 3.4323,
+      "step": 3655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947287600902438e-05,
+      "loss": 3.514,
+      "step": 3660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94713969001937e-05,
+      "loss": 3.582,
+      "step": 3665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946991573010893e-05,
+      "loss": 3.563,
+      "step": 3670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946843249883182e-05,
+      "loss": 3.4311,
+      "step": 3675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946694720642413e-05,
+      "loss": 3.4795,
+      "step": 3680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946545985294775e-05,
+      "loss": 3.522,
+      "step": 3685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946397043846469e-05,
+      "loss": 3.3382,
+      "step": 3690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946247896303695e-05,
+      "loss": 3.3988,
+      "step": 3695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946098542672672e-05,
+      "loss": 3.3815,
+      "step": 3700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945948982959621e-05,
+      "loss": 3.3522,
+      "step": 3705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945799217170773e-05,
+      "loss": 3.4279,
+      "step": 3710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94564924531237e-05,
+      "loss": 3.5861,
+      "step": 3715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945499067390658e-05,
+      "loss": 3.3549,
+      "step": 3720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945348683411894e-05,
+      "loss": 3.3517,
+      "step": 3725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945198093382349e-05,
+      "loss": 3.5562,
+      "step": 3730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94504729730829e-05,
+      "loss": 3.3431,
+      "step": 3735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944896295196007e-05,
+      "loss": 3.5508,
+      "step": 3740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944745087051788e-05,
+      "loss": 3.4056,
+      "step": 3745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944593672881932e-05,
+      "loss": 3.4244,
+      "step": 3750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944442052692751e-05,
+      "loss": 3.4169,
+      "step": 3755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94429022649056e-05,
+      "loss": 3.5636,
+      "step": 3760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944138194281685e-05,
+      "loss": 3.4949,
+      "step": 3765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943985956072461e-05,
+      "loss": 3.4169,
+      "step": 3770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943833511869232e-05,
+      "loss": 3.4883,
+      "step": 3775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94368086167835e-05,
+      "loss": 3.391,
+      "step": 3780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943528005506173e-05,
+      "loss": 3.4834,
+      "step": 3785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943374943359071e-05,
+      "loss": 3.5161,
+      "step": 3790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943221675243422e-05,
+      "loss": 3.4218,
+      "step": 3795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943068201165611e-05,
+      "loss": 3.4834,
+      "step": 3800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942914521132032e-05,
+      "loss": 3.5535,
+      "step": 3805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94276063514909e-05,
+      "loss": 3.4559,
+      "step": 3810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942606543223196e-05,
+      "loss": 3.2759,
+      "step": 3815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942452245360772e-05,
+      "loss": 3.6232,
+      "step": 3820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942297741568243e-05,
+      "loss": 3.4975,
+      "step": 3825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942143031852049e-05,
+      "loss": 3.3843,
+      "step": 3830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941988116218635e-05,
+      "loss": 3.4417,
+      "step": 3835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941832994674458e-05,
+      "loss": 3.4021,
+      "step": 3840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941677667225979e-05,
+      "loss": 3.5199,
+      "step": 3845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94152213387967e-05,
+      "loss": 3.3713,
+      "step": 3850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941366394642012e-05,
+      "loss": 3.5729,
+      "step": 3855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941210449519493e-05,
+      "loss": 3.4776,
+      "step": 3860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941054298518612e-05,
+      "loss": 3.461,
+      "step": 3865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940897941645875e-05,
+      "loss": 3.5183,
+      "step": 3870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940741378907796e-05,
+      "loss": 3.6227,
+      "step": 3875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940584610310899e-05,
+      "loss": 3.4034,
+      "step": 3880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940427635861713e-05,
+      "loss": 3.4379,
+      "step": 3885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940270455566783e-05,
+      "loss": 3.5005,
+      "step": 3890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940113069432656e-05,
+      "loss": 3.4232,
+      "step": 3895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939955477465887e-05,
+      "loss": 3.408,
+      "step": 3900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939797679673044e-05,
+      "loss": 3.1964,
+      "step": 3905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939639676060703e-05,
+      "loss": 3.4299,
+      "step": 3910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939481466635448e-05,
+      "loss": 3.3159,
+      "step": 3915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939323051403869e-05,
+      "loss": 3.4763,
+      "step": 3920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939164430372566e-05,
+      "loss": 3.5309,
+      "step": 3925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939005603548148e-05,
+      "loss": 3.4454,
+      "step": 3930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938846570937236e-05,
+      "loss": 3.4187,
+      "step": 3935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938687332546453e-05,
+      "loss": 3.4039,
+      "step": 3940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938527888382432e-05,
+      "loss": 3.597,
+      "step": 3945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93836823845182e-05,
+      "loss": 3.4083,
+      "step": 3950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938208382761269e-05,
+      "loss": 3.4528,
+      "step": 3955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938048321317436e-05,
+      "loss": 3.4723,
+      "step": 3960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937888054126995e-05,
+      "loss": 3.317,
+      "step": 3965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937727581196618e-05,
+      "loss": 3.4667,
+      "step": 3970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937566902532996e-05,
+      "loss": 3.4721,
+      "step": 3975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93740601814282e-05,
+      "loss": 3.3525,
+      "step": 3980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937244928032796e-05,
+      "loss": 3.4649,
+      "step": 3985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937083632209637e-05,
+      "loss": 3.5115,
+      "step": 3990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936922130680058e-05,
+      "loss": 3.5014,
+      "step": 3995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936760423450794e-05,
+      "loss": 3.3868,
+      "step": 4000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936598510528579e-05,
+      "loss": 3.431,
+      "step": 4005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936436391920161e-05,
+      "loss": 3.7137,
+      "step": 4010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936274067632294e-05,
+      "loss": 3.5045,
+      "step": 4015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936111537671742e-05,
+      "loss": 3.5427,
+      "step": 4020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935948802045275e-05,
+      "loss": 3.4491,
+      "step": 4025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935785860759675e-05,
+      "loss": 3.5644,
+      "step": 4030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935622713821732e-05,
+      "loss": 3.4632,
+      "step": 4035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935459361238242e-05,
+      "loss": 3.2483,
+      "step": 4040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935295803016012e-05,
+      "loss": 3.5098,
+      "step": 4045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935132039161857e-05,
+      "loss": 3.4656,
+      "step": 4050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934968069682597e-05,
+      "loss": 3.5054,
+      "step": 4055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934803894585069e-05,
+      "loss": 3.2499,
+      "step": 4060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934639513876111e-05,
+      "loss": 3.4477,
+      "step": 4065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934474927562572e-05,
+      "loss": 3.4182,
+      "step": 4070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93431013565131e-05,
+      "loss": 3.3554,
+      "step": 4075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934145138149189e-05,
+      "loss": 3.4281,
+      "step": 4080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933979935063087e-05,
+      "loss": 3.4125,
+      "step": 4085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933814526399884e-05,
+      "loss": 3.3887,
+      "step": 4090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933648912166476e-05,
+      "loss": 3.3311,
+      "step": 4095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933483092369761e-05,
+      "loss": 3.3876,
+      "step": 4100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933317067016648e-05,
+      "loss": 3.4084,
+      "step": 4105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933150836114054e-05,
+      "loss": 3.4303,
+      "step": 4110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932984399668905e-05,
+      "loss": 3.3564,
+      "step": 4115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932817757688138e-05,
+      "loss": 3.4824,
+      "step": 4120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932650910178694e-05,
+      "loss": 3.4804,
+      "step": 4125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932483857147524e-05,
+      "loss": 3.3346,
+      "step": 4130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932316598601592e-05,
+      "loss": 3.5383,
+      "step": 4135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932149134547863e-05,
+      "loss": 3.2788,
+      "step": 4140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931981464993317e-05,
+      "loss": 3.4572,
+      "step": 4145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931813589944938e-05,
+      "loss": 3.4392,
+      "step": 4150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931645509409724e-05,
+      "loss": 3.5488,
+      "step": 4155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931477223394674e-05,
+      "loss": 3.4756,
+      "step": 4160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931308731906804e-05,
+      "loss": 3.4326,
+      "step": 4165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93114003495313e-05,
+      "loss": 3.3961,
+      "step": 4170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930971132540683e-05,
+      "loss": 3.4428,
+      "step": 4175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9308020246765e-05,
+      "loss": 3.4421,
+      "step": 4180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930632711367629e-05,
+      "loss": 3.5235,
+      "step": 4185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930463192621122e-05,
+      "loss": 3.4518,
+      "step": 4190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930293468444043e-05,
+      "loss": 3.344,
+      "step": 4195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930123538843463e-05,
+      "loss": 3.4712,
+      "step": 4200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929953403826465e-05,
+      "loss": 3.4119,
+      "step": 4205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929783063400133e-05,
+      "loss": 3.4843,
+      "step": 4210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929612517571566e-05,
+      "loss": 3.4964,
+      "step": 4215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929441766347873e-05,
+      "loss": 3.4155,
+      "step": 4220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929270809736167e-05,
+      "loss": 3.354,
+      "step": 4225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929099647743567e-05,
+      "loss": 3.4001,
+      "step": 4230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.92892828037721e-05,
+      "loss": 3.2977,
+      "step": 4235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928756707644234e-05,
+      "loss": 3.3494,
+      "step": 4240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928584929551787e-05,
+      "loss": 3.3298,
+      "step": 4245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928412946107028e-05,
+      "loss": 3.3397,
+      "step": 4250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928240757317119e-05,
+      "loss": 3.4139,
+      "step": 4255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92806836318924e-05,
+      "loss": 3.3456,
+      "step": 4260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92789576373057e-05,
+      "loss": 3.404,
+      "step": 4265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9277229589483e-05,
+      "loss": 3.391,
+      "step": 4270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927549948849631e-05,
+      "loss": 3.3742,
+      "step": 4275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927376733441774e-05,
+      "loss": 3.3372,
+      "step": 4280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927203312731943e-05,
+      "loss": 3.4676,
+      "step": 4285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927029686727365e-05,
+      "loss": 3.3418,
+      "step": 4290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926855855435273e-05,
+      "loss": 3.2624,
+      "step": 4295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92668181886291e-05,
+      "loss": 3.3555,
+      "step": 4300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92650757701753e-05,
+      "loss": 3.4181,
+      "step": 4305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926333129906391e-05,
+      "loss": 3.3713,
+      "step": 4310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92615847753676e-05,
+      "loss": 3.4854,
+      "step": 4315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925983619915916e-05,
+      "loss": 3.6349,
+      "step": 4320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925808557051144e-05,
+      "loss": 3.421,
+      "step": 4325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925633288949737e-05,
+      "loss": 3.3582,
+      "step": 4330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925457815618999e-05,
+      "loss": 3.4565,
+      "step": 4335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925282137066241e-05,
+      "loss": 3.3415,
+      "step": 4340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925106253298784e-05,
+      "loss": 3.338,
+      "step": 4345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924930164323953e-05,
+      "loss": 3.3409,
+      "step": 4350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924753870149087e-05,
+      "loss": 3.2578,
+      "step": 4355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924577370781532e-05,
+      "loss": 3.4736,
+      "step": 4360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924400666228639e-05,
+      "loss": 3.4416,
+      "step": 4365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924223756497775e-05,
+      "loss": 3.3527,
+      "step": 4370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924046641596307e-05,
+      "loss": 3.3367,
+      "step": 4375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923869321531616e-05,
+      "loss": 3.3464,
+      "step": 4380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923691796311092e-05,
+      "loss": 3.413,
+      "step": 4385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923514065942129e-05,
+      "loss": 3.3151,
+      "step": 4390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923336130432133e-05,
+      "loss": 3.3298,
+      "step": 4395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923157989788519e-05,
+      "loss": 3.4176,
+      "step": 4400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922979644018707e-05,
+      "loss": 3.4418,
+      "step": 4405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92280109313013e-05,
+      "loss": 3.3828,
+      "step": 4410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922622337130228e-05,
+      "loss": 3.3114,
+      "step": 4415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922443376026445e-05,
+      "loss": 3.3399,
+      "step": 4420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922264209826242e-05,
+      "loss": 3.2739,
+      "step": 4425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922084838537081e-05,
+      "loss": 3.3195,
+      "step": 4430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921905262166438e-05,
+      "loss": 3.4492,
+      "step": 4435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921725480721792e-05,
+      "loss": 3.413,
+      "step": 4440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921545494210638e-05,
+      "loss": 3.3426,
+      "step": 4445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921365302640473e-05,
+      "loss": 3.4984,
+      "step": 4450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921184906018804e-05,
+      "loss": 3.4248,
+      "step": 4455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921004304353147e-05,
+      "loss": 3.2569,
+      "step": 4460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92082349765103e-05,
+      "loss": 3.4942,
+      "step": 4465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920642485919982e-05,
+      "loss": 3.4584,
+      "step": 4470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920461269167549e-05,
+      "loss": 3.4616,
+      "step": 4475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92027984740128e-05,
+      "loss": 3.3551,
+      "step": 4480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920098220628733e-05,
+      "loss": 3.3322,
+      "step": 4485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919916388857476e-05,
+      "loss": 3.4045,
+      "step": 4490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919734352095085e-05,
+      "loss": 3.2943,
+      "step": 4495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919552110349147e-05,
+      "loss": 3.3681,
+      "step": 4500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919369663627251e-05,
+      "loss": 3.432,
+      "step": 4505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919187011937004e-05,
+      "loss": 3.5231,
+      "step": 4510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919004155286012e-05,
+      "loss": 3.3164,
+      "step": 4515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918821093681896e-05,
+      "loss": 3.4082,
+      "step": 4520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918637827132282e-05,
+      "loss": 3.3823,
+      "step": 4525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918454355644807e-05,
+      "loss": 3.4488,
+      "step": 4530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918270679227114e-05,
+      "loss": 3.3504,
+      "step": 4535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918086797886857e-05,
+      "loss": 3.4217,
+      "step": 4540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9179027116317e-05,
+      "loss": 3.282,
+      "step": 4545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917718420469306e-05,
+      "loss": 3.3723,
+      "step": 4550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917533924407361e-05,
+      "loss": 3.4253,
+      "step": 4555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917349223453549e-05,
+      "loss": 3.1981,
+      "step": 4560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917164317615566e-05,
+      "loss": 3.3994,
+      "step": 4565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916979206901117e-05,
+      "loss": 3.4501,
+      "step": 4570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916793891317912e-05,
+      "loss": 3.2841,
+      "step": 4575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916608370873677e-05,
+      "loss": 3.3051,
+      "step": 4580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916422645576136e-05,
+      "loss": 3.3199,
+      "step": 4585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916236715433032e-05,
+      "loss": 3.3987,
+      "step": 4590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91605058045211e-05,
+      "loss": 3.4167,
+      "step": 4595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915864240641124e-05,
+      "loss": 3.401,
+      "step": 4600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915677696007841e-05,
+      "loss": 3.316,
+      "step": 4605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915490946560033e-05,
+      "loss": 3.4107,
+      "step": 4610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91530399230548e-05,
+      "loss": 3.3471,
+      "step": 4615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91511683325197e-05,
+      "loss": 3.3283,
+      "step": 4620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914929469407304e-05,
+      "loss": 3.2489,
+      "step": 4625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914741900779288e-05,
+      "loss": 3.3733,
+      "step": 4630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914554127375734e-05,
+      "loss": 3.3277,
+      "step": 4635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91436614920447e-05,
+      "loss": 3.4055,
+      "step": 4640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914177966273326e-05,
+      "loss": 3.4053,
+      "step": 4645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913989578590144e-05,
+      "loss": 3.3807,
+      "step": 4650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913800986162772e-05,
+      "loss": 3.3311,
+      "step": 4655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913612188999069e-05,
+      "loss": 3.3373,
+      "step": 4660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913423187106899e-05,
+      "loss": 3.338,
+      "step": 4665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91323398049414e-05,
+      "loss": 3.3736,
+      "step": 4670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913044569168673e-05,
+      "loss": 3.3806,
+      "step": 4675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91285495313839e-05,
+      "loss": 3.3473,
+      "step": 4680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912665132411194e-05,
+      "loss": 3.3368,
+      "step": 4685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912475106994992e-05,
+      "loss": 3.3955,
+      "step": 4690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912284876897702e-05,
+      "loss": 3.4938,
+      "step": 4695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912094442127248e-05,
+      "loss": 3.317,
+      "step": 4700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911903802691569e-05,
+      "loss": 3.3148,
+      "step": 4705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911712958598605e-05,
+      "loss": 3.2585,
+      "step": 4710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911521909856307e-05,
+      "loss": 3.401,
+      "step": 4715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911330656472637e-05,
+      "loss": 3.2326,
+      "step": 4720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911139198455562e-05,
+      "loss": 3.4285,
+      "step": 4725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910947535813063e-05,
+      "loss": 3.4089,
+      "step": 4730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91075566855312e-05,
+      "loss": 3.3312,
+      "step": 4735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91056359668373e-05,
+      "loss": 3.3555,
+      "step": 4740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910371320212898e-05,
+      "loss": 3.2947,
+      "step": 4745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910178839148632e-05,
+      "loss": 3.379,
+      "step": 4750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909986153498953e-05,
+      "loss": 3.552,
+      "step": 4755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90979326327189e-05,
+      "loss": 3.4651,
+      "step": 4760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90960016847548e-05,
+      "loss": 3.3804,
+      "step": 4765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909406869117766e-05,
+      "loss": 3.3608,
+      "step": 4770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909213365206806e-05,
+      "loss": 3.2871,
+      "step": 4775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909019656750658e-05,
+      "loss": 3.2554,
+      "step": 4780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908825743757395e-05,
+      "loss": 3.2254,
+      "step": 4785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908631626235099e-05,
+      "loss": 3.3795,
+      "step": 4790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908437304191854e-05,
+      "loss": 3.4515,
+      "step": 4795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908242777635759e-05,
+      "loss": 3.3555,
+      "step": 4800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908048046574917e-05,
+      "loss": 3.3118,
+      "step": 4805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907853111017444e-05,
+      "loss": 3.1734,
+      "step": 4810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90765797097146e-05,
+      "loss": 3.342,
+      "step": 4815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907462626445096e-05,
+      "loss": 3.2776,
+      "step": 4820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907267077446493e-05,
+      "loss": 3.36,
+      "step": 4825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907071323983797e-05,
+      "loss": 3.3706,
+      "step": 4830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906875366065165e-05,
+      "loss": 3.2819,
+      "step": 4835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906679203698761e-05,
+      "loss": 3.4339,
+      "step": 4840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906482836892758e-05,
+      "loss": 3.4962,
+      "step": 4845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906286265655338e-05,
+      "loss": 3.3725,
+      "step": 4850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906089489994691e-05,
+      "loss": 3.2538,
+      "step": 4855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905892509919016e-05,
+      "loss": 3.2235,
+      "step": 4860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905695325436522e-05,
+      "loss": 3.3044,
+      "step": 4865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905497936555423e-05,
+      "loss": 3.3252,
+      "step": 4870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905300343283944e-05,
+      "loss": 3.3909,
+      "step": 4875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905102545630316e-05,
+      "loss": 3.3027,
+      "step": 4880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904904543602782e-05,
+      "loss": 3.2724,
+      "step": 4885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904706337209593e-05,
+      "loss": 3.2842,
+      "step": 4890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904507926459006e-05,
+      "loss": 3.2806,
+      "step": 4895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904309311359285e-05,
+      "loss": 3.264,
+      "step": 4900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904110491918712e-05,
+      "loss": 3.3704,
+      "step": 4905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903911468145565e-05,
+      "loss": 3.3627,
+      "step": 4910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90371224004814e-05,
+      "loss": 3.2658,
+      "step": 4915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903512807634736e-05,
+      "loss": 3.4305,
+      "step": 4920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903313170913662e-05,
+      "loss": 3.1165,
+      "step": 4925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903113329893239e-05,
+      "loss": 3.3983,
+      "step": 4930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902913284581791e-05,
+      "loss": 3.2596,
+      "step": 4935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902713034987654e-05,
+      "loss": 3.23,
+      "step": 4940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90251258111917e-05,
+      "loss": 3.3106,
+      "step": 4945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902311922984693e-05,
+      "loss": 3.2676,
+      "step": 4950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902111060592583e-05,
+      "loss": 3.2527,
+      "step": 4955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901909993951208e-05,
+      "loss": 3.3949,
+      "step": 4960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901708723068947e-05,
+      "loss": 3.1646,
+      "step": 4965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901507247954185e-05,
+      "loss": 3.3327,
+      "step": 4970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901305568615317e-05,
+      "loss": 3.5181,
+      "step": 4975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901103685060746e-05,
+      "loss": 3.2731,
+      "step": 4980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.900901597298884e-05,
+      "loss": 3.2161,
+      "step": 4985
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.900699305338149e-05,
+      "loss": 3.3801,
+      "step": 4990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.900496809186972e-05,
+      "loss": 3.4155,
+      "step": 4995
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.900294108853792e-05,
+      "loss": 3.4498,
+      "step": 5000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.900091204347049e-05,
+      "loss": 3.5043,
+      "step": 5005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899888095675201e-05,
+      "loss": 3.3042,
+      "step": 5010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89968478284671e-05,
+      "loss": 3.2406,
+      "step": 5015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899481265870047e-05,
+      "loss": 3.1785,
+      "step": 5020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899277544753692e-05,
+      "loss": 3.3052,
+      "step": 5025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899073619506132e-05,
+      "loss": 3.2328,
+      "step": 5030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898869490135865e-05,
+      "loss": 3.3731,
+      "step": 5035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898665156651395e-05,
+      "loss": 3.2402,
+      "step": 5040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898460619061237e-05,
+      "loss": 3.2178,
+      "step": 5045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898255877373912e-05,
+      "loss": 3.2527,
+      "step": 5050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898050931597952e-05,
+      "loss": 3.3198,
+      "step": 5055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897845781741894e-05,
+      "loss": 3.2928,
+      "step": 5060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897640427814289e-05,
+      "loss": 3.4127,
+      "step": 5065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89743486982369e-05,
+      "loss": 3.2708,
+      "step": 5070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897229107778663e-05,
+      "loss": 3.2147,
+      "step": 5075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89702314168778e-05,
+      "loss": 3.2455,
+      "step": 5080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896816971559626e-05,
+      "loss": 3.3819,
+      "step": 5085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896610597402787e-05,
+      "loss": 3.2653,
+      "step": 5090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896404019225864e-05,
+      "loss": 3.337,
+      "step": 5095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896197237037464e-05,
+      "loss": 3.3397,
+      "step": 5100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895990250846203e-05,
+      "loss": 3.3709,
+      "step": 5105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895783060660705e-05,
+      "loss": 3.3993,
+      "step": 5110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895575666489603e-05,
+      "loss": 3.2931,
+      "step": 5115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895368068341537e-05,
+      "loss": 3.3668,
+      "step": 5120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895160266225156e-05,
+      "loss": 3.3328,
+      "step": 5125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89495226014912e-05,
+      "loss": 3.2212,
+      "step": 5130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894744050122096e-05,
+      "loss": 3.3364,
+      "step": 5135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89453563615276e-05,
+      "loss": 3.2062,
+      "step": 5140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894327018249792e-05,
+      "loss": 3.3156,
+      "step": 5145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894118196421887e-05,
+      "loss": 3.2379,
+      "step": 5150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893909170677745e-05,
+      "loss": 3.4647,
+      "step": 5155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893699941026074e-05,
+      "loss": 3.363,
+      "step": 5160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893490507475593e-05,
+      "loss": 3.235,
+      "step": 5165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89328087003503e-05,
+      "loss": 3.2403,
+      "step": 5170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893071028713117e-05,
+      "loss": 3.2681,
+      "step": 5175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892860983518597e-05,
+      "loss": 3.3184,
+      "step": 5180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892650734460222e-05,
+      "loss": 3.2799,
+      "step": 5185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892440281546755e-05,
+      "loss": 3.4178,
+      "step": 5190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892229624786959e-05,
+      "loss": 3.3043,
+      "step": 5195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892018764189617e-05,
+      "loss": 3.2258,
+      "step": 5200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89180769976351e-05,
+      "loss": 3.3841,
+      "step": 5205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891596431517437e-05,
+      "loss": 3.3292,
+      "step": 5210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891384959460197e-05,
+      "loss": 3.3446,
+      "step": 5215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.8911732836006e-05,
+      "loss": 3.1356,
+      "step": 5220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890961403947469e-05,
+      "loss": 3.2869,
+      "step": 5225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890749320509629e-05,
+      "loss": 3.2138,
+      "step": 5230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89053703329592e-05,
+      "loss": 3.3252,
+      "step": 5235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890324542315184e-05,
+      "loss": 3.396,
+      "step": 5240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890111847576275e-05,
+      "loss": 3.251,
+      "step": 5245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889898949088059e-05,
+      "loss": 3.3871,
+      "step": 5250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.8896858468594e-05,
+      "loss": 3.3117,
+      "step": 5255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889472540899183e-05,
+      "loss": 3.2663,
+      "step": 5260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889259031216291e-05,
+      "loss": 3.2196,
+      "step": 5265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889045317819623e-05,
+      "loss": 3.4824,
+      "step": 5270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888831400718082e-05,
+      "loss": 3.3215,
+      "step": 5275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88861727992058e-05,
+      "loss": 3.268,
+      "step": 5280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888402955436041e-05,
+      "loss": 3.3047,
+      "step": 5285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888188427273395e-05,
+      "loss": 3.2366,
+      "step": 5290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887973695441577e-05,
+      "loss": 3.3498,
+      "step": 5295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887758759949537e-05,
+      "loss": 3.3313,
+      "step": 5300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88754362080623e-05,
+      "loss": 3.3793,
+      "step": 5305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88732827802062e-05,
+      "loss": 3.1216,
+      "step": 5310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887112731601677e-05,
+      "loss": 3.4555,
+      "step": 5315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886896981558384e-05,
+      "loss": 3.3322,
+      "step": 5320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886681027899731e-05,
+      "loss": 3.2239,
+      "step": 5325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886464870634715e-05,
+      "loss": 3.2681,
+      "step": 5330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886248509772343e-05,
+      "loss": 3.4341,
+      "step": 5335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886031945321628e-05,
+      "loss": 3.2486,
+      "step": 5340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885815177291593e-05,
+      "loss": 3.31,
+      "step": 5345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885598205691273e-05,
+      "loss": 3.3249,
+      "step": 5350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885381030529706e-05,
+      "loss": 3.361,
+      "step": 5355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88516365181594e-05,
+      "loss": 3.348,
+      "step": 5360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884946069559033e-05,
+      "loss": 3.3854,
+      "step": 5365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884728283768052e-05,
+      "loss": 3.2278,
+      "step": 5370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88451029445207e-05,
+      "loss": 3.1594,
+      "step": 5375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88429210162017e-05,
+      "loss": 3.157,
+      "step": 5380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884073705281442e-05,
+      "loss": 3.21,
+      "step": 5385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883855105444985e-05,
+      "loss": 3.1377,
+      "step": 5390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883636302119912e-05,
+      "loss": 3.2774,
+      "step": 5395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883417295315334e-05,
+      "loss": 3.2314,
+      "step": 5400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883198085040378e-05,
+      "loss": 3.2331,
+      "step": 5405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882978671304178e-05,
+      "loss": 3.2857,
+      "step": 5410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882759054115876e-05,
+      "loss": 3.3553,
+      "step": 5415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882539233484623e-05,
+      "loss": 3.26,
+      "step": 5420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882319209419575e-05,
+      "loss": 3.3011,
+      "step": 5425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882098981929904e-05,
+      "loss": 3.2622,
+      "step": 5430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881878551024782e-05,
+      "loss": 3.5305,
+      "step": 5435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881657916713394e-05,
+      "loss": 3.2477,
+      "step": 5440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881437079004936e-05,
+      "loss": 3.2299,
+      "step": 5445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881216037908604e-05,
+      "loss": 3.3721,
+      "step": 5450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880994793433614e-05,
+      "loss": 3.2496,
+      "step": 5455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88077334558918e-05,
+      "loss": 3.3899,
+      "step": 5460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880551694384529e-05,
+      "loss": 3.223,
+      "step": 5465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880329839828897e-05,
+      "loss": 3.1735,
+      "step": 5470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880107781931529e-05,
+      "loss": 3.1821,
+      "step": 5475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879885520701675e-05,
+      "loss": 3.3143,
+      "step": 5480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879663056148595e-05,
+      "loss": 3.2519,
+      "step": 5485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879440388281564e-05,
+      "loss": 3.4357,
+      "step": 5490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87921751710985e-05,
+      "loss": 3.3272,
+      "step": 5495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878994442642748e-05,
+      "loss": 3.4096,
+      "step": 5500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878771164889548e-05,
+      "loss": 3.3676,
+      "step": 5505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878547683859554e-05,
+      "loss": 3.0889,
+      "step": 5510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878323999562078e-05,
+      "loss": 3.1253,
+      "step": 5515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878100112006438e-05,
+      "loss": 3.3062,
+      "step": 5520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877876021201963e-05,
+      "loss": 3.2663,
+      "step": 5525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877651727157993e-05,
+      "loss": 3.202,
+      "step": 5530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877427229883868e-05,
+      "loss": 3.2663,
+      "step": 5535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877202529388947e-05,
+      "loss": 3.4619,
+      "step": 5540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87697762568259e-05,
+      "loss": 3.3543,
+      "step": 5545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876752518774166e-05,
+      "loss": 3.1981,
+      "step": 5550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876527208673057e-05,
+      "loss": 3.4382,
+      "step": 5555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876301695388649e-05,
+      "loss": 3.2137,
+      "step": 5560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87607597893034e-05,
+      "loss": 3.1932,
+      "step": 5565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875850059307531e-05,
+      "loss": 3.1983,
+      "step": 5570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875623936529638e-05,
+      "loss": 3.2666,
+      "step": 5575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875397610606083e-05,
+      "loss": 3.2948,
+      "step": 5580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875171081546294e-05,
+      "loss": 3.0876,
+      "step": 5585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874944349359711e-05,
+      "loss": 3.352,
+      "step": 5590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87471741405578e-05,
+      "loss": 3.0911,
+      "step": 5595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874490275643958e-05,
+      "loss": 3.1554,
+      "step": 5600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874262934133706e-05,
+      "loss": 3.2854,
+      "step": 5605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874035389534497e-05,
+      "loss": 3.1887,
+      "step": 5610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873807641855814e-05,
+      "loss": 3.1257,
+      "step": 5615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873579691107146e-05,
+      "loss": 3.2649,
+      "step": 5620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873351537297988e-05,
+      "loss": 3.1761,
+      "step": 5625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873123180437848e-05,
+      "loss": 3.2603,
+      "step": 5630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87289462053624e-05,
+      "loss": 3.1649,
+      "step": 5635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872665857602687e-05,
+      "loss": 3.3469,
+      "step": 5640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872436891646724e-05,
+      "loss": 3.1168,
+      "step": 5645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872207722677885e-05,
+      "loss": 3.2759,
+      "step": 5650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871978350705722e-05,
+      "loss": 3.182,
+      "step": 5655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871748775739792e-05,
+      "loss": 3.2249,
+      "step": 5660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871518997789659e-05,
+      "loss": 3.3087,
+      "step": 5665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871289016864897e-05,
+      "loss": 3.2985,
+      "step": 5670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87105883297509e-05,
+      "loss": 3.262,
+      "step": 5675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870828446129827e-05,
+      "loss": 3.2911,
+      "step": 5680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870597856338708e-05,
+      "loss": 3.3671,
+      "step": 5685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870367063611341e-05,
+      "loss": 3.2685,
+      "step": 5690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870136067957341e-05,
+      "loss": 3.3442,
+      "step": 5695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869904869386332e-05,
+      "loss": 3.3203,
+      "step": 5700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.86967346790795e-05,
+      "loss": 3.4222,
+      "step": 5705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869441863531835e-05,
+      "loss": 3.3913,
+      "step": 5710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869210056267635e-05,
+      "loss": 3.2148,
+      "step": 5715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.86897804612501e-05,
+      "loss": 3.2466,
+      "step": 5720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.868745833113627e-05,
+      "loss": 3.2399,
+      "step": 5725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.868513417243161e-05,
+      "loss": 3.1875,
+      "step": 5730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.868280798523296e-05,
+      "loss": 3.1399,
+      "step": 5735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.868047976963724e-05,
+      "loss": 3.2215,
+      "step": 5740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867814952574146e-05,
+      "loss": 3.2249,
+      "step": 5745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.86758172536427e-05,
+      "loss": 3.2974,
+      "step": 5750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867348295343814e-05,
+      "loss": 3.2343,
+      "step": 5755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867114662522505e-05,
+      "loss": 3.2614,
+      "step": 5760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.866880826910076e-05,
+      "loss": 3.263,
+      "step": 5765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.866646788516272e-05,
+      "loss": 3.2841,
+      "step": 5770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.86641254735084e-05,
+      "loss": 3.2433,
+      "step": 5775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866178103423545e-05,
+      "loss": 3.2507,
+      "step": 5780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865943456744154e-05,
+      "loss": 3.2808,
+      "step": 5785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865708607322441e-05,
+      "loss": 3.3517,
+      "step": 5790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865473555168192e-05,
+      "loss": 3.2108,
+      "step": 5795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865238300291202e-05,
+      "loss": 3.2822,
+      "step": 5800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865002842701274e-05,
+      "loss": 3.2165,
+      "step": 5805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864767182408214e-05,
+      "loss": 3.299,
+      "step": 5810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864531319421846e-05,
+      "loss": 3.1632,
+      "step": 5815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864295253751995e-05,
+      "loss": 3.1587,
+      "step": 5820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864058985408496e-05,
+      "loss": 3.3005,
+      "step": 5825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863822514401195e-05,
+      "loss": 3.3271,
+      "step": 5830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863585840739944e-05,
+      "loss": 3.2023,
+      "step": 5835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863348964434604e-05,
+      "loss": 3.1882,
+      "step": 5840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863111885495044e-05,
+      "loss": 3.3558,
+      "step": 5845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862874603931144e-05,
+      "loss": 3.1752,
+      "step": 5850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862637119752788e-05,
+      "loss": 3.2036,
+      "step": 5855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862399432969873e-05,
+      "loss": 3.105,
+      "step": 5860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862161543592301e-05,
+      "loss": 3.2356,
+      "step": 5865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861923451629984e-05,
+      "loss": 3.2165,
+      "step": 5870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861685157092842e-05,
+      "loss": 3.1998,
+      "step": 5875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861446659990805e-05,
+      "loss": 3.2413,
+      "step": 5880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861207960333809e-05,
+      "loss": 3.3389,
+      "step": 5885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860969058131799e-05,
+      "loss": 3.1248,
+      "step": 5890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860729953394731e-05,
+      "loss": 3.1805,
+      "step": 5895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860490646132566e-05,
+      "loss": 3.0911,
+      "step": 5900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860251136355275e-05,
+      "loss": 3.257,
+      "step": 5905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860011424072837e-05,
+      "loss": 3.2618,
+      "step": 5910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859771509295241e-05,
+      "loss": 3.1992,
+      "step": 5915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859531392032482e-05,
+      "loss": 3.2655,
+      "step": 5920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859291072294565e-05,
+      "loss": 3.2498,
+      "step": 5925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859050550091503e-05,
+      "loss": 3.2435,
+      "step": 5930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858809825433318e-05,
+      "loss": 3.179,
+      "step": 5935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858568898330039e-05,
+      "loss": 3.1682,
+      "step": 5940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858327768791704e-05,
+      "loss": 3.1889,
+      "step": 5945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858086436828362e-05,
+      "loss": 3.1072,
+      "step": 5950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857844902450066e-05,
+      "loss": 3.164,
+      "step": 5955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857603165666882e-05,
+      "loss": 3.2654,
+      "step": 5960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857361226488879e-05,
+      "loss": 3.2922,
+      "step": 5965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85711908492614e-05,
+      "loss": 3.2869,
+      "step": 5970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856876740988754e-05,
+      "loss": 3.08,
+      "step": 5975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856634194686815e-05,
+      "loss": 3.3397,
+      "step": 5980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856391446030435e-05,
+      "loss": 3.2818,
+      "step": 5985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856148495029723e-05,
+      "loss": 3.1866,
+      "step": 5990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855905341694802e-05,
+      "loss": 3.2982,
+      "step": 5995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855661986035805e-05,
+      "loss": 3.3018,
+      "step": 6000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855418428062873e-05,
+      "loss": 3.0977,
+      "step": 6005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855174667786151e-05,
+      "loss": 3.3037,
+      "step": 6010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854930705215795e-05,
+      "loss": 3.1942,
+      "step": 6015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854686540361972e-05,
+      "loss": 3.2829,
+      "step": 6020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854442173234854e-05,
+      "loss": 3.1451,
+      "step": 6025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854197603844624e-05,
+      "loss": 3.236,
+      "step": 6030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853952832201472e-05,
+      "loss": 3.3377,
+      "step": 6035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853707858315596e-05,
+      "loss": 3.3082,
+      "step": 6040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853462682197202e-05,
+      "loss": 3.3145,
+      "step": 6045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853217303856506e-05,
+      "loss": 3.2886,
+      "step": 6050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852971723303732e-05,
+      "loss": 3.2119,
+      "step": 6055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852725940549112e-05,
+      "loss": 3.212,
+      "step": 6060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852479955602888e-05,
+      "loss": 3.2686,
+      "step": 6065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852233768475307e-05,
+      "loss": 3.1469,
+      "step": 6070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85198737917663e-05,
+      "loss": 3.3313,
+      "step": 6075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851740787717118e-05,
+      "loss": 3.1893,
+      "step": 6080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85149399410705e-05,
+      "loss": 3.3096,
+      "step": 6085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851246998356707e-05,
+      "loss": 3.0483,
+      "step": 6090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850999800476378e-05,
+      "loss": 3.1791,
+      "step": 6095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850752400476366e-05,
+      "loss": 3.2761,
+      "step": 6100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850504798366978e-05,
+      "loss": 3.345,
+      "step": 6105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850256994158529e-05,
+      "loss": 3.2325,
+      "step": 6110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850008987861346e-05,
+      "loss": 3.2484,
+      "step": 6115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84976077948576e-05,
+      "loss": 3.2422,
+      "step": 6120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849512369042116e-05,
+      "loss": 3.1258,
+      "step": 6125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84926375654076e-05,
+      "loss": 3.3186,
+      "step": 6130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849014941992057e-05,
+      "loss": 3.132,
+      "step": 6135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848765925406366e-05,
+      "loss": 3.2943,
+      "step": 6140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848516706794068e-05,
+      "loss": 3.0116,
+      "step": 6145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848267286165544e-05,
+      "loss": 3.2844,
+      "step": 6150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848017663531188e-05,
+      "loss": 3.2129,
+      "step": 6155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847767838901402e-05,
+      "loss": 3.2831,
+      "step": 6160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847517812286591e-05,
+      "loss": 3.2902,
+      "step": 6165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847267583697176e-05,
+      "loss": 3.321,
+      "step": 6170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847017153143581e-05,
+      "loss": 3.0165,
+      "step": 6175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84676652063624e-05,
+      "loss": 3.154,
+      "step": 6180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846515686185598e-05,
+      "loss": 3.216,
+      "step": 6185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846264649802105e-05,
+      "loss": 3.1194,
+      "step": 6190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84601341149622e-05,
+      "loss": 3.1794,
+      "step": 6195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84576197127841e-05,
+      "loss": 3.2332,
+      "step": 6200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845510329159154e-05,
+      "loss": 3.325,
+      "step": 6205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845258485148937e-05,
+      "loss": 3.1917,
+      "step": 6210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845006439258247e-05,
+      "loss": 3.205,
+      "step": 6215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844754191497592e-05,
+      "loss": 3.1951,
+      "step": 6220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84450174187748e-05,
+      "loss": 3.3032,
+      "step": 6225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844249090408429e-05,
+      "loss": 3.1107,
+      "step": 6230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843996237100965e-05,
+      "loss": 3.2097,
+      "step": 6235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843743181965624e-05,
+      "loss": 3.2463,
+      "step": 6240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843489925012949e-05,
+      "loss": 3.2417,
+      "step": 6245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843236466253496e-05,
+      "loss": 3.0329,
+      "step": 6250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84298280569782e-05,
+      "loss": 3.1021,
+      "step": 6255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842728943356492e-05,
+      "loss": 3.2262,
+      "step": 6260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84247487924009e-05,
+      "loss": 3.1869,
+      "step": 6265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842220613359201e-05,
+      "loss": 3.1684,
+      "step": 6270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841966145724416e-05,
+      "loss": 3.2364,
+      "step": 6275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84171147634634e-05,
+      "loss": 3.2157,
+      "step": 6280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841456605235582e-05,
+      "loss": 3.132,
+      "step": 6285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841201532402763e-05,
+      "loss": 3.256,
+      "step": 6290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84094625785851e-05,
+      "loss": 3.2753,
+      "step": 6295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840690781613458e-05,
+      "loss": 3.237,
+      "step": 6300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840435103678254e-05,
+      "loss": 3.1953,
+      "step": 6305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84017922406355e-05,
+      "loss": 3.2281,
+      "step": 6310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839923142780008e-05,
+      "loss": 3.2862,
+      "step": 6315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839666859838294e-05,
+      "loss": 3.2248,
+      "step": 6320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839410375249091e-05,
+      "loss": 3.1183,
+      "step": 6325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839153689023084e-05,
+      "loss": 3.2468,
+      "step": 6330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838896801170967e-05,
+      "loss": 3.2139,
+      "step": 6335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838639711703446e-05,
+      "loss": 3.1792,
+      "step": 6340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838382420631229e-05,
+      "loss": 3.1401,
+      "step": 6345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838124927965038e-05,
+      "loss": 3.098,
+      "step": 6350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837867233715601e-05,
+      "loss": 3.111,
+      "step": 6355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837609337893656e-05,
+      "loss": 3.3412,
+      "step": 6360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837351240509949e-05,
+      "loss": 3.1461,
+      "step": 6365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837092941575232e-05,
+      "loss": 3.072,
+      "step": 6370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836834441100268e-05,
+      "loss": 3.2039,
+      "step": 6375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836575739095827e-05,
+      "loss": 3.1099,
+      "step": 6380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836316835572688e-05,
+      "loss": 3.1851,
+      "step": 6385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83605773054164e-05,
+      "loss": 3.2625,
+      "step": 6390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835798424013477e-05,
+      "loss": 3.181,
+      "step": 6395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835538915999003e-05,
+      "loss": 3.1817,
+      "step": 6400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835279206509032e-05,
+      "loss": 3.0841,
+      "step": 6405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835019295554386e-05,
+      "loss": 3.2543,
+      "step": 6410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83475918314589e-05,
+      "loss": 3.4404,
+      "step": 6415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834498869294385e-05,
+      "loss": 3.2412,
+      "step": 6420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834238354010716e-05,
+      "loss": 3.2919,
+      "step": 6425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833977637305738e-05,
+      "loss": 3.1278,
+      "step": 6430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833716719190316e-05,
+      "loss": 3.2761,
+      "step": 6435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833455599675317e-05,
+      "loss": 3.1738,
+      "step": 6440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833194278771623e-05,
+      "loss": 3.1528,
+      "step": 6445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83293275649012e-05,
+      "loss": 3.2363,
+      "step": 6450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.832671032841709e-05,
+      "loss": 3.192,
+      "step": 6455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.832409107837293e-05,
+      "loss": 3.245,
+      "step": 6460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.832146981487782e-05,
+      "loss": 3.1436,
+      "step": 6465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.831884653804102e-05,
+      "loss": 3.2045,
+      "step": 6470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.831622124797179e-05,
+      "loss": 3.278,
+      "step": 6475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.831359394477957e-05,
+      "loss": 3.0327,
+      "step": 6480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.831096462857377e-05,
+      "loss": 3.2183,
+      "step": 6485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.830833329946398e-05,
+      "loss": 3.2146,
+      "step": 6490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.830569995755981e-05,
+      "loss": 3.2608,
+      "step": 6495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.830306460297099e-05,
+      "loss": 3.187,
+      "step": 6500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.830042723580732e-05,
+      "loss": 3.258,
+      "step": 6505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.829778785617872e-05,
+      "loss": 3.2885,
+      "step": 6510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.829514646419511e-05,
+      "loss": 3.2719,
+      "step": 6515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.829250305996658e-05,
+      "loss": 3.1173,
+      "step": 6520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.828985764360325e-05,
+      "loss": 3.0457,
+      "step": 6525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.828721021521534e-05,
+      "loss": 3.2153,
+      "step": 6530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.82845607749132e-05,
+      "loss": 3.3294,
+      "step": 6535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.828190932280716e-05,
+      "loss": 3.1074,
+      "step": 6540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.827925585900773e-05,
+      "loss": 3.2134,
+      "step": 6545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827660038362546e-05,
+      "loss": 3.2142,
+      "step": 6550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827394289677098e-05,
+      "loss": 3.2795,
+      "step": 6555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827128339855505e-05,
+      "loss": 3.234,
+      "step": 6560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826862188908846e-05,
+      "loss": 3.2671,
+      "step": 6565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826595836848208e-05,
+      "loss": 3.2937,
+      "step": 6570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826329283684693e-05,
+      "loss": 3.2306,
+      "step": 6575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826062529429402e-05,
+      "loss": 3.1279,
+      "step": 6580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825795574093453e-05,
+      "loss": 3.2051,
+      "step": 6585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82552841768797e-05,
+      "loss": 3.2819,
+      "step": 6590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825261060224081e-05,
+      "loss": 3.1094,
+      "step": 6595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824993501712927e-05,
+      "loss": 3.2636,
+      "step": 6600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824725742165656e-05,
+      "loss": 3.2625,
+      "step": 6605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824457781593426e-05,
+      "loss": 3.1657,
+      "step": 6610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824189620007397e-05,
+      "loss": 3.1899,
+      "step": 6615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823921257418749e-05,
+      "loss": 3.2282,
+      "step": 6620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823652693838656e-05,
+      "loss": 3.1414,
+      "step": 6625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823383929278313e-05,
+      "loss": 3.2208,
+      "step": 6630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823114963748918e-05,
+      "loss": 3.3004,
+      "step": 6635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822845797261675e-05,
+      "loss": 3.1262,
+      "step": 6640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822576429827802e-05,
+      "loss": 3.1481,
+      "step": 6645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822306861458516e-05,
+      "loss": 3.1702,
+      "step": 6650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822037092165058e-05,
+      "loss": 3.2181,
+      "step": 6655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821767121958662e-05,
+      "loss": 3.1597,
+      "step": 6660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821496950850578e-05,
+      "loss": 3.0988,
+      "step": 6665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821226578852063e-05,
+      "loss": 3.153,
+      "step": 6670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820956005974382e-05,
+      "loss": 3.2555,
+      "step": 6675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820685232228808e-05,
+      "loss": 3.2109,
+      "step": 6680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820414257626623e-05,
+      "loss": 3.2247,
+      "step": 6685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820143082179118e-05,
+      "loss": 3.1911,
+      "step": 6690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819871705897591e-05,
+      "loss": 3.2616,
+      "step": 6695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819600128793351e-05,
+      "loss": 3.2286,
+      "step": 6700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819328350877712e-05,
+      "loss": 3.2896,
+      "step": 6705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819056372161998e-05,
+      "loss": 3.2149,
+      "step": 6710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81878419265754e-05,
+      "loss": 3.1838,
+      "step": 6715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81851181237568e-05,
+      "loss": 3.1895,
+      "step": 6720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818239231327766e-05,
+      "loss": 3.2896,
+      "step": 6725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817966449525154e-05,
+      "loss": 3.1035,
+      "step": 6730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817693466979213e-05,
+      "loss": 3.1984,
+      "step": 6735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817420283701315e-05,
+      "loss": 3.0946,
+      "step": 6740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817146899702842e-05,
+      "loss": 3.2679,
+      "step": 6745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816873314995185e-05,
+      "loss": 3.2333,
+      "step": 6750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816599529589742e-05,
+      "loss": 3.3176,
+      "step": 6755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816325543497924e-05,
+      "loss": 3.1449,
+      "step": 6760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816051356731143e-05,
+      "loss": 3.1456,
+      "step": 6765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815776969300825e-05,
+      "loss": 3.2135,
+      "step": 6770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815502381218402e-05,
+      "loss": 3.2501,
+      "step": 6775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815227592495315e-05,
+      "loss": 3.1916,
+      "step": 6780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814952603143014e-05,
+      "loss": 3.2535,
+      "step": 6785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814677413172955e-05,
+      "loss": 3.3086,
+      "step": 6790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814402022596604e-05,
+      "loss": 3.2059,
+      "step": 6795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814126431425437e-05,
+      "loss": 3.2229,
+      "step": 6800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813850639670936e-05,
+      "loss": 3.208,
+      "step": 6805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81357464734459e-05,
+      "loss": 3.2839,
+      "step": 6810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813298454457902e-05,
+      "loss": 3.0291,
+      "step": 6815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813022061022377e-05,
+      "loss": 3.1229,
+      "step": 6820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812745467049531e-05,
+      "loss": 3.2546,
+      "step": 6825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81246867255089e-05,
+      "loss": 3.1599,
+      "step": 6830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812191677537987e-05,
+      "loss": 3.4715,
+      "step": 6835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811914482022362e-05,
+      "loss": 3.1362,
+      "step": 6840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811637086015563e-05,
+      "loss": 3.1678,
+      "step": 6845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811359489529153e-05,
+      "loss": 3.1433,
+      "step": 6850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811081692574693e-05,
+      "loss": 3.1371,
+      "step": 6855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81080369516376e-05,
+      "loss": 3.1863,
+      "step": 6860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810525497307934e-05,
+      "loss": 3.0523,
+      "step": 6865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810247099018812e-05,
+      "loss": 3.2554,
+      "step": 6870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809968500307989e-05,
+      "loss": 3.106,
+      "step": 6875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809689701187074e-05,
+      "loss": 3.2168,
+      "step": 6880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809410701667682e-05,
+      "loss": 3.3036,
+      "step": 6885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809131501761443e-05,
+      "loss": 3.2491,
+      "step": 6890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808852101479984e-05,
+      "loss": 3.2506,
+      "step": 6895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808572500834949e-05,
+      "loss": 3.2045,
+      "step": 6900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808292699837986e-05,
+      "loss": 3.1344,
+      "step": 6905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808012698500757e-05,
+      "loss": 3.1866,
+      "step": 6910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807732496834924e-05,
+      "loss": 3.2773,
+      "step": 6915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807452094852165e-05,
+      "loss": 3.2386,
+      "step": 6920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807171492564161e-05,
+      "loss": 3.1669,
+      "step": 6925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806890689982604e-05,
+      "loss": 3.1699,
+      "step": 6930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806609687119194e-05,
+      "loss": 3.0997,
+      "step": 6935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806328483985641e-05,
+      "loss": 3.1687,
+      "step": 6940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806047080593656e-05,
+      "loss": 3.2276,
+      "step": 6945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805765476954971e-05,
+      "loss": 3.2038,
+      "step": 6950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805483673081314e-05,
+      "loss": 3.2125,
+      "step": 6955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805201668984428e-05,
+      "loss": 3.0678,
+      "step": 6960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804919464676063e-05,
+      "loss": 3.0938,
+      "step": 6965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804637060167979e-05,
+      "loss": 3.2255,
+      "step": 6970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804354455471937e-05,
+      "loss": 3.2346,
+      "step": 6975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80407165059972e-05,
+      "loss": 3.2753,
+      "step": 6980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.803788645563104e-05,
+      "loss": 3.0766,
+      "step": 6985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.803505440373883e-05,
+      "loss": 3.0762,
+      "step": 6990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.803222035043859e-05,
+      "loss": 3.3248,
+      "step": 6995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802938429584839e-05,
+      "loss": 3.1464,
+      "step": 7000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802654624008638e-05,
+      "loss": 3.025,
+      "step": 7005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802370618327083e-05,
+      "loss": 3.2183,
+      "step": 7010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802086412552007e-05,
+      "loss": 3.2765,
+      "step": 7015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801802006695249e-05,
+      "loss": 3.2733,
+      "step": 7020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801517400768663e-05,
+      "loss": 3.2329,
+      "step": 7025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801232594784102e-05,
+      "loss": 3.1332,
+      "step": 7030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800947588753438e-05,
+      "loss": 3.1224,
+      "step": 7035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800662382688544e-05,
+      "loss": 3.1447,
+      "step": 7040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800376976601302e-05,
+      "loss": 3.1116,
+      "step": 7045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800091370503605e-05,
+      "loss": 3.1752,
+      "step": 7050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799805564407354e-05,
+      "loss": 3.1091,
+      "step": 7055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799519558324453e-05,
+      "loss": 3.0978,
+      "step": 7060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799233352266823e-05,
+      "loss": 3.177,
+      "step": 7065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798946946246388e-05,
+      "loss": 3.1987,
+      "step": 7070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798660340275079e-05,
+      "loss": 3.1548,
+      "step": 7075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79837353436484e-05,
+      "loss": 3.1021,
+      "step": 7080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79808652852762e-05,
+      "loss": 3.118,
+      "step": 7085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797799322775379e-05,
+      "loss": 3.1485,
+      "step": 7090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797511917120079e-05,
+      "loss": 3.217,
+      "step": 7095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.7972243115737e-05,
+      "loss": 3.2215,
+      "step": 7100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.796936506148222e-05,
+      "loss": 3.163,
+      "step": 7105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.796648500855639e-05,
+      "loss": 3.0794,
+      "step": 7110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79636029570795e-05,
+      "loss": 3.1988,
+      "step": 7115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.796071890717162e-05,
+      "loss": 3.2736,
+      "step": 7120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795783285895293e-05,
+      "loss": 3.2073,
+      "step": 7125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795494481254366e-05,
+      "loss": 3.1525,
+      "step": 7130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795205476806416e-05,
+      "loss": 3.0109,
+      "step": 7135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794916272563484e-05,
+      "loss": 3.1194,
+      "step": 7140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794626868537622e-05,
+      "loss": 3.0873,
+      "step": 7145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794337264740884e-05,
+      "loss": 3.2034,
+      "step": 7150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794047461185339e-05,
+      "loss": 3.2909,
+      "step": 7155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793757457883062e-05,
+      "loss": 2.9677,
+      "step": 7160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793467254846135e-05,
+      "loss": 3.1492,
+      "step": 7165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79317685208665e-05,
+      "loss": 3.0689,
+      "step": 7170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792886249616709e-05,
+      "loss": 3.137,
+      "step": 7175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792595447448416e-05,
+      "loss": 3.1696,
+      "step": 7180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79230444559389e-05,
+      "loss": 3.161,
+      "step": 7185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792013244065255e-05,
+      "loss": 3.2378,
+      "step": 7190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.791721842874644e-05,
+      "loss": 3.1803,
+      "step": 7195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.791430242034198e-05,
+      "loss": 3.0979,
+      "step": 7200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79113844155607e-05,
+      "loss": 2.9737,
+      "step": 7205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790846441452413e-05,
+      "loss": 2.9762,
+      "step": 7210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790554241735397e-05,
+      "loss": 3.1044,
+      "step": 7215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790261842417194e-05,
+      "loss": 3.1765,
+      "step": 7220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78996924350999e-05,
+      "loss": 3.0833,
+      "step": 7225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.789676445025975e-05,
+      "loss": 3.1933,
+      "step": 7230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.789383446977347e-05,
+      "loss": 3.2769,
+      "step": 7235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.789090249376315e-05,
+      "loss": 3.24,
+      "step": 7240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.788796852235098e-05,
+      "loss": 3.3102,
+      "step": 7245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.788503255565917e-05,
+      "loss": 3.125,
+      "step": 7250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.788209459381006e-05,
+      "loss": 3.1681,
+      "step": 7255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.787915463692606e-05,
+      "loss": 3.0513,
+      "step": 7260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.787621268512965e-05,
+      "loss": 3.2129,
+      "step": 7265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.787326873854345e-05,
+      "loss": 3.1531,
+      "step": 7270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.787032279729009e-05,
+      "loss": 3.1314,
+      "step": 7275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.786737486149231e-05,
+      "loss": 3.2413,
+      "step": 7280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.786442493127295e-05,
+      "loss": 3.2096,
+      "step": 7285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.786147300675492e-05,
+      "loss": 3.2428,
+      "step": 7290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.785851908806121e-05,
+      "loss": 3.0966,
+      "step": 7295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.785556317531491e-05,
+      "loss": 3.2244,
+      "step": 7300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.785260526863916e-05,
+      "loss": 3.1338,
+      "step": 7305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.784964536815722e-05,
+      "loss": 3.1389,
+      "step": 7310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.784668347399239e-05,
+      "loss": 3.219,
+      "step": 7315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.784371958626811e-05,
+      "loss": 3.2614,
+      "step": 7320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.784075370510786e-05,
+      "loss": 3.143,
+      "step": 7325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78377858306352e-05,
+      "loss": 3.0171,
+      "step": 7330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783481596297381e-05,
+      "loss": 3.125,
+      "step": 7335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783184410224741e-05,
+      "loss": 3.2852,
+      "step": 7340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782887024857986e-05,
+      "loss": 3.1528,
+      "step": 7345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782589440209504e-05,
+      "loss": 3.1535,
+      "step": 7350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782291656291694e-05,
+      "loss": 3.0512,
+      "step": 7355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.781993673116962e-05,
+      "loss": 3.1299,
+      "step": 7360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.781695490697729e-05,
+      "loss": 3.0069,
+      "step": 7365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.781397109046412e-05,
+      "loss": 3.1099,
+      "step": 7370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.781098528175449e-05,
+      "loss": 3.0891,
+      "step": 7375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780799748097277e-05,
+      "loss": 3.1509,
+      "step": 7380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780500768824345e-05,
+      "loss": 3.2237,
+      "step": 7385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780201590369113e-05,
+      "loss": 3.2066,
+      "step": 7390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779902212744045e-05,
+      "loss": 3.2969,
+      "step": 7395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779602635961613e-05,
+      "loss": 3.0487,
+      "step": 7400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7793028600343e-05,
+      "loss": 3.1249,
+      "step": 7405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7790028849746e-05,
+      "loss": 3.187,
+      "step": 7410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778702710795004e-05,
+      "loss": 3.028,
+      "step": 7415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778402337508026e-05,
+      "loss": 3.1999,
+      "step": 7420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778101765126178e-05,
+      "loss": 3.1794,
+      "step": 7425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777800993661983e-05,
+      "loss": 3.0901,
+      "step": 7430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777500023127975e-05,
+      "loss": 3.2831,
+      "step": 7435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777198853536694e-05,
+      "loss": 3.0975,
+      "step": 7440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776897484900685e-05,
+      "loss": 3.1071,
+      "step": 7445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776595917232507e-05,
+      "loss": 3.0061,
+      "step": 7450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776294150544727e-05,
+      "loss": 3.0034,
+      "step": 7455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775992184849916e-05,
+      "loss": 3.3711,
+      "step": 7460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775690020160655e-05,
+      "loss": 3.2837,
+      "step": 7465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775387656489535e-05,
+      "loss": 3.1972,
+      "step": 7470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775085093849155e-05,
+      "loss": 3.2037,
+      "step": 7475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774782332252119e-05,
+      "loss": 3.0422,
+      "step": 7480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774479371711043e-05,
+      "loss": 2.9882,
+      "step": 7485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77417621223855e-05,
+      "loss": 3.0568,
+      "step": 7490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.773872853847272e-05,
+      "loss": 3.1684,
+      "step": 7495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.773569296549849e-05,
+      "loss": 3.1133,
+      "step": 7500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.773265540358926e-05,
+      "loss": 3.1338,
+      "step": 7505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772961585287163e-05,
+      "loss": 3.2609,
+      "step": 7510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77265743134722e-05,
+      "loss": 3.0692,
+      "step": 7515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772353078551775e-05,
+      "loss": 3.1679,
+      "step": 7520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772048526913505e-05,
+      "loss": 3.0888,
+      "step": 7525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771743776445102e-05,
+      "loss": 3.1064,
+      "step": 7530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77143882715926e-05,
+      "loss": 3.1972,
+      "step": 7535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771133679068689e-05,
+      "loss": 3.1869,
+      "step": 7540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770828332186101e-05,
+      "loss": 3.0234,
+      "step": 7545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770522786524219e-05,
+      "loss": 3.1413,
+      "step": 7550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770217042095772e-05,
+      "loss": 3.0988,
+      "step": 7555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769911098913502e-05,
+      "loss": 3.1338,
+      "step": 7560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769604956990155e-05,
+      "loss": 3.0511,
+      "step": 7565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769298616338485e-05,
+      "loss": 3.1803,
+      "step": 7570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76899207697126e-05,
+      "loss": 3.1739,
+      "step": 7575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768685338901246e-05,
+      "loss": 3.177,
+      "step": 7580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768378402141229e-05,
+      "loss": 3.1869,
+      "step": 7585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768071266703995e-05,
+      "loss": 3.1314,
+      "step": 7590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76776393260234e-05,
+      "loss": 3.2719,
+      "step": 7595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767456399849073e-05,
+      "loss": 3.1799,
+      "step": 7600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767148668457005e-05,
+      "loss": 3.1303,
+      "step": 7605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766840738438957e-05,
+      "loss": 3.2289,
+      "step": 7610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766532609807761e-05,
+      "loss": 3.1074,
+      "step": 7615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766224282576254e-05,
+      "loss": 3.1015,
+      "step": 7620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765915756757283e-05,
+      "loss": 3.1517,
+      "step": 7625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765607032363704e-05,
+      "loss": 3.1644,
+      "step": 7630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765298109408379e-05,
+      "loss": 3.0278,
+      "step": 7635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.764988987904178e-05,
+      "loss": 3.203,
+      "step": 7640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.764679667863984e-05,
+      "loss": 3.1732,
+      "step": 7645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.764370149300683e-05,
+      "loss": 3.2541,
+      "step": 7650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.764060432227173e-05,
+      "loss": 3.1977,
+      "step": 7655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.763750516656354e-05,
+      "loss": 3.1448,
+      "step": 7660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.763440402601145e-05,
+      "loss": 3.2023,
+      "step": 7665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.763130090074463e-05,
+      "loss": 3.0268,
+      "step": 7670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.762819579089239e-05,
+      "loss": 3.0828,
+      "step": 7675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.762508869658408e-05,
+      "loss": 3.1244,
+      "step": 7680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.762197961794921e-05,
+      "loss": 3.1719,
+      "step": 7685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761886855511727e-05,
+      "loss": 3.1893,
+      "step": 7690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761575550821791e-05,
+      "loss": 3.0634,
+      "step": 7695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761264047738083e-05,
+      "loss": 3.1217,
+      "step": 7700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760952346273582e-05,
+      "loss": 3.0384,
+      "step": 7705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760640446441277e-05,
+      "loss": 3.0361,
+      "step": 7710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76032834825416e-05,
+      "loss": 3.1317,
+      "step": 7715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760016051725238e-05,
+      "loss": 2.9923,
+      "step": 7720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.759703556867521e-05,
+      "loss": 3.0823,
+      "step": 7725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75939086369403e-05,
+      "loss": 3.1186,
+      "step": 7730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.759077972217794e-05,
+      "loss": 3.1411,
+      "step": 7735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758764882451848e-05,
+      "loss": 3.1068,
+      "step": 7740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758451594409239e-05,
+      "loss": 3.1523,
+      "step": 7745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758138108103022e-05,
+      "loss": 3.2531,
+      "step": 7750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757824423546252e-05,
+      "loss": 3.2185,
+      "step": 7755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757510540752005e-05,
+      "loss": 3.0296,
+      "step": 7760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757196459733357e-05,
+      "loss": 3.1483,
+      "step": 7765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756882180503395e-05,
+      "loss": 3.1331,
+      "step": 7770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756567703075212e-05,
+      "loss": 3.1428,
+      "step": 7775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756253027461915e-05,
+      "loss": 3.0586,
+      "step": 7780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75593815367661e-05,
+      "loss": 3.0562,
+      "step": 7785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.755623081732418e-05,
+      "loss": 3.1922,
+      "step": 7790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.755307811642468e-05,
+      "loss": 3.1532,
+      "step": 7795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754992343419895e-05,
+      "loss": 3.1308,
+      "step": 7800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754676677077843e-05,
+      "loss": 3.1528,
+      "step": 7805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754360812629467e-05,
+      "loss": 3.1102,
+      "step": 7810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754044750087923e-05,
+      "loss": 3.0805,
+      "step": 7815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753728489466382e-05,
+      "loss": 3.2182,
+      "step": 7820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753412030778022e-05,
+      "loss": 3.1298,
+      "step": 7825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75309537403603e-05,
+      "loss": 3.1341,
+      "step": 7830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752778519253595e-05,
+      "loss": 3.0649,
+      "step": 7835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752461466443923e-05,
+      "loss": 3.1076,
+      "step": 7840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752144215620223e-05,
+      "loss": 3.11,
+      "step": 7845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.751826766795712e-05,
+      "loss": 3.1488,
+      "step": 7850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.751509119983618e-05,
+      "loss": 3.111,
+      "step": 7855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.751191275197176e-05,
+      "loss": 3.1678,
+      "step": 7860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75087323244963e-05,
+      "loss": 3.057,
+      "step": 7865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.75055499175423e-05,
+      "loss": 3.0621,
+      "step": 7870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.750236553124236e-05,
+      "loss": 3.0441,
+      "step": 7875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749917916572915e-05,
+      "loss": 3.1223,
+      "step": 7880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749599082113545e-05,
+      "loss": 3.1807,
+      "step": 7885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749280049759411e-05,
+      "loss": 3.0993,
+      "step": 7890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748960819523803e-05,
+      "loss": 3.0145,
+      "step": 7895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748641391420023e-05,
+      "loss": 3.171,
+      "step": 7900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748321765461382e-05,
+      "loss": 3.1874,
+      "step": 7905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748001941661194e-05,
+      "loss": 3.1567,
+      "step": 7910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747681920032785e-05,
+      "loss": 3.127,
+      "step": 7915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747361700589492e-05,
+      "loss": 3.0768,
+      "step": 7920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747041283344657e-05,
+      "loss": 3.0641,
+      "step": 7925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746720668311627e-05,
+      "loss": 3.0079,
+      "step": 7930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746399855503763e-05,
+      "loss": 3.1469,
+      "step": 7935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746078844934431e-05,
+      "loss": 3.0377,
+      "step": 7940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.745757636617005e-05,
+      "loss": 3.0256,
+      "step": 7945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.745436230564873e-05,
+      "loss": 3.1933,
+      "step": 7950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74511462679142e-05,
+      "loss": 3.099,
+      "step": 7955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74479282531005e-05,
+      "loss": 3.112,
+      "step": 7960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744470826134172e-05,
+      "loss": 3.2008,
+      "step": 7965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744148629277198e-05,
+      "loss": 3.1242,
+      "step": 7970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743826234752555e-05,
+      "loss": 3.0915,
+      "step": 7975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743503642573677e-05,
+      "loss": 3.1045,
+      "step": 7980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743180852754004e-05,
+      "loss": 3.0568,
+      "step": 7985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742857865306983e-05,
+      "loss": 3.1728,
+      "step": 7990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742534680246074e-05,
+      "loss": 3.0039,
+      "step": 7995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742211297584742e-05,
+      "loss": 3.1734,
+      "step": 8000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.741887717336459e-05,
+      "loss": 3.1151,
+      "step": 8005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.741563939514713e-05,
+      "loss": 3.1703,
+      "step": 8010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.741239964132988e-05,
+      "loss": 3.0476,
+      "step": 8015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.740915791204786e-05,
+      "loss": 3.1373,
+      "step": 8020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.740591420743612e-05,
+      "loss": 3.1507,
+      "step": 8025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.740266852762982e-05,
+      "loss": 3.0256,
+      "step": 8030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.73994208727642e-05,
+      "loss": 3.0622,
+      "step": 8035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.739617124297458e-05,
+      "loss": 3.2012,
+      "step": 8040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.739291963839632e-05,
+      "loss": 3.1474,
+      "step": 8045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.738966605916495e-05,
+      "loss": 3.0255,
+      "step": 8050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.738641050541599e-05,
+      "loss": 2.9893,
+      "step": 8055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.738315297728511e-05,
+      "loss": 3.1517,
+      "step": 8060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.737989347490804e-05,
+      "loss": 3.1304,
+      "step": 8065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.737663199842055e-05,
+      "loss": 3.1292,
+      "step": 8070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.73733685479586e-05,
+      "loss": 3.1523,
+      "step": 8075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.73701031236581e-05,
+      "loss": 3.0277,
+      "step": 8080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.736683572565513e-05,
+      "loss": 3.0757,
+      "step": 8085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.736356635408583e-05,
+      "loss": 3.0411,
+      "step": 8090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.736029500908642e-05,
+      "loss": 3.0219,
+      "step": 8095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735702169079319e-05,
+      "loss": 3.0651,
+      "step": 8100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735374639934253e-05,
+      "loss": 3.1686,
+      "step": 8105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735046913487093e-05,
+      "loss": 3.0159,
+      "step": 8110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734718989751492e-05,
+      "loss": 3.0136,
+      "step": 8115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734390868741113e-05,
+      "loss": 3.1465,
+      "step": 8120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734062550469627e-05,
+      "loss": 3.0931,
+      "step": 8125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733734034950714e-05,
+      "loss": 3.2148,
+      "step": 8130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733405322198064e-05,
+      "loss": 3.0297,
+      "step": 8135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733076412225368e-05,
+      "loss": 3.0846,
+      "step": 8140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.732747305046334e-05,
+      "loss": 3.1362,
+      "step": 8145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.732418000674675e-05,
+      "loss": 3.1285,
+      "step": 8150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73208849912411e-05,
+      "loss": 3.1584,
+      "step": 8155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.731758800408366e-05,
+      "loss": 3.1783,
+      "step": 8160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.731428904541184e-05,
+      "loss": 3.0478,
+      "step": 8165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.731098811536307e-05,
+      "loss": 3.123,
+      "step": 8170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73076852140749e-05,
+      "loss": 3.0301,
+      "step": 8175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.730438034168492e-05,
+      "loss": 3.089,
+      "step": 8180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.730107349833086e-05,
+      "loss": 3.0624,
+      "step": 8185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729776468415047e-05,
+      "loss": 3.116,
+      "step": 8190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729445389928164e-05,
+      "loss": 3.1164,
+      "step": 8195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729114114386232e-05,
+      "loss": 3.1465,
+      "step": 8200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728782641803052e-05,
+      "loss": 3.0485,
+      "step": 8205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728450972192435e-05,
+      "loss": 3.0158,
+      "step": 8210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728119105568201e-05,
+      "loss": 3.132,
+      "step": 8215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727787041944177e-05,
+      "loss": 3.1836,
+      "step": 8220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.7274547813342e-05,
+      "loss": 3.1055,
+      "step": 8225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727122323752112e-05,
+      "loss": 3.106,
+      "step": 8230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.726789669211764e-05,
+      "loss": 3.11,
+      "step": 8235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.726456817727019e-05,
+      "loss": 3.2393,
+      "step": 8240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.726123769311746e-05,
+      "loss": 3.0928,
+      "step": 8245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725790523979818e-05,
+      "loss": 3.0862,
+      "step": 8250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725457081745122e-05,
+      "loss": 2.9499,
+      "step": 8255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725123442621551e-05,
+      "loss": 3.0755,
+      "step": 8260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.724789606623006e-05,
+      "loss": 3.0437,
+      "step": 8265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.724455573763398e-05,
+      "loss": 3.2454,
+      "step": 8270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.724121344056642e-05,
+      "loss": 3.0655,
+      "step": 8275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723786917516666e-05,
+      "loss": 3.1203,
+      "step": 8280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723452294157402e-05,
+      "loss": 3.0875,
+      "step": 8285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723117473992795e-05,
+      "loss": 3.0065,
+      "step": 8290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.722782457036793e-05,
+      "loss": 3.1916,
+      "step": 8295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.722447243303356e-05,
+      "loss": 3.0467,
+      "step": 8300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72211183280645e-05,
+      "loss": 3.1495,
+      "step": 8305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721776225560052e-05,
+      "loss": 3.2251,
+      "step": 8310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721440421578143e-05,
+      "loss": 3.0682,
+      "step": 8315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721104420874717e-05,
+      "loss": 3.0991,
+      "step": 8320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72076822346377e-05,
+      "loss": 3.1851,
+      "step": 8325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.720431829359312e-05,
+      "loss": 3.1464,
+      "step": 8330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.720095238575359e-05,
+      "loss": 3.1811,
+      "step": 8335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719758451125935e-05,
+      "loss": 3.1505,
+      "step": 8340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719421467025074e-05,
+      "loss": 3.1036,
+      "step": 8345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719084286286813e-05,
+      "loss": 3.1088,
+      "step": 8350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718746908925205e-05,
+      "loss": 3.2003,
+      "step": 8355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718409334954304e-05,
+      "loss": 3.0104,
+      "step": 8360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718071564388178e-05,
+      "loss": 2.9075,
+      "step": 8365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717733597240897e-05,
+      "loss": 3.1996,
+      "step": 8370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717395433526545e-05,
+      "loss": 3.1436,
+      "step": 8375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717057073259211e-05,
+      "loss": 3.0003,
+      "step": 8380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.716718516452994e-05,
+      "loss": 3.0798,
+      "step": 8385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.716379763121996e-05,
+      "loss": 3.0509,
+      "step": 8390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.716040813280338e-05,
+      "loss": 3.1295,
+      "step": 8395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.715701666942138e-05,
+      "loss": 3.0803,
+      "step": 8400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.715362324121529e-05,
+      "loss": 3.1938,
+      "step": 8405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.715022784832647e-05,
+      "loss": 3.1098,
+      "step": 8410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714683049089641e-05,
+      "loss": 3.0906,
+      "step": 8415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714343116906667e-05,
+      "loss": 3.0223,
+      "step": 8420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714002988297887e-05,
+      "loss": 3.1205,
+      "step": 8425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.713662663277474e-05,
+      "loss": 3.0717,
+      "step": 8430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.713322141859607e-05,
+      "loss": 3.1261,
+      "step": 8435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712981424058473e-05,
+      "loss": 2.9771,
+      "step": 8440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712640509888271e-05,
+      "loss": 3.1545,
+      "step": 8445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712299399363203e-05,
+      "loss": 3.1112,
+      "step": 8450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711958092497482e-05,
+      "loss": 3.18,
+      "step": 8455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711616589305328e-05,
+      "loss": 3.1324,
+      "step": 8460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711274889800972e-05,
+      "loss": 3.1631,
+      "step": 8465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710932993998651e-05,
+      "loss": 3.2061,
+      "step": 8470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710590901912608e-05,
+      "loss": 3.1291,
+      "step": 8475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710248613557099e-05,
+      "loss": 3.0171,
+      "step": 8480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.709906128946383e-05,
+      "loss": 2.9191,
+      "step": 8485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.709563448094732e-05,
+      "loss": 3.1077,
+      "step": 8490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.709220571016425e-05,
+      "loss": 3.1577,
+      "step": 8495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.708877497725743e-05,
+      "loss": 3.1788,
+      "step": 8500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.708534228236986e-05,
+      "loss": 3.01,
+      "step": 8505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.708190762564452e-05,
+      "loss": 3.1142,
+      "step": 8510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707847100722456e-05,
+      "loss": 3.0865,
+      "step": 8515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707503242725316e-05,
+      "loss": 2.9671,
+      "step": 8520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707159188587356e-05,
+      "loss": 3.0413,
+      "step": 8525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706814938322915e-05,
+      "loss": 2.9862,
+      "step": 8530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706470491946331e-05,
+      "loss": 3.2372,
+      "step": 8535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706125849471962e-05,
+      "loss": 3.0728,
+      "step": 8540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705781010914164e-05,
+      "loss": 2.9774,
+      "step": 8545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705435976287305e-05,
+      "loss": 3.0771,
+      "step": 8550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705090745605763e-05,
+      "loss": 3.0516,
+      "step": 8555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.704745318883919e-05,
+      "loss": 3.1437,
+      "step": 8560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70439969613617e-05,
+      "loss": 3.0308,
+      "step": 8565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.704053877376913e-05,
+      "loss": 2.9736,
+      "step": 8570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703707862620556e-05,
+      "loss": 3.155,
+      "step": 8575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70336165188152e-05,
+      "loss": 3.2073,
+      "step": 8580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703015245174227e-05,
+      "loss": 2.983,
+      "step": 8585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.702668642513108e-05,
+      "loss": 3.0794,
+      "step": 8590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.702321843912612e-05,
+      "loss": 3.0414,
+      "step": 8595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.701974849387181e-05,
+      "loss": 3.044,
+      "step": 8600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.701627658951276e-05,
+      "loss": 3.1357,
+      "step": 8605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.701280272619362e-05,
+      "loss": 3.1138,
+      "step": 8610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700932690405915e-05,
+      "loss": 3.0056,
+      "step": 8615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700584912325414e-05,
+      "loss": 3.0225,
+      "step": 8620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700236938392352e-05,
+      "loss": 3.0486,
+      "step": 8625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.699888768621226e-05,
+      "loss": 3.0607,
+      "step": 8630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.699540403026544e-05,
+      "loss": 3.0681,
+      "step": 8635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69919184162282e-05,
+      "loss": 3.1566,
+      "step": 8640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.698843084424577e-05,
+      "loss": 3.0948,
+      "step": 8645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.698494131446346e-05,
+      "loss": 3.041,
+      "step": 8650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.698144982702669e-05,
+      "loss": 3.0946,
+      "step": 8655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697795638208089e-05,
+      "loss": 3.1098,
+      "step": 8660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697446097977163e-05,
+      "loss": 3.054,
+      "step": 8665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697096362024457e-05,
+      "loss": 2.9213,
+      "step": 8670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696746430364541e-05,
+      "loss": 3.0939,
+      "step": 8675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696396303011996e-05,
+      "loss": 3.0448,
+      "step": 8680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69604597998141e-05,
+      "loss": 3.069,
+      "step": 8685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.695695461287379e-05,
+      "loss": 3.0301,
+      "step": 8690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.695344746944507e-05,
+      "loss": 3.1801,
+      "step": 8695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69499383696741e-05,
+      "loss": 3.0969,
+      "step": 8700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694642731370705e-05,
+      "loss": 3.0162,
+      "step": 8705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69429143016902e-05,
+      "loss": 3.1083,
+      "step": 8710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.693939933376999e-05,
+      "loss": 3.0267,
+      "step": 8715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.693588241009281e-05,
+      "loss": 3.0722,
+      "step": 8720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.693236353080522e-05,
+      "loss": 3.2258,
+      "step": 8725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692884269605381e-05,
+      "loss": 3.1223,
+      "step": 8730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692531990598533e-05,
+      "loss": 3.0879,
+      "step": 8735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69217951607465e-05,
+      "loss": 3.1944,
+      "step": 8740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.691826846048421e-05,
+      "loss": 3.061,
+      "step": 8745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69147398053454e-05,
+      "loss": 3.1468,
+      "step": 8750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69112091954771e-05,
+      "loss": 3.0011,
+      "step": 8755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690767663102641e-05,
+      "loss": 3.148,
+      "step": 8760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690414211214052e-05,
+      "loss": 3.068,
+      "step": 8765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690060563896668e-05,
+      "loss": 3.0695,
+      "step": 8770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.689706721165225e-05,
+      "loss": 3.0573,
+      "step": 8775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.689352683034466e-05,
+      "loss": 3.0427,
+      "step": 8780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.688998449519142e-05,
+      "loss": 2.9388,
+      "step": 8785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.688644020634015e-05,
+      "loss": 3.1886,
+      "step": 8790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.688289396393848e-05,
+      "loss": 3.0232,
+      "step": 8795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.68793457681342e-05,
+      "loss": 3.168,
+      "step": 8800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.687579561907512e-05,
+      "loss": 3.0781,
+      "step": 8805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.687224351690918e-05,
+      "loss": 3.1277,
+      "step": 8810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.686868946178439e-05,
+      "loss": 3.1389,
+      "step": 8815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.686513345384882e-05,
+      "loss": 3.0873,
+      "step": 8820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.686157549325061e-05,
+      "loss": 3.0645,
+      "step": 8825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.685801558013802e-05,
+      "loss": 3.1024,
+      "step": 8830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.68544537146594e-05,
+      "loss": 3.0089,
+      "step": 8835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.685088989696312e-05,
+      "loss": 3.0579,
+      "step": 8840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.68473241271977e-05,
+      "loss": 2.9389,
+      "step": 8845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.684375640551169e-05,
+      "loss": 3.1608,
+      "step": 8850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.684018673205374e-05,
+      "loss": 2.9941,
+      "step": 8855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.68366151069726e-05,
+      "loss": 3.0974,
+      "step": 8860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.683304153041708e-05,
+      "loss": 3.0113,
+      "step": 8865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.682946600253606e-05,
+      "loss": 3.1033,
+      "step": 8870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.682588852347852e-05,
+      "loss": 3.035,
+      "step": 8875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.682230909339351e-05,
+      "loss": 3.0473,
+      "step": 8880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68187277124302e-05,
+      "loss": 3.0496,
+      "step": 8885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68151443807378e-05,
+      "loss": 3.002,
+      "step": 8890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.681155909846558e-05,
+      "loss": 3.1419,
+      "step": 8895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680797186576295e-05,
+      "loss": 3.0026,
+      "step": 8900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680438268277938e-05,
+      "loss": 3.0419,
+      "step": 8905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68007915496644e-05,
+      "loss": 3.0298,
+      "step": 8910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.679719846656764e-05,
+      "loss": 3.1236,
+      "step": 8915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.679360343363882e-05,
+      "loss": 3.1918,
+      "step": 8920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.679000645102771e-05,
+      "loss": 3.1717,
+      "step": 8925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67864075188842e-05,
+      "loss": 3.0164,
+      "step": 8930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.678280663735822e-05,
+      "loss": 3.0562,
+      "step": 8935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.677920380659981e-05,
+      "loss": 3.061,
+      "step": 8940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67755990267591e-05,
+      "loss": 3.1318,
+      "step": 8945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.677199229798628e-05,
+      "loss": 3.1634,
+      "step": 8950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676838362043161e-05,
+      "loss": 3.0123,
+      "step": 8955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676477299424545e-05,
+      "loss": 3.0182,
+      "step": 8960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676116041957826e-05,
+      "loss": 2.9559,
+      "step": 8965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675754589658053e-05,
+      "loss": 3.0943,
+      "step": 8970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675392942540288e-05,
+      "loss": 3.1389,
+      "step": 8975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6750311006196e-05,
+      "loss": 3.077,
+      "step": 8980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.674669063911064e-05,
+      "loss": 2.938,
+      "step": 8985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.674306832429763e-05,
+      "loss": 2.9988,
+      "step": 8990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673944406190794e-05,
+      "loss": 2.8766,
+      "step": 8995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673581785209254e-05,
+      "loss": 2.9333,
+      "step": 9000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673218969500251e-05,
+      "loss": 3.0978,
+      "step": 9005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.672855959078906e-05,
+      "loss": 3.1324,
+      "step": 9010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.672492753960341e-05,
+      "loss": 3.0456,
+      "step": 9015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67212935415969e-05,
+      "loss": 3.0887,
+      "step": 9020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671765759692094e-05,
+      "loss": 3.1573,
+      "step": 9025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671401970572704e-05,
+      "loss": 3.1215,
+      "step": 9030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671037986816675e-05,
+      "loss": 3.0844,
+      "step": 9035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670673808439172e-05,
+      "loss": 3.011,
+      "step": 9040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670309435455374e-05,
+      "loss": 3.1114,
+      "step": 9045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669944867880456e-05,
+      "loss": 3.1812,
+      "step": 9050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669580105729612e-05,
+      "loss": 3.1606,
+      "step": 9055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669215149018039e-05,
+      "loss": 3.0977,
+      "step": 9060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.668849997760945e-05,
+      "loss": 3.0898,
+      "step": 9065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.668484651973541e-05,
+      "loss": 3.0246,
+      "step": 9070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66811911167105e-05,
+      "loss": 3.021,
+      "step": 9075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.667753376868704e-05,
+      "loss": 3.0226,
+      "step": 9080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.667387447581742e-05,
+      "loss": 3.1839,
+      "step": 9085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.667021323825407e-05,
+      "loss": 3.0614,
+      "step": 9090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.666655005614958e-05,
+      "loss": 2.9196,
+      "step": 9095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.666288492965656e-05,
+      "loss": 3.1501,
+      "step": 9100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665921785892771e-05,
+      "loss": 2.9856,
+      "step": 9105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665554884411585e-05,
+      "loss": 3.0595,
+      "step": 9110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665187788537382e-05,
+      "loss": 3.2525,
+      "step": 9115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.664820498285458e-05,
+      "loss": 3.0507,
+      "step": 9120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.664453013671119e-05,
+      "loss": 3.0718,
+      "step": 9125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.664085334709673e-05,
+      "loss": 2.9356,
+      "step": 9130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66371746141644e-05,
+      "loss": 3.1444,
+      "step": 9135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.663349393806748e-05,
+      "loss": 3.1108,
+      "step": 9140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662981131895936e-05,
+      "loss": 3.0581,
+      "step": 9145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662612675699344e-05,
+      "loss": 3.0076,
+      "step": 9150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662244025232326e-05,
+      "loss": 3.0833,
+      "step": 9155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661875180510239e-05,
+      "loss": 3.1162,
+      "step": 9160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661506141548454e-05,
+      "loss": 3.0012,
+      "step": 9165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661136908362347e-05,
+      "loss": 3.0928,
+      "step": 9170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660767480967302e-05,
+      "loss": 2.9461,
+      "step": 9175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66039785937871e-05,
+      "loss": 3.1244,
+      "step": 9180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660028043611972e-05,
+      "loss": 3.0412,
+      "step": 9185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.659658033682499e-05,
+      "loss": 3.0223,
+      "step": 9190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.659287829605706e-05,
+      "loss": 3.0322,
+      "step": 9195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.658917431397017e-05,
+      "loss": 3.026,
+      "step": 9200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.658546839071865e-05,
+      "loss": 3.0245,
+      "step": 9205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.658176052645693e-05,
+      "loss": 3.1889,
+      "step": 9210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.657805072133948e-05,
+      "loss": 3.1692,
+      "step": 9215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.657433897552086e-05,
+      "loss": 3.0661,
+      "step": 9220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.657062528915576e-05,
+      "loss": 3.009,
+      "step": 9225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.656690966239887e-05,
+      "loss": 3.1478,
+      "step": 9230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.656319209540504e-05,
+      "loss": 3.0103,
+      "step": 9235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655947258832916e-05,
+      "loss": 2.9966,
+      "step": 9240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655575114132618e-05,
+      "loss": 2.9566,
+      "step": 9245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655202775455117e-05,
+      "loss": 3.0863,
+      "step": 9250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.654830242815927e-05,
+      "loss": 2.892,
+      "step": 9255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.654457516230568e-05,
+      "loss": 3.0263,
+      "step": 9260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.654084595714575e-05,
+      "loss": 3.0518,
+      "step": 9265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.65371148128348e-05,
+      "loss": 3.1444,
+      "step": 9270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.653338172952831e-05,
+      "loss": 3.1013,
+      "step": 9275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.652964670738184e-05,
+      "loss": 3.0417,
+      "step": 9280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6525909746551e-05,
+      "loss": 3.1456,
+      "step": 9285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.652217084719148e-05,
+      "loss": 3.0906,
+      "step": 9290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651843000945907e-05,
+      "loss": 3.1403,
+      "step": 9295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651468723350965e-05,
+      "loss": 3.0753,
+      "step": 9300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651094251949914e-05,
+      "loss": 3.0271,
+      "step": 9305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.650719586758359e-05,
+      "loss": 3.1187,
+      "step": 9310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.650344727791908e-05,
+      "loss": 3.031,
+      "step": 9315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649969675066182e-05,
+      "loss": 3.0688,
+      "step": 9320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649594428596807e-05,
+      "loss": 2.9957,
+      "step": 9325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649218988399417e-05,
+      "loss": 2.9746,
+      "step": 9330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.648843354489656e-05,
+      "loss": 2.942,
+      "step": 9335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.648467526883176e-05,
+      "loss": 3.1022,
+      "step": 9340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.648091505595634e-05,
+      "loss": 3.0181,
+      "step": 9345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.647715290642697e-05,
+      "loss": 3.0359,
+      "step": 9350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.647338882040042e-05,
+      "loss": 3.0419,
+      "step": 9355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646962279803351e-05,
+      "loss": 2.9927,
+      "step": 9360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646585483948315e-05,
+      "loss": 3.0558,
+      "step": 9365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646208494490636e-05,
+      "loss": 3.0578,
+      "step": 9370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64583131144602e-05,
+      "loss": 3.1693,
+      "step": 9375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.645453934830181e-05,
+      "loss": 3.1727,
+      "step": 9380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.645076364658846e-05,
+      "loss": 3.1086,
+      "step": 9385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.644698600947741e-05,
+      "loss": 3.0099,
+      "step": 9390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64432064371261e-05,
+      "loss": 3.0451,
+      "step": 9395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643942492969201e-05,
+      "loss": 3.0464,
+      "step": 9400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643564148733269e-05,
+      "loss": 3.1135,
+      "step": 9405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643185611020577e-05,
+      "loss": 3.1644,
+      "step": 9410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.642806879846899e-05,
+      "loss": 3.0,
+      "step": 9415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.642427955228014e-05,
+      "loss": 3.084,
+      "step": 9420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.642048837179709e-05,
+      "loss": 2.9501,
+      "step": 9425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.641669525717782e-05,
+      "loss": 3.0017,
+      "step": 9430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.641290020858037e-05,
+      "loss": 3.0272,
+      "step": 9435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.640910322616284e-05,
+      "loss": 3.087,
+      "step": 9440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.640530431008345e-05,
+      "loss": 3.0558,
+      "step": 9445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64015034605005e-05,
+      "loss": 3.0535,
+      "step": 9450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639770067757232e-05,
+      "loss": 3.0309,
+      "step": 9455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63938959614574e-05,
+      "loss": 3.1304,
+      "step": 9460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639008931231421e-05,
+      "loss": 3.0171,
+      "step": 9465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63862807303014e-05,
+      "loss": 3.074,
+      "step": 9470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.638247021557764e-05,
+      "loss": 3.0353,
+      "step": 9475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637865776830172e-05,
+      "loss": 3.0715,
+      "step": 9480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637484338863245e-05,
+      "loss": 3.1232,
+      "step": 9485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637102707672879e-05,
+      "loss": 3.0652,
+      "step": 9490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63672088327497e-05,
+      "loss": 3.1041,
+      "step": 9495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.636338865685434e-05,
+      "loss": 2.9954,
+      "step": 9500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635956654920182e-05,
+      "loss": 3.1519,
+      "step": 9505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635574250995143e-05,
+      "loss": 3.0118,
+      "step": 9510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635191653926248e-05,
+      "loss": 3.2593,
+      "step": 9515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.634808863729438e-05,
+      "loss": 3.1208,
+      "step": 9520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.634425880420665e-05,
+      "loss": 3.0196,
+      "step": 9525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.634042704015883e-05,
+      "loss": 3.0733,
+      "step": 9530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.633659334531057e-05,
+      "loss": 3.0407,
+      "step": 9535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.633275771982162e-05,
+      "loss": 3.2746,
+      "step": 9540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63289201638518e-05,
+      "loss": 3.1094,
+      "step": 9545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.632508067756098e-05,
+      "loss": 2.9824,
+      "step": 9550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.632123926110916e-05,
+      "loss": 2.9984,
+      "step": 9555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.631739591465639e-05,
+      "loss": 2.9264,
+      "step": 9560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.631355063836278e-05,
+      "loss": 3.1238,
+      "step": 9565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.630970343238857e-05,
+      "loss": 3.0339,
+      "step": 9570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.630585429689404e-05,
+      "loss": 2.9693,
+      "step": 9575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63020032320396e-05,
+      "loss": 2.9192,
+      "step": 9580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.629815023798567e-05,
+      "loss": 3.1488,
+      "step": 9585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.62942953148928e-05,
+      "loss": 3.2314,
+      "step": 9590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.629043846292161e-05,
+      "loss": 3.0222,
+      "step": 9595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.628657968223278e-05,
+      "loss": 3.0851,
+      "step": 9600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.628271897298713e-05,
+      "loss": 3.0279,
+      "step": 9605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.627885633534548e-05,
+      "loss": 2.9932,
+      "step": 9610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.627499176946878e-05,
+      "loss": 3.0784,
+      "step": 9615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.627112527551805e-05,
+      "loss": 2.9275,
+      "step": 9620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.626725685365439e-05,
+      "loss": 3.0516,
+      "step": 9625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.626338650403899e-05,
+      "loss": 2.8633,
+      "step": 9630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.625951422683309e-05,
+      "loss": 3.0118,
+      "step": 9635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.625564002219805e-05,
+      "loss": 3.0229,
+      "step": 9640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.625176389029528e-05,
+      "loss": 3.1926,
+      "step": 9645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.624788583128628e-05,
+      "loss": 3.1251,
+      "step": 9650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.624400584533264e-05,
+      "loss": 2.9993,
+      "step": 9655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.624012393259601e-05,
+      "loss": 2.928,
+      "step": 9660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623624009323816e-05,
+      "loss": 2.8959,
+      "step": 9665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623235432742088e-05,
+      "loss": 3.0474,
+      "step": 9670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.622846663530607e-05,
+      "loss": 3.0743,
+      "step": 9675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.622457701705574e-05,
+      "loss": 3.1771,
+      "step": 9680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.622068547283194e-05,
+      "loss": 3.0046,
+      "step": 9685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.621679200279682e-05,
+      "loss": 3.1735,
+      "step": 9690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.621289660711259e-05,
+      "loss": 2.9677,
+      "step": 9695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620899928594156e-05,
+      "loss": 3.0956,
+      "step": 9700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620510003944612e-05,
+      "loss": 3.0151,
+      "step": 9705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620119886778873e-05,
+      "loss": 3.1317,
+      "step": 9710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.619729577113194e-05,
+      "loss": 2.9835,
+      "step": 9715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.619339074963835e-05,
+      "loss": 2.9828,
+      "step": 9720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61894838034707e-05,
+      "loss": 3.1507,
+      "step": 9725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.618557493279177e-05,
+      "loss": 3.1033,
+      "step": 9730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.618166413776439e-05,
+      "loss": 2.9334,
+      "step": 9735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.617775141855154e-05,
+      "loss": 2.8825,
+      "step": 9740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.617383677531624e-05,
+      "loss": 3.0824,
+      "step": 9745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61699202082216e-05,
+      "loss": 3.0675,
+      "step": 9750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.616600171743078e-05,
+      "loss": 2.962,
+      "step": 9755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.616208130310706e-05,
+      "loss": 3.0103,
+      "step": 9760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61581589654138e-05,
+      "loss": 3.0173,
+      "step": 9765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615423470451444e-05,
+      "loss": 3.0313,
+      "step": 9770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615030852057243e-05,
+      "loss": 3.1718,
+      "step": 9775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61463804137514e-05,
+      "loss": 2.9381,
+      "step": 9780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.6142450384215e-05,
+      "loss": 2.9305,
+      "step": 9785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.6138518432127e-05,
+      "loss": 2.9251,
+      "step": 9790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.613458455765121e-05,
+      "loss": 2.968,
+      "step": 9795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.613064876095153e-05,
+      "loss": 3.0544,
+      "step": 9800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612671104219196e-05,
+      "loss": 3.1497,
+      "step": 9805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612277140153655e-05,
+      "loss": 3.1843,
+      "step": 9810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611882983914945e-05,
+      "loss": 3.0463,
+      "step": 9815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611488635519492e-05,
+      "loss": 3.1118,
+      "step": 9820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611094094983725e-05,
+      "loss": 3.1061,
+      "step": 9825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61069936232408e-05,
+      "loss": 3.1507,
+      "step": 9830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.610304437557008e-05,
+      "loss": 2.9632,
+      "step": 9835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.609909320698961e-05,
+      "loss": 2.932,
+      "step": 9840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.609514011766403e-05,
+      "loss": 3.0069,
+      "step": 9845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.609118510775805e-05,
+      "loss": 2.9971,
+      "step": 9850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608722817743642e-05,
+      "loss": 3.0617,
+      "step": 9855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608326932686407e-05,
+      "loss": 3.2009,
+      "step": 9860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.607930855620591e-05,
+      "loss": 3.0494,
+      "step": 9865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.607534586562697e-05,
+      "loss": 2.978,
+      "step": 9870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.607138125529236e-05,
+      "loss": 3.0345,
+      "step": 9875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.606741472536726e-05,
+      "loss": 3.0238,
+      "step": 9880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.606344627601696e-05,
+      "loss": 3.0813,
+      "step": 9885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60594759074068e-05,
+      "loss": 3.0183,
+      "step": 9890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.605550361970219e-05,
+      "loss": 2.9841,
+      "step": 9895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.605152941306865e-05,
+      "loss": 3.0191,
+      "step": 9900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.604755328767178e-05,
+      "loss": 3.0207,
+      "step": 9905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.604357524367722e-05,
+      "loss": 3.1091,
+      "step": 9910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.603959528125075e-05,
+      "loss": 2.9516,
+      "step": 9915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.603561340055818e-05,
+      "loss": 2.9397,
+      "step": 9920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60316296017654e-05,
+      "loss": 3.1102,
+      "step": 9925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.602764388503843e-05,
+      "loss": 2.9715,
+      "step": 9930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.602365625054333e-05,
+      "loss": 2.9806,
+      "step": 9935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.601966669844622e-05,
+      "loss": 3.0743,
+      "step": 9940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.601567522891335e-05,
+      "loss": 3.023,
+      "step": 9945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.601168184211102e-05,
+      "loss": 3.1373,
+      "step": 9950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.600768653820564e-05,
+      "loss": 3.0451,
+      "step": 9955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.600368931736364e-05,
+      "loss": 3.1844,
+      "step": 9960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59996901797516e-05,
+      "loss": 2.9124,
+      "step": 9965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.599568912553611e-05,
+      "loss": 3.0334,
+      "step": 9970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59916861548839e-05,
+      "loss": 3.0976,
+      "step": 9975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.598768126796176e-05,
+      "loss": 2.9482,
+      "step": 9980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.598367446493653e-05,
+      "loss": 2.9235,
+      "step": 9985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.597966574597519e-05,
+      "loss": 3.1066,
+      "step": 9990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.597565511124474e-05,
+      "loss": 3.0373,
+      "step": 9995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59716425609123e-05,
+      "loss": 3.0391,
+      "step": 10000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.596762809514505e-05,
+      "loss": 3.0411,
+      "step": 10005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.596361171411026e-05,
+      "loss": 2.9186,
+      "step": 10010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595959341797525e-05,
+      "loss": 2.9814,
+      "step": 10015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595557320690748e-05,
+      "loss": 3.0184,
+      "step": 10020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595155108107443e-05,
+      "loss": 3.0501,
+      "step": 10025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.594752704064369e-05,
+      "loss": 3.1321,
+      "step": 10030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.594350108578293e-05,
+      "loss": 2.8987,
+      "step": 10035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59394732166599e-05,
+      "loss": 3.0516,
+      "step": 10040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.593544343344241e-05,
+      "loss": 3.056,
+      "step": 10045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.593141173629836e-05,
+      "loss": 3.0088,
+      "step": 10050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.592737812539575e-05,
+      "loss": 3.024,
+      "step": 10055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.592334260090263e-05,
+      "loss": 2.9656,
+      "step": 10060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.591930516298714e-05,
+      "loss": 2.9727,
+      "step": 10065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.591526581181751e-05,
+      "loss": 2.9837,
+      "step": 10070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.591122454756203e-05,
+      "loss": 2.9519,
+      "step": 10075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59071813703891e-05,
+      "loss": 3.1327,
+      "step": 10080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.590313628046717e-05,
+      "loss": 3.0469,
+      "step": 10085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.589908927796478e-05,
+      "loss": 2.9852,
+      "step": 10090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.589504036305056e-05,
+      "loss": 2.9849,
+      "step": 10095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58909895358932e-05,
+      "loss": 3.0088,
+      "step": 10100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.588693679666148e-05,
+      "loss": 3.0251,
+      "step": 10105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.588288214552427e-05,
+      "loss": 3.1727,
+      "step": 10110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58788255826505e-05,
+      "loss": 3.0288,
+      "step": 10115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58747671082092e-05,
+      "loss": 2.9235,
+      "step": 10120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.587070672236944e-05,
+      "loss": 3.1028,
+      "step": 10125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.586664442530044e-05,
+      "loss": 3.1047,
+      "step": 10130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.586258021717143e-05,
+      "loss": 3.012,
+      "step": 10135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.585851409815175e-05,
+      "loss": 3.0674,
+      "step": 10140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.585444606841083e-05,
+      "loss": 2.9766,
+      "step": 10145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.585037612811815e-05,
+      "loss": 3.1404,
+      "step": 10150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58463042774433e-05,
+      "loss": 3.1025,
+      "step": 10155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.584223051655594e-05,
+      "loss": 2.9923,
+      "step": 10160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583815484562578e-05,
+      "loss": 2.9661,
+      "step": 10165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583407726482265e-05,
+      "loss": 3.04,
+      "step": 10170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.582999777431648e-05,
+      "loss": 3.0588,
+      "step": 10175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.582591637427717e-05,
+      "loss": 2.9247,
+      "step": 10180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.582183306487482e-05,
+      "loss": 3.0768,
+      "step": 10185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.581774784627957e-05,
+      "loss": 3.0121,
+      "step": 10190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.581366071866163e-05,
+      "loss": 2.9903,
+      "step": 10195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580957168219127e-05,
+      "loss": 3.0572,
+      "step": 10200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580548073703888e-05,
+      "loss": 3.0256,
+      "step": 10205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580138788337489e-05,
+      "loss": 3.0382,
+      "step": 10210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.579729312136987e-05,
+      "loss": 3.0479,
+      "step": 10215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.57931964511944e-05,
+      "loss": 2.9969,
+      "step": 10220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.578909787301918e-05,
+      "loss": 3.0698,
+      "step": 10225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.578499738701498e-05,
+      "loss": 2.9617,
+      "step": 10230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.578089499335266e-05,
+      "loss": 3.082,
+      "step": 10235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.577679069220312e-05,
+      "loss": 3.0271,
+      "step": 10240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.577268448373738e-05,
+      "loss": 2.9909,
+      "step": 10245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.576857636812653e-05,
+      "loss": 3.0619,
+      "step": 10250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.576446634554175e-05,
+      "loss": 2.9944,
+      "step": 10255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.576035441615428e-05,
+      "loss": 2.9325,
+      "step": 10260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.575624058013544e-05,
+      "loss": 2.9523,
+      "step": 10265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.575212483765662e-05,
+      "loss": 2.9817,
+      "step": 10270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.574800718888933e-05,
+      "loss": 3.0532,
+      "step": 10275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.574388763400512e-05,
+      "loss": 3.1156,
+      "step": 10280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573976617317564e-05,
+      "loss": 3.0461,
+      "step": 10285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573564280657264e-05,
+      "loss": 3.0491,
+      "step": 10290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573151753436787e-05,
+      "loss": 3.1158,
+      "step": 10295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.572739035673323e-05,
+      "loss": 2.9911,
+      "step": 10300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.572326127384069e-05,
+      "loss": 2.9767,
+      "step": 10305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.571913028586231e-05,
+      "loss": 3.0019,
+      "step": 10310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.571499739297016e-05,
+      "loss": 3.0471,
+      "step": 10315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.571086259533649e-05,
+      "loss": 3.179,
+      "step": 10320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.570672589313353e-05,
+      "loss": 3.0684,
+      "step": 10325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.570258728653367e-05,
+      "loss": 3.1917,
+      "step": 10330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.569844677570936e-05,
+      "loss": 2.9524,
+      "step": 10335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.569430436083308e-05,
+      "loss": 2.8908,
+      "step": 10340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.569016004207746e-05,
+      "loss": 3.0154,
+      "step": 10345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.568601381961515e-05,
+      "loss": 3.0216,
+      "step": 10350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.568186569361892e-05,
+      "loss": 2.9785,
+      "step": 10355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.56777156642616e-05,
+      "loss": 3.1033,
+      "step": 10360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.56735637317161e-05,
+      "loss": 3.0115,
+      "step": 10365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.566940989615541e-05,
+      "loss": 3.0761,
+      "step": 10370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.566525415775261e-05,
+      "loss": 2.7643,
+      "step": 10375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.566109651668084e-05,
+      "loss": 2.9495,
+      "step": 10380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.565693697311335e-05,
+      "loss": 2.9922,
+      "step": 10385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.565277552722344e-05,
+      "loss": 3.0491,
+      "step": 10390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.564861217918448e-05,
+      "loss": 2.949,
+      "step": 10395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.564444692916998e-05,
+      "loss": 3.0129,
+      "step": 10400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.564027977735344e-05,
+      "loss": 2.9872,
+      "step": 10405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.563611072390852e-05,
+      "loss": 3.0959,
+      "step": 10410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.563193976900892e-05,
+      "loss": 2.9236,
+      "step": 10415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56277669128284e-05,
+      "loss": 2.9829,
+      "step": 10420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.562359215554086e-05,
+      "loss": 3.0917,
+      "step": 10425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.561941549732023e-05,
+      "loss": 2.9864,
+      "step": 10430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.561523693834054e-05,
+      "loss": 3.0164,
+      "step": 10435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.561105647877587e-05,
+      "loss": 3.0873,
+      "step": 10440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.560687411880043e-05,
+      "loss": 3.0087,
+      "step": 10445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.560268985858845e-05,
+      "loss": 2.9056,
+      "step": 10450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559850369831429e-05,
+      "loss": 2.9397,
+      "step": 10455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559431563815237e-05,
+      "loss": 2.9321,
+      "step": 10460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559012567827716e-05,
+      "loss": 2.9997,
+      "step": 10465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.558593381886329e-05,
+      "loss": 2.9446,
+      "step": 10470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.558174006008536e-05,
+      "loss": 3.1028,
+      "step": 10475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557754440211815e-05,
+      "loss": 3.0152,
+      "step": 10480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557334684513643e-05,
+      "loss": 3.0475,
+      "step": 10485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.556914738931515e-05,
+      "loss": 3.1285,
+      "step": 10490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.556494603482925e-05,
+      "loss": 3.0335,
+      "step": 10495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.556074278185375e-05,
+      "loss": 2.9184,
+      "step": 10500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.555653763056383e-05,
+      "loss": 2.973,
+      "step": 10505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.555233058113469e-05,
+      "loss": 2.8948,
+      "step": 10510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.55481216337416e-05,
+      "loss": 3.0063,
+      "step": 10515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.554391078855996e-05,
+      "loss": 3.0058,
+      "step": 10520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.553969804576519e-05,
+      "loss": 3.0168,
+      "step": 10525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.553548340553282e-05,
+      "loss": 3.0156,
+      "step": 10530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.553126686803847e-05,
+      "loss": 2.98,
+      "step": 10535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.552704843345781e-05,
+      "loss": 2.9482,
+      "step": 10540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.552282810196662e-05,
+      "loss": 3.0166,
+      "step": 10545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.55186058737407e-05,
+      "loss": 3.0491,
+      "step": 10550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.551438174895604e-05,
+      "loss": 3.0329,
+      "step": 10555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.551015572778859e-05,
+      "loss": 3.0222,
+      "step": 10560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.550592781041444e-05,
+      "loss": 3.0374,
+      "step": 10565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.550169799700974e-05,
+      "loss": 2.8931,
+      "step": 10570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.549746628775075e-05,
+      "loss": 2.9774,
+      "step": 10575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.549323268281377e-05,
+      "loss": 2.9789,
+      "step": 10580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548899718237521e-05,
+      "loss": 3.194,
+      "step": 10585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548475978661154e-05,
+      "loss": 2.8937,
+      "step": 10590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548052049569929e-05,
+      "loss": 2.9664,
+      "step": 10595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.547627930981512e-05,
+      "loss": 2.9839,
+      "step": 10600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.547203622913574e-05,
+      "loss": 2.9792,
+      "step": 10605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.546779125383791e-05,
+      "loss": 3.1818,
+      "step": 10610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.546354438409854e-05,
+      "loss": 3.0574,
+      "step": 10615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545929562009457e-05,
+      "loss": 2.9793,
+      "step": 10620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545504496200302e-05,
+      "loss": 3.1335,
+      "step": 10625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545079241000098e-05,
+      "loss": 3.0083,
+      "step": 10630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.544653796426564e-05,
+      "loss": 3.0137,
+      "step": 10635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.544228162497429e-05,
+      "loss": 3.0155,
+      "step": 10640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.543802339230425e-05,
+      "loss": 2.959,
+      "step": 10645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.543376326643296e-05,
+      "loss": 2.9026,
+      "step": 10650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.542950124753789e-05,
+      "loss": 3.0512,
+      "step": 10655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.542523733579665e-05,
+      "loss": 2.9118,
+      "step": 10660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.542097153138688e-05,
+      "loss": 3.1191,
+      "step": 10665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.541670383448634e-05,
+      "loss": 2.9705,
+      "step": 10670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.54124342452728e-05,
+      "loss": 2.9174,
+      "step": 10675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.54081627639242e-05,
+      "loss": 2.8306,
+      "step": 10680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.54038893906185e-05,
+      "loss": 2.8426,
+      "step": 10685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.539961412553375e-05,
+      "loss": 2.7915,
+      "step": 10690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.539533696884808e-05,
+      "loss": 2.9979,
+      "step": 10695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53910579207397e-05,
+      "loss": 3.0691,
+      "step": 10700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.538677698138689e-05,
+      "loss": 2.8454,
+      "step": 10705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.538249415096804e-05,
+      "loss": 3.0519,
+      "step": 10710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.537820942966161e-05,
+      "loss": 2.913,
+      "step": 10715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.537392281764605e-05,
+      "loss": 2.9987,
+      "step": 10720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.536963431510005e-05,
+      "loss": 2.9949,
+      "step": 10725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.536534392220226e-05,
+      "loss": 2.9874,
+      "step": 10730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.536105163913143e-05,
+      "loss": 2.9548,
+      "step": 10735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53567574660664e-05,
+      "loss": 2.9539,
+      "step": 10740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.535246140318612e-05,
+      "loss": 2.8845,
+      "step": 10745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.534816345066955e-05,
+      "loss": 3.0482,
+      "step": 10750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.534386360869578e-05,
+      "loss": 2.9973,
+      "step": 10755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.533956187744399e-05,
+      "loss": 3.0684,
+      "step": 10760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.533525825709337e-05,
+      "loss": 2.894,
+      "step": 10765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.533095274782326e-05,
+      "loss": 3.0566,
+      "step": 10770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.532664534981306e-05,
+      "loss": 3.0271,
+      "step": 10775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.532233606324222e-05,
+      "loss": 2.9799,
+      "step": 10780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53180248882903e-05,
+      "loss": 2.9081,
+      "step": 10785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.531371182513693e-05,
+      "loss": 3.0741,
+      "step": 10790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53093968739618e-05,
+      "loss": 3.1121,
+      "step": 10795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.530508003494472e-05,
+      "loss": 3.0762,
+      "step": 10800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.53007613082655e-05,
+      "loss": 3.0669,
+      "step": 10805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.529644069410416e-05,
+      "loss": 3.0094,
+      "step": 10810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.529211819264067e-05,
+      "loss": 2.9362,
+      "step": 10815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.528779380405515e-05,
+      "loss": 2.937,
+      "step": 10820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.528346752852775e-05,
+      "loss": 2.9323,
+      "step": 10825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.527913936623876e-05,
+      "loss": 3.0706,
+      "step": 10830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52748093173685e-05,
+      "loss": 3.1589,
+      "step": 10835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.527047738209738e-05,
+      "loss": 3.1129,
+      "step": 10840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52661435606059e-05,
+      "loss": 2.9398,
+      "step": 10845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.526180785307463e-05,
+      "loss": 3.1795,
+      "step": 10850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.525747025968423e-05,
+      "loss": 3.0604,
+      "step": 10855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.525313078061541e-05,
+      "loss": 2.9888,
+      "step": 10860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.524878941604898e-05,
+      "loss": 3.0682,
+      "step": 10865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.524444616616583e-05,
+      "loss": 3.069,
+      "step": 10870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.524010103114694e-05,
+      "loss": 2.9918,
+      "step": 10875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52357540111733e-05,
+      "loss": 3.0054,
+      "step": 10880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52314051064261e-05,
+      "loss": 2.8349,
+      "step": 10885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.522705431708649e-05,
+      "loss": 3.0574,
+      "step": 10890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.522270164333578e-05,
+      "loss": 2.9515,
+      "step": 10895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.521834708535531e-05,
+      "loss": 3.0448,
+      "step": 10900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52139906433265e-05,
+      "loss": 3.1005,
+      "step": 10905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52096323174309e-05,
+      "loss": 2.8489,
+      "step": 10910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.520527210785007e-05,
+      "loss": 3.0906,
+      "step": 10915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.52009100147657e-05,
+      "loss": 2.9085,
+      "step": 10920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.519654603835952e-05,
+      "loss": 3.0881,
+      "step": 10925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.519218017881338e-05,
+      "loss": 3.0413,
+      "step": 10930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.518781243630917e-05,
+      "loss": 3.0533,
+      "step": 10935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.518344281102889e-05,
+      "loss": 2.9912,
+      "step": 10940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.517907130315459e-05,
+      "loss": 3.0114,
+      "step": 10945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.517469791286841e-05,
+      "loss": 3.1392,
+      "step": 10950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.517032264035257e-05,
+      "loss": 3.0184,
+      "step": 10955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51659454857894e-05,
+      "loss": 2.9913,
+      "step": 10960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.516156644936122e-05,
+      "loss": 2.9444,
+      "step": 10965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.515718553125053e-05,
+      "loss": 3.0167,
+      "step": 10970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.515280273163984e-05,
+      "loss": 3.0302,
+      "step": 10975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.514841805071177e-05,
+      "loss": 3.0433,
+      "step": 10980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5144031488649e-05,
+      "loss": 3.0435,
+      "step": 10985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.513964304563434e-05,
+      "loss": 2.979,
+      "step": 10990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.513525272185057e-05,
+      "loss": 2.9131,
+      "step": 10995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.513086051748068e-05,
+      "loss": 3.1787,
+      "step": 11000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.512646643270764e-05,
+      "loss": 3.0343,
+      "step": 11005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.512207046771452e-05,
+      "loss": 2.9334,
+      "step": 11010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.511767262268452e-05,
+      "loss": 2.9565,
+      "step": 11015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.511327289780083e-05,
+      "loss": 3.0382,
+      "step": 11020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.510887129324682e-05,
+      "loss": 2.9975,
+      "step": 11025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.510446780920583e-05,
+      "loss": 2.9654,
+      "step": 11030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.510006244586138e-05,
+      "loss": 3.0411,
+      "step": 11035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5095655203397e-05,
+      "loss": 3.0479,
+      "step": 11040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.509124608199633e-05,
+      "loss": 3.1457,
+      "step": 11045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.508683508184306e-05,
+      "loss": 3.0963,
+      "step": 11050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5082422203121e-05,
+      "loss": 2.983,
+      "step": 11055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5078007446014e-05,
+      "loss": 3.1513,
+      "step": 11060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.507359081070601e-05,
+      "loss": 3.0343,
+      "step": 11065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.506917229738105e-05,
+      "loss": 2.9824,
+      "step": 11070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.506475190622322e-05,
+      "loss": 3.0451,
+      "step": 11075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.506032963741672e-05,
+      "loss": 2.9432,
+      "step": 11080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.505590549114576e-05,
+      "loss": 3.1253,
+      "step": 11085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.505147946759469e-05,
+      "loss": 2.9927,
+      "step": 11090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.504705156694794e-05,
+      "loss": 3.0106,
+      "step": 11095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.504262178939001e-05,
+      "loss": 3.0498,
+      "step": 11100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.503819013510542e-05,
+      "loss": 2.9814,
+      "step": 11105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.503375660427887e-05,
+      "loss": 3.0722,
+      "step": 11110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.502932119709507e-05,
+      "loss": 2.9499,
+      "step": 11115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.502488391373881e-05,
+      "loss": 2.9925,
+      "step": 11120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.502044475439498e-05,
+      "loss": 2.8923,
+      "step": 11125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.501600371924854e-05,
+      "loss": 3.046,
+      "step": 11130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.501156080848453e-05,
+      "loss": 3.0347,
+      "step": 11135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.500711602228807e-05,
+      "loss": 2.9686,
+      "step": 11140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.500266936084435e-05,
+      "loss": 2.9177,
+      "step": 11145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.499822082433863e-05,
+      "loss": 3.0285,
+      "step": 11150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.49937704129563e-05,
+      "loss": 3.0405,
+      "step": 11155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.498931812688274e-05,
+      "loss": 2.9351,
+      "step": 11160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.498486396630349e-05,
+      "loss": 2.8521,
+      "step": 11165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.498040793140411e-05,
+      "loss": 2.9727,
+      "step": 11170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.497595002237029e-05,
+      "loss": 3.0528,
+      "step": 11175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.497149023938776e-05,
+      "loss": 2.9748,
+      "step": 11180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.496702858264233e-05,
+      "loss": 3.0351,
+      "step": 11185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.49625650523199e-05,
+      "loss": 3.065,
+      "step": 11190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.495809964860649e-05,
+      "loss": 2.989,
+      "step": 11195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.495363237168808e-05,
+      "loss": 3.064,
+      "step": 11200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.494916322175084e-05,
+      "loss": 3.1504,
+      "step": 11205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.494469219898099e-05,
+      "loss": 2.9174,
+      "step": 11210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.494021930356479e-05,
+      "loss": 2.9734,
+      "step": 11215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.493574453568863e-05,
+      "loss": 3.1219,
+      "step": 11220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.493126789553894e-05,
+      "loss": 2.9903,
+      "step": 11225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.492678938330225e-05,
+      "loss": 3.1443,
+      "step": 11230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.492230899916515e-05,
+      "loss": 3.0026,
+      "step": 11235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.491782674331432e-05,
+      "loss": 2.9489,
+      "step": 11240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.491334261593653e-05,
+      "loss": 3.0312,
+      "step": 11245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.49088566172186e-05,
+      "loss": 3.0144,
+      "step": 11250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.490436874734743e-05,
+      "loss": 3.039,
+      "step": 11255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.489987900651005e-05,
+      "loss": 3.1102,
+      "step": 11260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.489538739489347e-05,
+      "loss": 3.039,
+      "step": 11265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.489089391268489e-05,
+      "loss": 2.9264,
+      "step": 11270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.488639856007151e-05,
+      "loss": 2.9919,
+      "step": 11275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.488190133724062e-05,
+      "loss": 3.0517,
+      "step": 11280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.487740224437962e-05,
+      "loss": 2.9167,
+      "step": 11285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.487290128167597e-05,
+      "loss": 3.0228,
+      "step": 11290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.486839844931717e-05,
+      "loss": 3.0336,
+      "step": 11295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.486389374749087e-05,
+      "loss": 2.9314,
+      "step": 11300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485938717638475e-05,
+      "loss": 2.9561,
+      "step": 11305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485487873618657e-05,
+      "loss": 2.9721,
+      "step": 11310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485036842708418e-05,
+      "loss": 2.9345,
+      "step": 11315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.484585624926553e-05,
+      "loss": 3.1496,
+      "step": 11320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.484134220291859e-05,
+      "loss": 3.0211,
+      "step": 11325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.483682628823145e-05,
+      "loss": 3.1103,
+      "step": 11330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.483230850539227e-05,
+      "loss": 3.151,
+      "step": 11335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.482778885458927e-05,
+      "loss": 2.9801,
+      "step": 11340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.482326733601079e-05,
+      "loss": 3.0672,
+      "step": 11345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.481874394984522e-05,
+      "loss": 2.8812,
+      "step": 11350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.481421869628102e-05,
+      "loss": 2.9604,
+      "step": 11355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.480969157550672e-05,
+      "loss": 3.0995,
+      "step": 11360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.480516258771097e-05,
+      "loss": 2.9095,
+      "step": 11365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.480063173308247e-05,
+      "loss": 3.076,
+      "step": 11370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.479609901181e-05,
+      "loss": 2.9421,
+      "step": 11375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.479156442408242e-05,
+      "loss": 3.0179,
+      "step": 11380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.478702797008863e-05,
+      "loss": 3.0073,
+      "step": 11385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.47824896500177e-05,
+      "loss": 2.9612,
+      "step": 11390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.477794946405869e-05,
+      "loss": 2.9618,
+      "step": 11395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.477340741240078e-05,
+      "loss": 3.1407,
+      "step": 11400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476886349523321e-05,
+      "loss": 3.0269,
+      "step": 11405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476431771274532e-05,
+      "loss": 2.9365,
+      "step": 11410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.475977006512648e-05,
+      "loss": 2.9588,
+      "step": 11415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.475522055256621e-05,
+      "loss": 2.9613,
+      "step": 11420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.475066917525404e-05,
+      "loss": 2.9819,
+      "step": 11425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.474611593337961e-05,
+      "loss": 2.9758,
+      "step": 11430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.474156082713264e-05,
+      "loss": 3.0576,
+      "step": 11435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.473700385670292e-05,
+      "loss": 3.0425,
+      "step": 11440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.473244502228031e-05,
+      "loss": 2.9972,
+      "step": 11445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.472788432405478e-05,
+      "loss": 2.9148,
+      "step": 11450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.472332176221634e-05,
+      "loss": 3.0469,
+      "step": 11455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.471875733695508e-05,
+      "loss": 2.9497,
+      "step": 11460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.471419104846119e-05,
+      "loss": 2.9041,
+      "step": 11465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.470962289692493e-05,
+      "loss": 3.1136,
+      "step": 11470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.470505288253664e-05,
+      "loss": 2.9411,
+      "step": 11475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.470048100548671e-05,
+      "loss": 2.9982,
+      "step": 11480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.469590726596564e-05,
+      "loss": 2.9989,
+      "step": 11485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.469133166416402e-05,
+      "loss": 2.8732,
+      "step": 11490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.468675420027246e-05,
+      "loss": 3.0765,
+      "step": 11495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.46821748744817e-05,
+      "loss": 2.8651,
+      "step": 11500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.467759368698255e-05,
+      "loss": 2.9841,
+      "step": 11505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.467301063796587e-05,
+      "loss": 3.0367,
+      "step": 11510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.466842572762262e-05,
+      "loss": 3.0966,
+      "step": 11515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.466383895614384e-05,
+      "loss": 2.978,
+      "step": 11520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465925032372063e-05,
+      "loss": 2.9506,
+      "step": 11525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465465983054419e-05,
+      "loss": 2.9547,
+      "step": 11530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.465006747680577e-05,
+      "loss": 2.9946,
+      "step": 11535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.464547326269673e-05,
+      "loss": 2.9875,
+      "step": 11540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.464087718840849e-05,
+      "loss": 2.9336,
+      "step": 11545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.463627925413252e-05,
+      "loss": 2.9347,
+      "step": 11550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.463167946006042e-05,
+      "loss": 3.0036,
+      "step": 11555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.462707780638384e-05,
+      "loss": 3.0882,
+      "step": 11560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.462247429329453e-05,
+      "loss": 2.888,
+      "step": 11565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.461786892098427e-05,
+      "loss": 2.9685,
+      "step": 11570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.461326168964495e-05,
+      "loss": 3.0604,
+      "step": 11575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.460865259946852e-05,
+      "loss": 2.9369,
+      "step": 11580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.460404165064706e-05,
+      "loss": 2.9407,
+      "step": 11585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.459942884337264e-05,
+      "loss": 2.9697,
+      "step": 11590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.45948141778375e-05,
+      "loss": 2.955,
+      "step": 11595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.45901976542339e-05,
+      "loss": 3.1162,
+      "step": 11600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.458557927275417e-05,
+      "loss": 2.979,
+      "step": 11605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.458095903359075e-05,
+      "loss": 2.8902,
+      "step": 11610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.457633693693614e-05,
+      "loss": 3.1007,
+      "step": 11615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.457171298298293e-05,
+      "loss": 2.9604,
+      "step": 11620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.456708717192379e-05,
+      "loss": 2.9845,
+      "step": 11625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.456245950395143e-05,
+      "loss": 2.9693,
+      "step": 11630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.455782997925868e-05,
+      "loss": 2.8531,
+      "step": 11635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.455319859803843e-05,
+      "loss": 2.9128,
+      "step": 11640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.454856536048365e-05,
+      "loss": 2.8907,
+      "step": 11645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.454393026678738e-05,
+      "loss": 2.962,
+      "step": 11650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.453929331714277e-05,
+      "loss": 3.1157,
+      "step": 11655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.4534654511743e-05,
+      "loss": 2.8758,
+      "step": 11660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.453001385078133e-05,
+      "loss": 3.0131,
+      "step": 11665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.452537133445114e-05,
+      "loss": 3.0319,
+      "step": 11670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.452072696294586e-05,
+      "loss": 2.9186,
+      "step": 11675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.4516080736459e-05,
+      "loss": 2.9952,
+      "step": 11680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.451143265518414e-05,
+      "loss": 2.9595,
+      "step": 11685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.450678271931497e-05,
+      "loss": 3.0124,
+      "step": 11690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.450213092904519e-05,
+      "loss": 3.0463,
+      "step": 11695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.449747728456865e-05,
+      "loss": 3.0672,
+      "step": 11700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.449282178607925e-05,
+      "loss": 2.9932,
+      "step": 11705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.448816443377096e-05,
+      "loss": 2.9237,
+      "step": 11710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.448350522783779e-05,
+      "loss": 2.9919,
+      "step": 11715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.447884416847395e-05,
+      "loss": 2.9133,
+      "step": 11720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.447418125587357e-05,
+      "loss": 2.9213,
+      "step": 11725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.446951649023098e-05,
+      "loss": 3.1396,
+      "step": 11730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44648498717405e-05,
+      "loss": 3.0829,
+      "step": 11735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44601814005966e-05,
+      "loss": 2.9883,
+      "step": 11740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.445551107699377e-05,
+      "loss": 2.8766,
+      "step": 11745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.445083890112664e-05,
+      "loss": 2.9368,
+      "step": 11750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.444616487318985e-05,
+      "loss": 3.046,
+      "step": 11755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.444148899337815e-05,
+      "loss": 2.9385,
+      "step": 11760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.443681126188635e-05,
+      "loss": 2.9931,
+      "step": 11765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.443213167890937e-05,
+      "loss": 2.8884,
+      "step": 11770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.44274502446422e-05,
+      "loss": 2.9049,
+      "step": 11775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.442276695927986e-05,
+      "loss": 3.0332,
+      "step": 11780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.441808182301751e-05,
+      "loss": 2.9995,
+      "step": 11785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.441339483605033e-05,
+      "loss": 3.079,
+      "step": 11790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.440870599857363e-05,
+      "loss": 2.9816,
+      "step": 11795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.440401531078277e-05,
+      "loss": 3.0469,
+      "step": 11800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.439932277287318e-05,
+      "loss": 3.0981,
+      "step": 11805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.439462838504039e-05,
+      "loss": 3.0511,
+      "step": 11810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.438993214747999e-05,
+      "loss": 3.0456,
+      "step": 11815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.438523406038763e-05,
+      "loss": 3.0452,
+      "step": 11820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.43805341239591e-05,
+      "loss": 2.9907,
+      "step": 11825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.43758323383902e-05,
+      "loss": 2.9466,
+      "step": 11830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.437112870387684e-05,
+      "loss": 2.9204,
+      "step": 11835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.436642322061499e-05,
+      "loss": 2.9223,
+      "step": 11840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.436171588880071e-05,
+      "loss": 3.1225,
+      "step": 11845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.435700670863014e-05,
+      "loss": 3.0706,
+      "step": 11850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.435229568029948e-05,
+      "loss": 3.0274,
+      "step": 11855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.434758280400504e-05,
+      "loss": 2.9842,
+      "step": 11860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.434286807994316e-05,
+      "loss": 2.9409,
+      "step": 11865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.43381515083103e-05,
+      "loss": 2.9488,
+      "step": 11870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.433343308930295e-05,
+      "loss": 3.1277,
+      "step": 11875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.432871282311772e-05,
+      "loss": 3.0536,
+      "step": 11880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.43239907099513e-05,
+      "loss": 3.0321,
+      "step": 11885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.431926675000044e-05,
+      "loss": 3.0221,
+      "step": 11890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.431454094346195e-05,
+      "loss": 2.9224,
+      "step": 11895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.430981329053272e-05,
+      "loss": 2.9104,
+      "step": 11900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.430508379140976e-05,
+      "loss": 2.9579,
+      "step": 11905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.430035244629012e-05,
+      "loss": 3.0822,
+      "step": 11910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.429561925537092e-05,
+      "loss": 2.9949,
+      "step": 11915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.429088421884938e-05,
+      "loss": 2.8658,
+      "step": 11920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.42861473369228e-05,
+      "loss": 2.8716,
+      "step": 11925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.428140860978853e-05,
+      "loss": 2.9928,
+      "step": 11930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.4276668037644e-05,
+      "loss": 3.1144,
+      "step": 11935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.427192562068675e-05,
+      "loss": 2.8251,
+      "step": 11940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.426718135911438e-05,
+      "loss": 2.8464,
+      "step": 11945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.426243525312453e-05,
+      "loss": 2.8456,
+      "step": 11950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.4257687302915e-05,
+      "loss": 2.9737,
+      "step": 11955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.425293750868356e-05,
+      "loss": 3.0444,
+      "step": 11960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.424818587062816e-05,
+      "loss": 3.0515,
+      "step": 11965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.424343238894674e-05,
+      "loss": 3.0031,
+      "step": 11970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.423867706383738e-05,
+      "loss": 3.0397,
+      "step": 11975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.423391989549823e-05,
+      "loss": 2.9491,
+      "step": 11980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.422916088412745e-05,
+      "loss": 2.8305,
+      "step": 11985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.422440002992338e-05,
+      "loss": 2.9087,
+      "step": 11990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.421963733308434e-05,
+      "loss": 3.0152,
+      "step": 11995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.42148727938088e-05,
+      "loss": 2.8599,
+      "step": 12000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.421010641229528e-05,
+      "loss": 3.0352,
+      "step": 12005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.420533818874235e-05,
+      "loss": 2.9518,
+      "step": 12010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.42005681233487e-05,
+      "loss": 2.9375,
+      "step": 12015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.419579621631304e-05,
+      "loss": 2.9376,
+      "step": 12020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.419102246783427e-05,
+      "loss": 2.9481,
+      "step": 12025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.418624687811121e-05,
+      "loss": 2.9265,
+      "step": 12030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41814694473429e-05,
+      "loss": 2.971,
+      "step": 12035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.417669017572835e-05,
+      "loss": 3.0215,
+      "step": 12040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.417190906346672e-05,
+      "loss": 2.9694,
+      "step": 12045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.416712611075718e-05,
+      "loss": 2.9448,
+      "step": 12050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.416234131779907e-05,
+      "loss": 2.9309,
+      "step": 12055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.415755468479171e-05,
+      "loss": 3.0214,
+      "step": 12060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.415276621193455e-05,
+      "loss": 3.0238,
+      "step": 12065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41479758994271e-05,
+      "loss": 2.913,
+      "step": 12070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.414318374746895e-05,
+      "loss": 2.9468,
+      "step": 12075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.413838975625979e-05,
+      "loss": 2.9253,
+      "step": 12080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.413359392599934e-05,
+      "loss": 2.92,
+      "step": 12085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.412879625688742e-05,
+      "loss": 2.9561,
+      "step": 12090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.412399674912396e-05,
+      "loss": 3.0225,
+      "step": 12095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.411919540290887e-05,
+      "loss": 2.9652,
+      "step": 12100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.411439221844227e-05,
+      "loss": 2.996,
+      "step": 12105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.410958719592427e-05,
+      "loss": 3.0436,
+      "step": 12110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.410478033555504e-05,
+      "loss": 3.0967,
+      "step": 12115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.409997163753488e-05,
+      "loss": 3.0579,
+      "step": 12120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.409516110206417e-05,
+      "loss": 2.9395,
+      "step": 12125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.40903487293433e-05,
+      "loss": 2.9491,
+      "step": 12130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.408553451957283e-05,
+      "loss": 2.9399,
+      "step": 12135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.40807184729533e-05,
+      "loss": 3.0054,
+      "step": 12140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.407590058968541e-05,
+      "loss": 3.0768,
+      "step": 12145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.407108086996987e-05,
+      "loss": 3.2292,
+      "step": 12150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.406625931400753e-05,
+      "loss": 3.0617,
+      "step": 12155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.406143592199924e-05,
+      "loss": 2.8568,
+      "step": 12160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.4056610694146e-05,
+      "loss": 3.0824,
+      "step": 12165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.405178363064885e-05,
+      "loss": 2.9408,
+      "step": 12170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.404695473170892e-05,
+      "loss": 2.8991,
+      "step": 12175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.40421239975274e-05,
+      "loss": 3.0645,
+      "step": 12180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.403729142830555e-05,
+      "loss": 2.9931,
+      "step": 12185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.403245702424475e-05,
+      "loss": 2.9836,
+      "step": 12190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.402762078554641e-05,
+      "loss": 2.8983,
+      "step": 12195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.402278271241204e-05,
+      "loss": 3.033,
+      "step": 12200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.401794280504324e-05,
+      "loss": 2.9537,
+      "step": 12205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.401310106364162e-05,
+      "loss": 2.9914,
+      "step": 12210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.400825748840896e-05,
+      "loss": 2.887,
+      "step": 12215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.400341207954704e-05,
+      "loss": 2.8646,
+      "step": 12220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.399856483725778e-05,
+      "loss": 2.8998,
+      "step": 12225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.39937157617431e-05,
+      "loss": 2.9067,
+      "step": 12230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.398886485320507e-05,
+      "loss": 3.0589,
+      "step": 12235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.39840121118458e-05,
+      "loss": 3.102,
+      "step": 12240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.397915753786748e-05,
+      "loss": 3.0147,
+      "step": 12245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.397430113147237e-05,
+      "loss": 2.973,
+      "step": 12250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.396944289286284e-05,
+      "loss": 3.0201,
+      "step": 12255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.396458282224127e-05,
+      "loss": 3.0181,
+      "step": 12260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.395972091981022e-05,
+      "loss": 2.9198,
+      "step": 12265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.39548571857722e-05,
+      "loss": 2.9166,
+      "step": 12270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.39499916203299e-05,
+      "loss": 3.051,
+      "step": 12275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.394512422368602e-05,
+      "loss": 2.8469,
+      "step": 12280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.394025499604338e-05,
+      "loss": 2.9202,
+      "step": 12285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.393538393760485e-05,
+      "loss": 3.0227,
+      "step": 12290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.393051104857342e-05,
+      "loss": 2.9974,
+      "step": 12295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.392563632915206e-05,
+      "loss": 3.0322,
+      "step": 12300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.392075977954394e-05,
+      "loss": 2.9969,
+      "step": 12305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.391588139995218e-05,
+      "loss": 2.8248,
+      "step": 12310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.39110011905801e-05,
+      "loss": 2.8892,
+      "step": 12315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.390611915163102e-05,
+      "loss": 3.013,
+      "step": 12320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.390123528330832e-05,
+      "loss": 2.961,
+      "step": 12325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.389634958581552e-05,
+      "loss": 3.0334,
+      "step": 12330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.389146205935619e-05,
+      "loss": 2.9564,
+      "step": 12335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.388657270413394e-05,
+      "loss": 2.9704,
+      "step": 12340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.388168152035253e-05,
+      "loss": 3.0291,
+      "step": 12345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.387678850821573e-05,
+      "loss": 2.8282,
+      "step": 12350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.387189366792742e-05,
+      "loss": 2.9047,
+      "step": 12355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.386699699969153e-05,
+      "loss": 2.988,
+      "step": 12360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.386209850371208e-05,
+      "loss": 2.9515,
+      "step": 12365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.385719818019321e-05,
+      "loss": 3.0176,
+      "step": 12370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.385229602933903e-05,
+      "loss": 2.8627,
+      "step": 12375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.384739205135385e-05,
+      "loss": 3.0336,
+      "step": 12380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.384248624644197e-05,
+      "loss": 2.9908,
+      "step": 12385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.38375786148078e-05,
+      "loss": 2.9152,
+      "step": 12390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.38326691566558e-05,
+      "loss": 3.0232,
+      "step": 12395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.382775787219056e-05,
+      "loss": 3.0021,
+      "step": 12400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.38228447616167e-05,
+      "loss": 2.9667,
+      "step": 12405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.381792982513888e-05,
+      "loss": 2.9475,
+      "step": 12410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.381301306296196e-05,
+      "loss": 3.0315,
+      "step": 12415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.380809447529076e-05,
+      "loss": 2.9147,
+      "step": 12420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.380317406233019e-05,
+      "loss": 3.0464,
+      "step": 12425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.379825182428533e-05,
+      "loss": 2.9307,
+      "step": 12430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.37933277613612e-05,
+      "loss": 3.0401,
+      "step": 12435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.3788401873763e-05,
+      "loss": 2.9333,
+      "step": 12440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.378347416169597e-05,
+      "loss": 2.9506,
+      "step": 12445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.377854462536542e-05,
+      "loss": 2.8773,
+      "step": 12450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.377361326497674e-05,
+      "loss": 2.9851,
+      "step": 12455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.376868008073538e-05,
+      "loss": 3.0231,
+      "step": 12460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.376374507284693e-05,
+      "loss": 2.9875,
+      "step": 12465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.375880824151697e-05,
+      "loss": 2.8422,
+      "step": 12470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.37538695869512e-05,
+      "loss": 3.0913,
+      "step": 12475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.374892910935542e-05,
+      "loss": 3.0021,
+      "step": 12480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.374398680893544e-05,
+      "loss": 3.0115,
+      "step": 12485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.373904268589722e-05,
+      "loss": 2.8879,
+      "step": 12490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.373409674044673e-05,
+      "loss": 2.8389,
+      "step": 12495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.372914897279006e-05,
+      "loss": 2.8946,
+      "step": 12500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.372419938313335e-05,
+      "loss": 2.9602,
+      "step": 12505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.371924797168286e-05,
+      "loss": 3.0124,
+      "step": 12510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.371429473864483e-05,
+      "loss": 2.8823,
+      "step": 12515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.37093396842257e-05,
+      "loss": 2.9257,
+      "step": 12520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.370438280863192e-05,
+      "loss": 3.0228,
+      "step": 12525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.369942411207e-05,
+      "loss": 2.9471,
+      "step": 12530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.369446359474654e-05,
+      "loss": 2.9616,
+      "step": 12535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.368950125686824e-05,
+      "loss": 2.9532,
+      "step": 12540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.368453709864184e-05,
+      "loss": 2.9838,
+      "step": 12545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.367957112027421e-05,
+      "loss": 2.966,
+      "step": 12550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.367460332197223e-05,
+      "loss": 2.913,
+      "step": 12555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.366963370394289e-05,
+      "loss": 2.9304,
+      "step": 12560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.366466226639327e-05,
+      "loss": 3.0484,
+      "step": 12565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.365968900953048e-05,
+      "loss": 3.0055,
+      "step": 12570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.365471393356176e-05,
+      "loss": 3.1115,
+      "step": 12575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.364973703869437e-05,
+      "loss": 2.9958,
+      "step": 12580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.36447583251357e-05,
+      "loss": 2.8839,
+      "step": 12585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.363977779309321e-05,
+      "loss": 2.9581,
+      "step": 12590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.363479544277436e-05,
+      "loss": 2.9742,
+      "step": 12595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.362981127438677e-05,
+      "loss": 2.9066,
+      "step": 12600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.362482528813815e-05,
+      "loss": 2.9986,
+      "step": 12605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.361983748423615e-05,
+      "loss": 3.0361,
+      "step": 12610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.361484786288868e-05,
+      "loss": 3.1252,
+      "step": 12615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.36098564243036e-05,
+      "loss": 3.0347,
+      "step": 12620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.360486316868886e-05,
+      "loss": 2.9769,
+      "step": 12625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.359986809625254e-05,
+      "loss": 2.9662,
+      "step": 12630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.359487120720276e-05,
+      "loss": 2.9433,
+      "step": 12635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.358987250174769e-05,
+      "loss": 2.9276,
+      "step": 12640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.358487198009564e-05,
+      "loss": 2.893,
+      "step": 12645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.357986964245494e-05,
+      "loss": 3.0764,
+      "step": 12650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.3574865489034e-05,
+      "loss": 2.914,
+      "step": 12655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.356985952004136e-05,
+      "loss": 2.9787,
+      "step": 12660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.356485173568556e-05,
+      "loss": 2.8958,
+      "step": 12665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.355984213617528e-05,
+      "loss": 2.9618,
+      "step": 12670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.355483072171922e-05,
+      "loss": 2.8932,
+      "step": 12675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.354981749252621e-05,
+      "loss": 2.8999,
+      "step": 12680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.354480244880513e-05,
+      "loss": 2.961,
+      "step": 12685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.353978559076491e-05,
+      "loss": 3.0882,
+      "step": 12690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.353476691861458e-05,
+      "loss": 2.9626,
+      "step": 12695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.352974643256327e-05,
+      "loss": 2.943,
+      "step": 12700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.352472413282015e-05,
+      "loss": 2.988,
+      "step": 12705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.351970001959447e-05,
+      "loss": 2.8394,
+      "step": 12710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.351467409309558e-05,
+      "loss": 2.8633,
+      "step": 12715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.350964635353287e-05,
+      "loss": 2.938,
+      "step": 12720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.350461680111584e-05,
+      "loss": 2.8293,
+      "step": 12725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.349958543605404e-05,
+      "loss": 3.0516,
+      "step": 12730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.34945522585571e-05,
+      "loss": 3.0327,
+      "step": 12735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.348951726883473e-05,
+      "loss": 2.8453,
+      "step": 12740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.348448046709672e-05,
+      "loss": 2.8827,
+      "step": 12745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.347944185355295e-05,
+      "loss": 2.8211,
+      "step": 12750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.347440142841333e-05,
+      "loss": 2.8354,
+      "step": 12755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.346935919188787e-05,
+      "loss": 3.0236,
+      "step": 12760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.346431514418668e-05,
+      "loss": 2.7448,
+      "step": 12765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.345926928551991e-05,
+      "loss": 2.9333,
+      "step": 12770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.34542216160978e-05,
+      "loss": 3.021,
+      "step": 12775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.344917213613067e-05,
+      "loss": 2.9773,
+      "step": 12780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.34441208458289e-05,
+      "loss": 2.89,
+      "step": 12785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.343906774540295e-05,
+      "loss": 2.8555,
+      "step": 12790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.343401283506338e-05,
+      "loss": 3.0151,
+      "step": 12795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.342895611502079e-05,
+      "loss": 2.8109,
+      "step": 12800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.342389758548588e-05,
+      "loss": 3.0413,
+      "step": 12805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.34188372466694e-05,
+      "loss": 3.0184,
+      "step": 12810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.341377509878223e-05,
+      "loss": 2.9875,
+      "step": 12815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.340871114203524e-05,
+      "loss": 2.9517,
+      "step": 12820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.340364537663946e-05,
+      "loss": 3.0377,
+      "step": 12825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.339857780280594e-05,
+      "loss": 3.0372,
+      "step": 12830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.339350842074582e-05,
+      "loss": 2.911,
+      "step": 12835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.338843723067032e-05,
+      "loss": 2.993,
+      "step": 12840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.338336423279075e-05,
+      "loss": 2.9139,
+      "step": 12845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.337828942731846e-05,
+      "loss": 2.9193,
+      "step": 12850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.33732128144649e-05,
+      "loss": 3.0664,
+      "step": 12855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.336813439444161e-05,
+      "loss": 3.0146,
+      "step": 12860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.336305416746016e-05,
+      "loss": 2.8752,
+      "step": 12865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.335797213373222e-05,
+      "loss": 2.9491,
+      "step": 12870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.335288829346955e-05,
+      "loss": 2.8167,
+      "step": 12875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.334780264688396e-05,
+      "loss": 3.0811,
+      "step": 12880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.334271519418736e-05,
+      "loss": 3.0396,
+      "step": 12885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.33376259355917e-05,
+      "loss": 3.0145,
+      "step": 12890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.333253487130906e-05,
+      "loss": 2.9339,
+      "step": 12895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.332744200155154e-05,
+      "loss": 2.8554,
+      "step": 12900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.332234732653133e-05,
+      "loss": 3.0231,
+      "step": 12905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.331725084646072e-05,
+      "loss": 2.9948,
+      "step": 12910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.331215256155203e-05,
+      "loss": 3.0469,
+      "step": 12915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.330705247201773e-05,
+      "loss": 2.9725,
+      "step": 12920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.330195057807027e-05,
+      "loss": 2.8088,
+      "step": 12925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.329684687992226e-05,
+      "loss": 2.9652,
+      "step": 12930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.329174137778634e-05,
+      "loss": 3.0526,
+      "step": 12935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.32866340718752e-05,
+      "loss": 2.917,
+      "step": 12940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.328152496240169e-05,
+      "loss": 2.9542,
+      "step": 12945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.327641404957866e-05,
+      "loss": 3.028,
+      "step": 12950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.327130133361904e-05,
+      "loss": 2.9979,
+      "step": 12955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.326618681473589e-05,
+      "loss": 2.8613,
+      "step": 12960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.326107049314228e-05,
+      "loss": 2.9775,
+      "step": 12965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.32559523690514e-05,
+      "loss": 2.878,
+      "step": 12970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.32508324426765e-05,
+      "loss": 2.9421,
+      "step": 12975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.32457107142309e-05,
+      "loss": 2.9552,
+      "step": 12980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.324058718392798e-05,
+      "loss": 3.0313,
+      "step": 12985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.323546185198127e-05,
+      "loss": 2.9888,
+      "step": 12990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.323033471860427e-05,
+      "loss": 2.9188,
+      "step": 12995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.322520578401062e-05,
+      "loss": 3.0178,
+      "step": 13000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.322007504841401e-05,
+      "loss": 2.9416,
+      "step": 13005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.321494251202824e-05,
+      "loss": 2.9491,
+      "step": 13010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.320980817506715e-05,
+      "loss": 2.9358,
+      "step": 13015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.320467203774465e-05,
+      "loss": 3.0592,
+      "step": 13020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.319953410027475e-05,
+      "loss": 2.9129,
+      "step": 13025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.319439436287154e-05,
+      "loss": 2.9924,
+      "step": 13030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.318925282574914e-05,
+      "loss": 3.0187,
+      "step": 13035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.31841094891218e-05,
+      "loss": 2.8792,
+      "step": 13040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.31789643532038e-05,
+      "loss": 2.996,
+      "step": 13045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.317381741820955e-05,
+      "loss": 2.9112,
+      "step": 13050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.316866868435345e-05,
+      "loss": 2.9155,
+      "step": 13055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.316351815185006e-05,
+      "loss": 2.7358,
+      "step": 13060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.315836582091397e-05,
+      "loss": 3.0136,
+      "step": 13065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.315321169175986e-05,
+      "loss": 3.0198,
+      "step": 13070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.314805576460246e-05,
+      "loss": 2.9674,
+      "step": 13075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.314289803965663e-05,
+      "loss": 2.9487,
+      "step": 13080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.313773851713725e-05,
+      "loss": 2.853,
+      "step": 13085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.313257719725927e-05,
+      "loss": 2.9407,
+      "step": 13090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.31274140802378e-05,
+      "loss": 2.9535,
+      "step": 13095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.31222491662879e-05,
+      "loss": 2.8551,
+      "step": 13100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.31170824556248e-05,
+      "loss": 2.8434,
+      "step": 13105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.311191394846378e-05,
+      "loss": 2.8759,
+      "step": 13110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.310674364502018e-05,
+      "loss": 2.8944,
+      "step": 13115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.310157154550942e-05,
+      "loss": 2.9992,
+      "step": 13120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.3096397650147e-05,
+      "loss": 2.8724,
+      "step": 13125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.309122195914848e-05,
+      "loss": 2.9355,
+      "step": 13130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.308604447272955e-05,
+      "loss": 2.9413,
+      "step": 13135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.30808651911059e-05,
+      "loss": 2.9572,
+      "step": 13140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.307568411449333e-05,
+      "loss": 3.0392,
+      "step": 13145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.30705012431077e-05,
+      "loss": 2.9721,
+      "step": 13150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.3065316577165e-05,
+      "loss": 2.983,
+      "step": 13155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.306013011688121e-05,
+      "loss": 3.0551,
+      "step": 13160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.305494186247246e-05,
+      "loss": 3.0609,
+      "step": 13165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.304975181415489e-05,
+      "loss": 3.0145,
+      "step": 13170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.304455997214476e-05,
+      "loss": 3.0406,
+      "step": 13175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.30393663366584e-05,
+      "loss": 3.0239,
+      "step": 13180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.303417090791218e-05,
+      "loss": 2.9775,
+      "step": 13185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.30289736861226e-05,
+      "loss": 2.953,
+      "step": 13190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.302377467150619e-05,
+      "loss": 2.8562,
+      "step": 13195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.301857386427956e-05,
+      "loss": 2.9672,
+      "step": 13200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.301337126465945e-05,
+      "loss": 3.0083,
+      "step": 13205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.300816687286258e-05,
+      "loss": 2.8941,
+      "step": 13210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.300296068910581e-05,
+      "loss": 2.8636,
+      "step": 13215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.299775271360606e-05,
+      "loss": 2.9339,
+      "step": 13220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.299254294658033e-05,
+      "loss": 2.9911,
+      "step": 13225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.298733138824567e-05,
+      "loss": 2.8789,
+      "step": 13230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.298211803881924e-05,
+      "loss": 2.8568,
+      "step": 13235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.297690289851824e-05,
+      "loss": 2.9081,
+      "step": 13240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.297168596755999e-05,
+      "loss": 2.9291,
+      "step": 13245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.296646724616182e-05,
+      "loss": 2.8986,
+      "step": 13250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.29612467345412e-05,
+      "loss": 2.9865,
+      "step": 13255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.295602443291563e-05,
+      "loss": 2.9314,
+      "step": 13260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.295080034150272e-05,
+      "loss": 3.0028,
+      "step": 13265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.29455744605201e-05,
+      "loss": 2.8921,
+      "step": 13270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.294034679018554e-05,
+      "loss": 2.8478,
+      "step": 13275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.293511733071684e-05,
+      "loss": 2.9377,
+      "step": 13280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.29298860823319e-05,
+      "loss": 2.9277,
+      "step": 13285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.292465304524866e-05,
+      "loss": 2.9483,
+      "step": 13290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.29194182196852e-05,
+      "loss": 2.8785,
+      "step": 13295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.291418160585958e-05,
+      "loss": 2.9814,
+      "step": 13300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.290894320399002e-05,
+      "loss": 2.891,
+      "step": 13305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.290370301429476e-05,
+      "loss": 2.9392,
+      "step": 13310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.289846103699216e-05,
+      "loss": 2.8724,
+      "step": 13315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.289321727230062e-05,
+      "loss": 3.0409,
+      "step": 13320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.288797172043862e-05,
+      "loss": 3.0617,
+      "step": 13325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.288272438162471e-05,
+      "loss": 3.0514,
+      "step": 13330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.287747525607753e-05,
+      "loss": 3.1126,
+      "step": 13335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.287222434401582e-05,
+      "loss": 2.8811,
+      "step": 13340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.28669716456583e-05,
+      "loss": 2.945,
+      "step": 13345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.28617171612239e-05,
+      "loss": 2.9667,
+      "step": 13350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.285646089093148e-05,
+      "loss": 2.8849,
+      "step": 13355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.285120283500008e-05,
+      "loss": 2.9593,
+      "step": 13360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.28459429936488e-05,
+      "loss": 2.9355,
+      "step": 13365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.284068136709675e-05,
+      "loss": 2.9289,
+      "step": 13370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.283541795556319e-05,
+      "loss": 2.9723,
+      "step": 13375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.283015275926741e-05,
+      "loss": 2.9423,
+      "step": 13380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.282488577842879e-05,
+      "loss": 3.0351,
+      "step": 13385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.281961701326678e-05,
+      "loss": 3.0467,
+      "step": 13390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.281434646400092e-05,
+      "loss": 2.9559,
+      "step": 13395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.280907413085079e-05,
+      "loss": 2.9564,
+      "step": 13400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.280380001403608e-05,
+      "loss": 2.9526,
+      "step": 13405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.279852411377652e-05,
+      "loss": 2.9588,
+      "step": 13410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.279324643029198e-05,
+      "loss": 2.9292,
+      "step": 13415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.278796696380227e-05,
+      "loss": 2.9512,
+      "step": 13420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.278268571452746e-05,
+      "loss": 3.0541,
+      "step": 13425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.277740268268752e-05,
+      "loss": 2.8527,
+      "step": 13430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.277211786850264e-05,
+      "loss": 2.928,
+      "step": 13435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.276683127219297e-05,
+      "loss": 3.0076,
+      "step": 13440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.276154289397877e-05,
+      "loss": 2.9502,
+      "step": 13445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.275625273408041e-05,
+      "loss": 2.9008,
+      "step": 13450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.275096079271829e-05,
+      "loss": 2.949,
+      "step": 13455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.274566707011291e-05,
+      "loss": 2.8938,
+      "step": 13460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.274037156648485e-05,
+      "loss": 2.97,
+      "step": 13465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.273507428205472e-05,
+      "loss": 2.9413,
+      "step": 13470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.272977521704325e-05,
+      "loss": 2.9706,
+      "step": 13475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.272447437167125e-05,
+      "loss": 2.9515,
+      "step": 13480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.271917174615953e-05,
+      "loss": 2.9532,
+      "step": 13485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.271386734072908e-05,
+      "loss": 2.9177,
+      "step": 13490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.270856115560089e-05,
+      "loss": 2.8936,
+      "step": 13495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.270325319099602e-05,
+      "loss": 2.8836,
+      "step": 13500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.269794344713568e-05,
+      "loss": 2.9325,
+      "step": 13505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.269263192424107e-05,
+      "loss": 2.9614,
+      "step": 13510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.26873186225335e-05,
+      "loss": 2.9393,
+      "step": 13515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.268200354223437e-05,
+      "loss": 2.906,
+      "step": 13520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.267668668356512e-05,
+      "loss": 3.0312,
+      "step": 13525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.267136804674727e-05,
+      "loss": 2.823,
+      "step": 13530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.266604763200245e-05,
+      "loss": 2.9449,
+      "step": 13535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.266072543955233e-05,
+      "loss": 2.8243,
+      "step": 13540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.265540146961864e-05,
+      "loss": 2.9815,
+      "step": 13545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.265007572242323e-05,
+      "loss": 3.0259,
+      "step": 13550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.264474819818799e-05,
+      "loss": 2.9064,
+      "step": 13555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.26394188971349e-05,
+      "loss": 3.0546,
+      "step": 13560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.263408781948601e-05,
+      "loss": 2.9182,
+      "step": 13565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.262875496546344e-05,
+      "loss": 2.9958,
+      "step": 13570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.262342033528939e-05,
+      "loss": 3.0118,
+      "step": 13575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.261808392918611e-05,
+      "loss": 2.8782,
+      "step": 13580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.261274574737598e-05,
+      "loss": 2.9033,
+      "step": 13585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.260740579008138e-05,
+      "loss": 2.9572,
+      "step": 13590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.260206405752483e-05,
+      "loss": 2.9777,
+      "step": 13595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.259672054992888e-05,
+      "loss": 2.9119,
+      "step": 13600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.25913752675162e-05,
+      "loss": 2.9799,
+      "step": 13605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.258602821050946e-05,
+      "loss": 2.8399,
+      "step": 13610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.258067937913147e-05,
+      "loss": 2.9522,
+      "step": 13615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.25753287736051e-05,
+      "loss": 2.999,
+      "step": 13620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.256997639415327e-05,
+      "loss": 2.9351,
+      "step": 13625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.2564622240999e-05,
+      "loss": 3.0358,
+      "step": 13630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.255926631436538e-05,
+      "loss": 3.0005,
+      "step": 13635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.255390861447555e-05,
+      "loss": 2.8364,
+      "step": 13640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.254854914155276e-05,
+      "loss": 2.8407,
+      "step": 13645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.25431878958203e-05,
+      "loss": 2.8911,
+      "step": 13650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.253782487750156e-05,
+      "loss": 2.9608,
+      "step": 13655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.253246008681999e-05,
+      "loss": 2.9256,
+      "step": 13660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.252709352399912e-05,
+      "loss": 2.8842,
+      "step": 13665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.252172518926254e-05,
+      "loss": 2.809,
+      "step": 13670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.251635508283394e-05,
+      "loss": 2.9176,
+      "step": 13675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.251098320493706e-05,
+      "loss": 3.004,
+      "step": 13680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.250560955579571e-05,
+      "loss": 3.0108,
+      "step": 13685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.250023413563383e-05,
+      "loss": 2.8325,
+      "step": 13690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.249485694467532e-05,
+      "loss": 2.939,
+      "step": 13695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.248947798314429e-05,
+      "loss": 2.8357,
+      "step": 13700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.248409725126481e-05,
+      "loss": 2.966,
+      "step": 13705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.247871474926111e-05,
+      "loss": 2.8037,
+      "step": 13710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.247333047735743e-05,
+      "loss": 2.9772,
+      "step": 13715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.246794443577811e-05,
+      "loss": 2.9486,
+      "step": 13720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.246255662474758e-05,
+      "loss": 2.963,
+      "step": 13725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.24571670444903e-05,
+      "loss": 3.0796,
+      "step": 13730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.245177569523085e-05,
+      "loss": 3.0045,
+      "step": 13735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.244638257719387e-05,
+      "loss": 2.9876,
+      "step": 13740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.244098769060404e-05,
+      "loss": 2.8888,
+      "step": 13745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.243559103568618e-05,
+      "loss": 2.8463,
+      "step": 13750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.243019261266508e-05,
+      "loss": 2.9915,
+      "step": 13755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.242479242176573e-05,
+      "loss": 2.8971,
+      "step": 13760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.241939046321312e-05,
+      "loss": 2.9582,
+      "step": 13765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.241398673723231e-05,
+      "loss": 2.9391,
+      "step": 13770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.240858124404847e-05,
+      "loss": 2.7893,
+      "step": 13775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.240317398388679e-05,
+      "loss": 2.8989,
+      "step": 13780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.239776495697259e-05,
+      "loss": 2.8157,
+      "step": 13785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.239235416353124e-05,
+      "loss": 2.9562,
+      "step": 13790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.238694160378818e-05,
+      "loss": 2.8863,
+      "step": 13795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.238152727796894e-05,
+      "loss": 2.9099,
+      "step": 13800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.237611118629907e-05,
+      "loss": 2.8401,
+      "step": 13805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.237069332900429e-05,
+      "loss": 2.8424,
+      "step": 13810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.236527370631029e-05,
+      "loss": 2.9589,
+      "step": 13815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.23598523184429e-05,
+      "loss": 3.0488,
+      "step": 13820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.235442916562805e-05,
+      "loss": 3.0959,
+      "step": 13825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.234900424809162e-05,
+      "loss": 2.9314,
+      "step": 13830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.234357756605968e-05,
+      "loss": 2.9688,
+      "step": 13835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.233814911975834e-05,
+      "loss": 2.807,
+      "step": 13840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.233271890941378e-05,
+      "loss": 2.873,
+      "step": 13845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.232728693525225e-05,
+      "loss": 2.8972,
+      "step": 13850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.232185319750006e-05,
+      "loss": 3.0023,
+      "step": 13855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.231641769638363e-05,
+      "loss": 2.8929,
+      "step": 13860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.231098043212942e-05,
+      "loss": 2.8631,
+      "step": 13865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.2305541404964e-05,
+      "loss": 2.9238,
+      "step": 13870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.230010061511396e-05,
+      "loss": 2.8682,
+      "step": 13875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.229465806280601e-05,
+      "loss": 2.8638,
+      "step": 13880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.228921374826691e-05,
+      "loss": 2.9762,
+      "step": 13885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.228376767172351e-05,
+      "loss": 2.9536,
+      "step": 13890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.227831983340271e-05,
+      "loss": 3.003,
+      "step": 13895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.227287023353152e-05,
+      "loss": 2.8504,
+      "step": 13900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.226741887233698e-05,
+      "loss": 2.8681,
+      "step": 13905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.226196575004622e-05,
+      "loss": 2.9176,
+      "step": 13910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.225651086688647e-05,
+      "loss": 2.9276,
+      "step": 13915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.2251054223085e-05,
+      "loss": 2.8129,
+      "step": 13920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.224559581886916e-05,
+      "loss": 2.8694,
+      "step": 13925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.224013565446638e-05,
+      "loss": 2.8118,
+      "step": 13930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.223467373010415e-05,
+      "loss": 2.989,
+      "step": 13935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.222921004601007e-05,
+      "loss": 3.0057,
+      "step": 13940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.222374460241177e-05,
+      "loss": 2.8829,
+      "step": 13945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.221827739953697e-05,
+      "loss": 2.9551,
+      "step": 13950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.221280843761349e-05,
+      "loss": 2.7609,
+      "step": 13955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.220733771686915e-05,
+      "loss": 2.919,
+      "step": 13960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.220186523753191e-05,
+      "loss": 2.838,
+      "step": 13965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.219639099982981e-05,
+      "loss": 2.9994,
+      "step": 13970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.219091500399091e-05,
+      "loss": 3.0366,
+      "step": 13975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.218543725024338e-05,
+      "loss": 2.7743,
+      "step": 13980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.217995773881545e-05,
+      "loss": 2.8914,
+      "step": 13985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.217447646993544e-05,
+      "loss": 3.0376,
+      "step": 13990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.21689934438317e-05,
+      "loss": 2.9026,
+      "step": 13995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.216350866073271e-05,
+      "loss": 3.0362,
+      "step": 14000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.2158022120867e-05,
+      "loss": 3.0947,
+      "step": 14005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.215253382446314e-05,
+      "loss": 3.0046,
+      "step": 14010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.214704377174983e-05,
+      "loss": 2.8791,
+      "step": 14015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.214155196295582e-05,
+      "loss": 3.0459,
+      "step": 14020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.21360583983099e-05,
+      "loss": 2.8902,
+      "step": 14025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.213056307804098e-05,
+      "loss": 2.9952,
+      "step": 14030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.212506600237803e-05,
+      "loss": 2.8816,
+      "step": 14035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.21195671715501e-05,
+      "loss": 2.8371,
+      "step": 14040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.211406658578628e-05,
+      "loss": 2.907,
+      "step": 14045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.210856424531573e-05,
+      "loss": 2.8836,
+      "step": 14050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.210306015036778e-05,
+      "loss": 3.0754,
+      "step": 14055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.209755430117169e-05,
+      "loss": 2.9896,
+      "step": 14060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.209204669795691e-05,
+      "loss": 3.0156,
+      "step": 14065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.208653734095291e-05,
+      "loss": 3.0452,
+      "step": 14070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.208102623038923e-05,
+      "loss": 2.8998,
+      "step": 14075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.207551336649549e-05,
+      "loss": 2.787,
+      "step": 14080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.20699987495014e-05,
+      "loss": 2.9392,
+      "step": 14085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.20644823796367e-05,
+      "loss": 2.8171,
+      "step": 14090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.205896425713127e-05,
+      "loss": 2.8908,
+      "step": 14095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.205344438221502e-05,
+      "loss": 2.8849,
+      "step": 14100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.204792275511793e-05,
+      "loss": 2.8844,
+      "step": 14105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.204239937607007e-05,
+      "loss": 3.0461,
+      "step": 14110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.203687424530154e-05,
+      "loss": 2.8959,
+      "step": 14115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.20313473630426e-05,
+      "loss": 2.9401,
+      "step": 14120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.202581872952348e-05,
+      "loss": 3.0823,
+      "step": 14125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.202028834497458e-05,
+      "loss": 2.9268,
+      "step": 14130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.20147562096263e-05,
+      "loss": 2.8288,
+      "step": 14135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.200922232370914e-05,
+      "loss": 2.857,
+      "step": 14140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.200368668745368e-05,
+      "loss": 2.9768,
+      "step": 14145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.199814930109057e-05,
+      "loss": 3.057,
+      "step": 14150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.19926101648505e-05,
+      "loss": 2.8786,
+      "step": 14155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.19870692789643e-05,
+      "loss": 2.8993,
+      "step": 14160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.198152664366281e-05,
+      "loss": 2.8734,
+      "step": 14165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.197598225917699e-05,
+      "loss": 2.913,
+      "step": 14170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.197043612573781e-05,
+      "loss": 2.9349,
+      "step": 14175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.196488824357639e-05,
+      "loss": 2.972,
+      "step": 14180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.195933861292388e-05,
+      "loss": 2.9776,
+      "step": 14185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.195378723401149e-05,
+      "loss": 3.0217,
+      "step": 14190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.194823410707053e-05,
+      "loss": 2.8182,
+      "step": 14195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.194267923233237e-05,
+      "loss": 2.9257,
+      "step": 14200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.193712261002849e-05,
+      "loss": 3.1178,
+      "step": 14205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.193156424039036e-05,
+      "loss": 2.8631,
+      "step": 14210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.19260041236496e-05,
+      "loss": 3.0071,
+      "step": 14215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.192044226003789e-05,
+      "loss": 2.9554,
+      "step": 14220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.191487864978692e-05,
+      "loss": 2.9001,
+      "step": 14225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.190931329312854e-05,
+      "loss": 2.8313,
+      "step": 14230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.190374619029464e-05,
+      "loss": 2.9193,
+      "step": 14235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.189817734151716e-05,
+      "loss": 2.9855,
+      "step": 14240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.189260674702812e-05,
+      "loss": 2.8269,
+      "step": 14245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.188703440705964e-05,
+      "loss": 3.0357,
+      "step": 14250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.188146032184387e-05,
+      "loss": 2.8285,
+      "step": 14255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.187588449161309e-05,
+      "loss": 2.9113,
+      "step": 14260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.18703069165996e-05,
+      "loss": 2.9674,
+      "step": 14265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.186472759703579e-05,
+      "loss": 2.8318,
+      "step": 14270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.185914653315414e-05,
+      "loss": 2.9868,
+      "step": 14275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.185356372518718e-05,
+      "loss": 2.9657,
+      "step": 14280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.184797917336754e-05,
+      "loss": 2.7754,
+      "step": 14285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.184239287792787e-05,
+      "loss": 3.0196,
+      "step": 14290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.183680483910093e-05,
+      "loss": 3.0068,
+      "step": 14295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.183121505711958e-05,
+      "loss": 2.8214,
+      "step": 14300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.18256235322167e-05,
+      "loss": 2.8588,
+      "step": 14305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.182003026462525e-05,
+      "loss": 2.8649,
+      "step": 14310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.181443525457831e-05,
+      "loss": 2.9415,
+      "step": 14315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.180883850230898e-05,
+      "loss": 2.9417,
+      "step": 14320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.180324000805045e-05,
+      "loss": 3.0032,
+      "step": 14325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.1797639772036e-05,
+      "loss": 2.95,
+      "step": 14330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.179203779449894e-05,
+      "loss": 2.8904,
+      "step": 14335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.17864340756727e-05,
+      "loss": 3.0578,
+      "step": 14340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.178082861579076e-05,
+      "loss": 3.0032,
+      "step": 14345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.177522141508666e-05,
+      "loss": 2.9726,
+      "step": 14350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.176961247379405e-05,
+      "loss": 2.9182,
+      "step": 14355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.176400179214663e-05,
+      "loss": 2.8351,
+      "step": 14360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.175838937037816e-05,
+      "loss": 2.7611,
+      "step": 14365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.175277520872246e-05,
+      "loss": 2.8828,
+      "step": 14370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.17471593074135e-05,
+      "loss": 3.0094,
+      "step": 14375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.174154166668522e-05,
+      "loss": 3.0309,
+      "step": 14380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.173592228677172e-05,
+      "loss": 2.9035,
+      "step": 14385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.173030116790713e-05,
+      "loss": 3.004,
+      "step": 14390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.172467831032562e-05,
+      "loss": 2.8411,
+      "step": 14395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.171905371426152e-05,
+      "loss": 2.9033,
+      "step": 14400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.171342737994916e-05,
+      "loss": 2.881,
+      "step": 14405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.170779930762295e-05,
+      "loss": 3.0595,
+      "step": 14410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.17021694975174e-05,
+      "loss": 2.9886,
+      "step": 14415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.169653794986709e-05,
+      "loss": 2.9375,
+      "step": 14420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.169090466490664e-05,
+      "loss": 2.8564,
+      "step": 14425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.168526964287079e-05,
+      "loss": 3.0056,
+      "step": 14430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.16796328839943e-05,
+      "loss": 2.9924,
+      "step": 14435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.167399438851204e-05,
+      "loss": 2.9779,
+      "step": 14440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.166835415665896e-05,
+      "loss": 2.9352,
+      "step": 14445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.166271218867002e-05,
+      "loss": 2.8627,
+      "step": 14450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.165706848478033e-05,
+      "loss": 2.9249,
+      "step": 14455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.165142304522502e-05,
+      "loss": 2.8584,
+      "step": 14460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.164577587023932e-05,
+      "loss": 2.8828,
+      "step": 14465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.164012696005855e-05,
+      "loss": 2.758,
+      "step": 14470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.163447631491802e-05,
+      "loss": 2.9551,
+      "step": 14475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.16288239350532e-05,
+      "loss": 2.9153,
+      "step": 14480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.162316982069959e-05,
+      "loss": 2.9199,
+      "step": 14485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.161751397209279e-05,
+      "loss": 3.0896,
+      "step": 14490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.161185638946843e-05,
+      "loss": 2.9206,
+      "step": 14495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.160619707306225e-05,
+      "loss": 2.8857,
+      "step": 14500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.160053602311006e-05,
+      "loss": 2.9279,
+      "step": 14505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.159487323984769e-05,
+      "loss": 2.9166,
+      "step": 14510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.158920872351112e-05,
+      "loss": 2.9772,
+      "step": 14515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.158354247433636e-05,
+      "loss": 2.7452,
+      "step": 14520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.157787449255948e-05,
+      "loss": 2.8393,
+      "step": 14525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.157220477841666e-05,
+      "loss": 2.9516,
+      "step": 14530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.156653333214412e-05,
+      "loss": 2.823,
+      "step": 14535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.156086015397818e-05,
+      "loss": 2.8503,
+      "step": 14540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.155518524415518e-05,
+      "loss": 2.8659,
+      "step": 14545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.154950860291162e-05,
+      "loss": 2.9746,
+      "step": 14550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.154383023048399e-05,
+      "loss": 2.9347,
+      "step": 14555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.153815012710887e-05,
+      "loss": 3.0403,
+      "step": 14560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.153246829302295e-05,
+      "loss": 2.8964,
+      "step": 14565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.152678472846295e-05,
+      "loss": 2.9736,
+      "step": 14570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.15210994336657e-05,
+      "loss": 2.8599,
+      "step": 14575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.151541240886804e-05,
+      "loss": 2.9569,
+      "step": 14580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.150972365430696e-05,
+      "loss": 3.0535,
+      "step": 14585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.15040331702195e-05,
+      "loss": 2.9169,
+      "step": 14590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.149834095684271e-05,
+      "loss": 3.04,
+      "step": 14595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.149264701441379e-05,
+      "loss": 3.0156,
+      "step": 14600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.148695134316997e-05,
+      "loss": 2.913,
+      "step": 14605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.148125394334857e-05,
+      "loss": 3.1244,
+      "step": 14610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.147555481518697e-05,
+      "loss": 2.8803,
+      "step": 14615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.146985395892263e-05,
+      "loss": 2.9557,
+      "step": 14620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.146415137479309e-05,
+      "loss": 3.0193,
+      "step": 14625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.145844706303593e-05,
+      "loss": 2.9869,
+      "step": 14630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.145274102388883e-05,
+      "loss": 3.0293,
+      "step": 14635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.144703325758955e-05,
+      "loss": 2.889,
+      "step": 14640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.144132376437589e-05,
+      "loss": 2.7007,
+      "step": 14645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.143561254448575e-05,
+      "loss": 2.9577,
+      "step": 14650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.142989959815708e-05,
+      "loss": 2.8537,
+      "step": 14655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.142418492562794e-05,
+      "loss": 2.8993,
+      "step": 14660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.14184685271364e-05,
+      "loss": 2.8936,
+      "step": 14665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.141275040292065e-05,
+      "loss": 2.8729,
+      "step": 14670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.140703055321896e-05,
+      "loss": 2.9589,
+      "step": 14675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.140130897826959e-05,
+      "loss": 3.0099,
+      "step": 14680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.139558567831101e-05,
+      "loss": 2.8174,
+      "step": 14685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.138986065358166e-05,
+      "loss": 3.0164,
+      "step": 14690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.138413390432004e-05,
+      "loss": 2.9493,
+      "step": 14695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.13784054307648e-05,
+      "loss": 2.9233,
+      "step": 14700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.13726752331546e-05,
+      "loss": 3.0215,
+      "step": 14705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.136694331172819e-05,
+      "loss": 2.8687,
+      "step": 14710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.136120966672441e-05,
+      "loss": 2.9628,
+      "step": 14715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.135547429838214e-05,
+      "loss": 2.9466,
+      "step": 14720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.134973720694036e-05,
+      "loss": 3.0167,
+      "step": 14725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.13439983926381e-05,
+      "loss": 2.9754,
+      "step": 14730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.133825785571447e-05,
+      "loss": 2.9759,
+      "step": 14735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.133251559640865e-05,
+      "loss": 2.9889,
+      "step": 14740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.132677161495991e-05,
+      "loss": 3.0146,
+      "step": 14745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.132102591160758e-05,
+      "loss": 2.8833,
+      "step": 14750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.131527848659102e-05,
+      "loss": 2.8721,
+      "step": 14755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.130952934014973e-05,
+      "loss": 2.8025,
+      "step": 14760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.130377847252328e-05,
+      "loss": 3.0597,
+      "step": 14765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.12980258839512e-05,
+      "loss": 3.0459,
+      "step": 14770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.129227157467325e-05,
+      "loss": 2.9425,
+      "step": 14775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.128651554492917e-05,
+      "loss": 2.9414,
+      "step": 14780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.128075779495875e-05,
+      "loss": 2.8727,
+      "step": 14785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.127499832500195e-05,
+      "loss": 2.9526,
+      "step": 14790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.126923713529868e-05,
+      "loss": 2.9644,
+      "step": 14795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.126347422608905e-05,
+      "loss": 2.8824,
+      "step": 14800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.12577095976131e-05,
+      "loss": 2.941,
+      "step": 14805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.125194325011109e-05,
+      "loss": 2.8629,
+      "step": 14810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.124617518382323e-05,
+      "loss": 2.9381,
+      "step": 14815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.124040539898986e-05,
+      "loss": 2.9214,
+      "step": 14820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.123463389585138e-05,
+      "loss": 2.9813,
+      "step": 14825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.122886067464829e-05,
+      "loss": 2.8459,
+      "step": 14830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.12230857356211e-05,
+      "loss": 2.9658,
+      "step": 14835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.121730907901042e-05,
+      "loss": 2.9205,
+      "step": 14840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.121153070505699e-05,
+      "loss": 2.8212,
+      "step": 14845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.120575061400152e-05,
+      "loss": 2.9313,
+      "step": 14850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.119996880608485e-05,
+      "loss": 2.9562,
+      "step": 14855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.11941852815479e-05,
+      "loss": 2.9223,
+      "step": 14860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.118840004063162e-05,
+      "loss": 2.9886,
+      "step": 14865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.118261308357708e-05,
+      "loss": 2.9196,
+      "step": 14870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.117682441062539e-05,
+      "loss": 2.8797,
+      "step": 14875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.11710340220177e-05,
+      "loss": 3.0169,
+      "step": 14880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.116524191799533e-05,
+      "loss": 2.9184,
+      "step": 14885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.115944809879959e-05,
+      "loss": 2.9502,
+      "step": 14890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.115365256467187e-05,
+      "loss": 3.0194,
+      "step": 14895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.114785531585364e-05,
+      "loss": 2.9836,
+      "step": 14900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.114205635258647e-05,
+      "loss": 2.9222,
+      "step": 14905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.113625567511196e-05,
+      "loss": 2.9104,
+      "step": 14910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.11304532836718e-05,
+      "loss": 3.0078,
+      "step": 14915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.112464917850776e-05,
+      "loss": 2.988,
+      "step": 14920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.111884335986166e-05,
+      "loss": 2.9713,
+      "step": 14925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.111303582797541e-05,
+      "loss": 3.0201,
+      "step": 14930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.110722658309097e-05,
+      "loss": 2.9177,
+      "step": 14935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.110141562545042e-05,
+      "loss": 2.8958,
+      "step": 14940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.109560295529583e-05,
+      "loss": 2.8591,
+      "step": 14945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.108978857286943e-05,
+      "loss": 2.8368,
+      "step": 14950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.108397247841345e-05,
+      "loss": 2.8887,
+      "step": 14955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.107815467217022e-05,
+      "loss": 2.9913,
+      "step": 14960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.107233515438218e-05,
+      "loss": 2.971,
+      "step": 14965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.106651392529177e-05,
+      "loss": 2.8674,
+      "step": 14970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.106069098514154e-05,
+      "loss": 2.9155,
+      "step": 14975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.10548663341741e-05,
+      "loss": 2.9289,
+      "step": 14980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.104903997263215e-05,
+      "loss": 2.8678,
+      "step": 14985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.104321190075846e-05,
+      "loss": 2.8147,
+      "step": 14990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.103738211879582e-05,
+      "loss": 2.9369,
+      "step": 14995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.103155062698718e-05,
+      "loss": 2.914,
+      "step": 15000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.102571742557548e-05,
+      "loss": 2.9289,
+      "step": 15005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.101988251480377e-05,
+      "loss": 2.8827,
+      "step": 15010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.101404589491517e-05,
+      "loss": 2.8012,
+      "step": 15015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.100820756615286e-05,
+      "loss": 2.8695,
+      "step": 15020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.10023675287601e-05,
+      "loss": 3.0379,
+      "step": 15025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.099652578298022e-05,
+      "loss": 2.9063,
+      "step": 15030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.099068232905663e-05,
+      "loss": 2.9357,
+      "step": 15035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.098483716723276e-05,
+      "loss": 2.9693,
+      "step": 15040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.09789902977522e-05,
+      "loss": 2.924,
+      "step": 15045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.097314172085856e-05,
+      "loss": 2.9657,
+      "step": 15050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.09672914367955e-05,
+      "loss": 3.0392,
+      "step": 15055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.096143944580677e-05,
+      "loss": 3.0439,
+      "step": 15060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.095558574813623e-05,
+      "loss": 2.9132,
+      "step": 15065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.094973034402774e-05,
+      "loss": 2.8257,
+      "step": 15070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.09438732337253e-05,
+      "loss": 2.9525,
+      "step": 15075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.093801441747294e-05,
+      "loss": 2.8485,
+      "step": 15080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.093215389551477e-05,
+      "loss": 2.9232,
+      "step": 15085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.092629166809495e-05,
+      "loss": 2.9656,
+      "step": 15090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.092042773545777e-05,
+      "loss": 2.9171,
+      "step": 15095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.091456209784755e-05,
+      "loss": 2.9534,
+      "step": 15100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.090869475550865e-05,
+      "loss": 3.0913,
+      "step": 15105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.090282570868555e-05,
+      "loss": 2.9505,
+      "step": 15110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.089695495762283e-05,
+      "loss": 3.019,
+      "step": 15115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.089108250256504e-05,
+      "loss": 2.8725,
+      "step": 15120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.088520834375689e-05,
+      "loss": 3.0222,
+      "step": 15125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.087933248144313e-05,
+      "loss": 2.8959,
+      "step": 15130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.087345491586856e-05,
+      "loss": 2.7822,
+      "step": 15135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.08675756472781e-05,
+      "loss": 2.9564,
+      "step": 15140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.08616946759167e-05,
+      "loss": 2.8417,
+      "step": 15145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.085581200202937e-05,
+      "loss": 2.8523,
+      "step": 15150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.084992762586127e-05,
+      "loss": 2.9134,
+      "step": 15155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.084404154765753e-05,
+      "loss": 2.8156,
+      "step": 15160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.083815376766342e-05,
+      "loss": 2.8755,
+      "step": 15165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.083226428612424e-05,
+      "loss": 2.9008,
+      "step": 15170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.082637310328538e-05,
+      "loss": 2.9002,
+      "step": 15175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.082048021939233e-05,
+      "loss": 2.9642,
+      "step": 15180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.081458563469059e-05,
+      "loss": 3.0325,
+      "step": 15185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.080868934942577e-05,
+      "loss": 2.9166,
+      "step": 15190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.080279136384354e-05,
+      "loss": 3.04,
+      "step": 15195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.079689167818965e-05,
+      "loss": 2.9301,
+      "step": 15200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.07909902927099e-05,
+      "loss": 2.7821,
+      "step": 15205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.078508720765018e-05,
+      "loss": 2.8989,
+      "step": 15210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.077918242325645e-05,
+      "loss": 2.9468,
+      "step": 15215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.077327593977474e-05,
+      "loss": 2.88,
+      "step": 15220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.076736775745114e-05,
+      "loss": 2.963,
+      "step": 15225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.076145787653181e-05,
+      "loss": 2.8988,
+      "step": 15230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.075554629726302e-05,
+      "loss": 2.9213,
+      "step": 15235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.074963301989103e-05,
+      "loss": 3.0474,
+      "step": 15240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.074371804466226e-05,
+      "loss": 2.8621,
+      "step": 15245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.073780137182314e-05,
+      "loss": 2.9533,
+      "step": 15250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.073188300162021e-05,
+      "loss": 2.8962,
+      "step": 15255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.072596293430004e-05,
+      "loss": 2.9426,
+      "step": 15260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.072004117010933e-05,
+      "loss": 2.7958,
+      "step": 15265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.071411770929477e-05,
+      "loss": 2.8594,
+      "step": 15270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.07081925521032e-05,
+      "loss": 2.9323,
+      "step": 15275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.070226569878144e-05,
+      "loss": 2.8143,
+      "step": 15280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.06963371495765e-05,
+      "loss": 2.9239,
+      "step": 15285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.069040690473537e-05,
+      "loss": 2.9223,
+      "step": 15290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.068447496450513e-05,
+      "loss": 2.8835,
+      "step": 15295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.067854132913294e-05,
+      "loss": 2.9752,
+      "step": 15300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.067260599886604e-05,
+      "loss": 2.8083,
+      "step": 15305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.066666897395173e-05,
+      "loss": 2.9604,
+      "step": 15310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.066073025463738e-05,
+      "loss": 2.9557,
+      "step": 15315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.065478984117039e-05,
+      "loss": 2.9215,
+      "step": 15320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.064884773379832e-05,
+      "loss": 2.8224,
+      "step": 15325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.064290393276873e-05,
+      "loss": 2.8688,
+      "step": 15330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.063695843832929e-05,
+      "loss": 2.9486,
+      "step": 15335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.063101125072769e-05,
+      "loss": 2.9227,
+      "step": 15340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.062506237021176e-05,
+      "loss": 2.9268,
+      "step": 15345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.061911179702934e-05,
+      "loss": 2.8142,
+      "step": 15350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.061315953142835e-05,
+      "loss": 2.971,
+      "step": 15355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.060720557365683e-05,
+      "loss": 3.0382,
+      "step": 15360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.060124992396283e-05,
+      "loss": 2.9673,
+      "step": 15365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.059529258259449e-05,
+      "loss": 2.8966,
+      "step": 15370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.058933354980005e-05,
+      "loss": 2.7845,
+      "step": 15375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.058337282582781e-05,
+      "loss": 2.7952,
+      "step": 15380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.057741041092609e-05,
+      "loss": 2.9584,
+      "step": 15385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.057144630534332e-05,
+      "loss": 2.9031,
+      "step": 15390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.056548050932803e-05,
+      "loss": 2.829,
+      "step": 15395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.055951302312875e-05,
+      "loss": 2.952,
+      "step": 15400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.055354384699413e-05,
+      "loss": 2.9091,
+      "step": 15405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.05475729811729e-05,
+      "loss": 2.9191,
+      "step": 15410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.054160042591381e-05,
+      "loss": 2.8856,
+      "step": 15415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.053562618146573e-05,
+      "loss": 2.8697,
+      "step": 15420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.052965024807758e-05,
+      "loss": 2.9969,
+      "step": 15425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.052367262599834e-05,
+      "loss": 2.811,
+      "step": 15430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.051769331547708e-05,
+      "loss": 2.9699,
+      "step": 15435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.051171231676292e-05,
+      "loss": 2.8996,
+      "step": 15440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.050572963010509e-05,
+      "loss": 2.8619,
+      "step": 15445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.049974525575282e-05,
+      "loss": 3.024,
+      "step": 15450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.049375919395548e-05,
+      "loss": 2.793,
+      "step": 15455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.048777144496248e-05,
+      "loss": 2.8259,
+      "step": 15460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.048178200902328e-05,
+      "loss": 3.0531,
+      "step": 15465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.047579088638747e-05,
+      "loss": 2.941,
+      "step": 15470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.046979807730467e-05,
+      "loss": 2.828,
+      "step": 15475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.046380358202454e-05,
+      "loss": 2.9694,
+      "step": 15480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.045780740079687e-05,
+      "loss": 2.8758,
+      "step": 15485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.045180953387149e-05,
+      "loss": 2.9455,
+      "step": 15490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.04458099814983e-05,
+      "loss": 2.9017,
+      "step": 15495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.043980874392728e-05,
+      "loss": 2.8909,
+      "step": 15500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.043380582140848e-05,
+      "loss": 2.8828,
+      "step": 15505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.0427801214192e-05,
+      "loss": 2.8408,
+      "step": 15510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.042179492252802e-05,
+      "loss": 2.7602,
+      "step": 15515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.041578694666684e-05,
+      "loss": 2.8463,
+      "step": 15520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.040977728685873e-05,
+      "loss": 2.8985,
+      "step": 15525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.040376594335412e-05,
+      "loss": 2.909,
+      "step": 15530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.039775291640347e-05,
+      "loss": 2.7874,
+      "step": 15535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.039173820625732e-05,
+      "loss": 2.8352,
+      "step": 15540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.038572181316624e-05,
+      "loss": 2.9311,
+      "step": 15545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.037970373738096e-05,
+      "loss": 2.836,
+      "step": 15550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.03736839791522e-05,
+      "loss": 2.8369,
+      "step": 15555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.036766253873078e-05,
+      "loss": 3.0102,
+      "step": 15560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.036163941636759e-05,
+      "loss": 2.8303,
+      "step": 15565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.035561461231357e-05,
+      "loss": 3.0345,
+      "step": 15570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.034958812681977e-05,
+      "loss": 2.9677,
+      "step": 15575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.034355996013729e-05,
+      "loss": 2.9675,
+      "step": 15580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.033753011251725e-05,
+      "loss": 2.8437,
+      "step": 15585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.033149858421094e-05,
+      "loss": 3.0561,
+      "step": 15590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.032546537546965e-05,
+      "loss": 2.9947,
+      "step": 15595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.031943048654474e-05,
+      "loss": 2.8889,
+      "step": 15600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.031339391768768e-05,
+      "loss": 2.8406,
+      "step": 15605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.030735566914999e-05,
+      "loss": 2.9709,
+      "step": 15610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.030131574118324e-05,
+      "loss": 2.9851,
+      "step": 15615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.029527413403909e-05,
+      "loss": 2.9361,
+      "step": 15620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.028923084796927e-05,
+      "loss": 2.9182,
+      "step": 15625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.028318588322558e-05,
+      "loss": 2.8493,
+      "step": 15630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.027713924005987e-05,
+      "loss": 3.025,
+      "step": 15635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.02710909187241e-05,
+      "loss": 2.9706,
+      "step": 15640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.026504091947028e-05,
+      "loss": 2.8667,
+      "step": 15645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.025898924255047e-05,
+      "loss": 2.9226,
+      "step": 15650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.025293588821682e-05,
+      "loss": 3.0279,
+      "step": 15655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.024688085672152e-05,
+      "loss": 2.8137,
+      "step": 15660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.024082414831691e-05,
+      "loss": 2.8758,
+      "step": 15665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.023476576325534e-05,
+      "loss": 2.8617,
+      "step": 15670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.022870570178918e-05,
+      "loss": 2.8106,
+      "step": 15675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.022264396417097e-05,
+      "loss": 2.9286,
+      "step": 15680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.021658055065328e-05,
+      "loss": 2.9158,
+      "step": 15685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.021051546148874e-05,
+      "loss": 2.8751,
+      "step": 15690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.020444869693003e-05,
+      "loss": 2.9731,
+      "step": 15695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.019838025722995e-05,
+      "loss": 2.9774,
+      "step": 15700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.019231014264134e-05,
+      "loss": 2.8865,
+      "step": 15705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.01862383534171e-05,
+      "loss": 2.8585,
+      "step": 15710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.018016488981025e-05,
+      "loss": 3.0217,
+      "step": 15715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.017408975207382e-05,
+      "loss": 2.7702,
+      "step": 15720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.016801294046094e-05,
+      "loss": 2.8855,
+      "step": 15725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.016193445522478e-05,
+      "loss": 2.9458,
+      "step": 15730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.015585429661864e-05,
+      "loss": 2.8603,
+      "step": 15735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.014977246489585e-05,
+      "loss": 2.9666,
+      "step": 15740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.014368896030978e-05,
+      "loss": 2.8277,
+      "step": 15745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.013760378311393e-05,
+      "loss": 2.9125,
+      "step": 15750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.013151693356183e-05,
+      "loss": 2.8727,
+      "step": 15755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.01254284119071e-05,
+      "loss": 2.8955,
+      "step": 15760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.011933821840342e-05,
+      "loss": 2.8829,
+      "step": 15765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.011324635330455e-05,
+      "loss": 2.9138,
+      "step": 15770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.010715281686429e-05,
+      "loss": 2.8526,
+      "step": 15775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.010105760933656e-05,
+      "loss": 2.8117,
+      "step": 15780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.009496073097528e-05,
+      "loss": 2.8872,
+      "step": 15785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.008886218203452e-05,
+      "loss": 2.8384,
+      "step": 15790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.008276196276835e-05,
+      "loss": 2.9459,
+      "step": 15795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.007666007343097e-05,
+      "loss": 2.8583,
+      "step": 15800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.007055651427658e-05,
+      "loss": 2.9844,
+      "step": 15805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.006445128555951e-05,
+      "loss": 2.9027,
+      "step": 15810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.005834438753415e-05,
+      "loss": 2.8196,
+      "step": 15815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.005223582045492e-05,
+      "loss": 2.8459,
+      "step": 15820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.004612558457637e-05,
+      "loss": 2.8454,
+      "step": 15825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.004001368015305e-05,
+      "loss": 2.8003,
+      "step": 15830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.003390010743964e-05,
+      "loss": 3.07,
+      "step": 15835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.002778486669086e-05,
+      "loss": 2.9711,
+      "step": 15840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.002166795816149e-05,
+      "loss": 2.9453,
+      "step": 15845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.001554938210643e-05,
+      "loss": 2.8381,
+      "step": 15850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.000942913878058e-05,
+      "loss": 2.9014,
+      "step": 15855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.000330722843895e-05,
+      "loss": 2.927,
+      "step": 15860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.999718365133662e-05,
+      "loss": 2.8709,
+      "step": 15865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.999105840772874e-05,
+      "loss": 2.9192,
+      "step": 15870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.998493149787052e-05,
+      "loss": 2.8193,
+      "step": 15875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.997880292201722e-05,
+      "loss": 2.815,
+      "step": 15880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.997267268042419e-05,
+      "loss": 2.912,
+      "step": 15885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.996654077334689e-05,
+      "loss": 2.902,
+      "step": 15890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.996040720104079e-05,
+      "loss": 2.9041,
+      "step": 15895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.995427196376143e-05,
+      "loss": 3.0575,
+      "step": 15900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.994813506176445e-05,
+      "loss": 2.9191,
+      "step": 15905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.994199649530554e-05,
+      "loss": 2.88,
+      "step": 15910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.993585626464049e-05,
+      "loss": 2.9582,
+      "step": 15915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.992971437002512e-05,
+      "loss": 2.8531,
+      "step": 15920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.992357081171532e-05,
+      "loss": 2.8788,
+      "step": 15925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.99174255899671e-05,
+      "loss": 2.8374,
+      "step": 15930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.991127870503648e-05,
+      "loss": 2.9372,
+      "step": 15935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.990513015717959e-05,
+      "loss": 2.9695,
+      "step": 15940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.989897994665259e-05,
+      "loss": 2.8146,
+      "step": 15945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.989282807371175e-05,
+      "loss": 2.8822,
+      "step": 15950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.988667453861337e-05,
+      "loss": 2.9706,
+      "step": 15955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.988051934161388e-05,
+      "loss": 2.9698,
+      "step": 15960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.987436248296969e-05,
+      "loss": 2.9162,
+      "step": 15965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.986820396293737e-05,
+      "loss": 2.8656,
+      "step": 15970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.986204378177352e-05,
+      "loss": 2.8815,
+      "step": 15975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.985588193973477e-05,
+      "loss": 2.903,
+      "step": 15980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.984971843707788e-05,
+      "loss": 2.8543,
+      "step": 15985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.984355327405966e-05,
+      "loss": 3.0389,
+      "step": 15990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.983738645093698e-05,
+      "loss": 2.8891,
+      "step": 15995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.983121796796677e-05,
+      "loss": 2.9022,
+      "step": 16000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.982504782540607e-05,
+      "loss": 2.8749,
+      "step": 16005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.981887602351195e-05,
+      "loss": 3.0147,
+      "step": 16010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.981270256254155e-05,
+      "loss": 2.7061,
+      "step": 16015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.980652744275211e-05,
+      "loss": 2.7784,
+      "step": 16020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.98003506644009e-05,
+      "loss": 2.8237,
+      "step": 16025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.979417222774529e-05,
+      "loss": 2.8773,
+      "step": 16030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.978799213304273e-05,
+      "loss": 2.8614,
+      "step": 16035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.978181038055067e-05,
+      "loss": 2.8638,
+      "step": 16040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.977562697052671e-05,
+      "loss": 2.9791,
+      "step": 16045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.976944190322847e-05,
+      "loss": 2.8412,
+      "step": 16050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.976325517891367e-05,
+      "loss": 2.9321,
+      "step": 16055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.975706679784009e-05,
+      "loss": 2.9026,
+      "step": 16060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.975087676026552e-05,
+      "loss": 2.7902,
+      "step": 16065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.974468506644795e-05,
+      "loss": 2.8182,
+      "step": 16070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.973849171664529e-05,
+      "loss": 2.8143,
+      "step": 16075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.973229671111563e-05,
+      "loss": 2.8454,
+      "step": 16080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.972610005011706e-05,
+      "loss": 3.0309,
+      "step": 16085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.971990173390781e-05,
+      "loss": 2.954,
+      "step": 16090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.97137017627461e-05,
+      "loss": 2.8431,
+      "step": 16095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.970750013689026e-05,
+      "loss": 2.9447,
+      "step": 16100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.97012968565987e-05,
+      "loss": 2.8812,
+      "step": 16105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.969509192212988e-05,
+      "loss": 2.8988,
+      "step": 16110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.968888533374231e-05,
+      "loss": 3.0051,
+      "step": 16115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.96826770916946e-05,
+      "loss": 2.8034,
+      "step": 16120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.967646719624545e-05,
+      "loss": 2.9661,
+      "step": 16125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.967025564765357e-05,
+      "loss": 2.8991,
+      "step": 16130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.966404244617777e-05,
+      "loss": 3.024,
+      "step": 16135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.965782759207691e-05,
+      "loss": 2.8936,
+      "step": 16140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.965161108560998e-05,
+      "loss": 2.919,
+      "step": 16145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.964539292703596e-05,
+      "loss": 2.983,
+      "step": 16150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.963917311661394e-05,
+      "loss": 2.9676,
+      "step": 16155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.963295165460309e-05,
+      "loss": 2.9163,
+      "step": 16160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.962672854126261e-05,
+      "loss": 2.9913,
+      "step": 16165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.962050377685178e-05,
+      "loss": 2.8432,
+      "step": 16170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.961427736163e-05,
+      "loss": 2.8438,
+      "step": 16175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.960804929585665e-05,
+      "loss": 2.9563,
+      "step": 16180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.960181957979125e-05,
+      "loss": 2.8772,
+      "step": 16185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.959558821369336e-05,
+      "loss": 2.9206,
+      "step": 16190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.958935519782264e-05,
+      "loss": 2.8078,
+      "step": 16195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.958312053243874e-05,
+      "loss": 2.8722,
+      "step": 16200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.957688421780146e-05,
+      "loss": 2.8105,
+      "step": 16205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.957064625417065e-05,
+      "loss": 2.8047,
+      "step": 16210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.956440664180619e-05,
+      "loss": 2.8313,
+      "step": 16215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.955816538096807e-05,
+      "loss": 3.0411,
+      "step": 16220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.955192247191636e-05,
+      "loss": 2.796,
+      "step": 16225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.954567791491113e-05,
+      "loss": 2.8191,
+      "step": 16230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.953943171021259e-05,
+      "loss": 2.8839,
+      "step": 16235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.9533183858081e-05,
+      "loss": 2.922,
+      "step": 16240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.952693435877667e-05,
+      "loss": 2.8911,
+      "step": 16245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.952068321255997e-05,
+      "loss": 2.8841,
+      "step": 16250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.95144304196914e-05,
+      "loss": 2.8589,
+      "step": 16255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.950817598043144e-05,
+      "loss": 2.7504,
+      "step": 16260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.950191989504071e-05,
+      "loss": 2.7807,
+      "step": 16265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.949566216377987e-05,
+      "loss": 2.9462,
+      "step": 16270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.948940278690964e-05,
+      "loss": 2.8994,
+      "step": 16275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.948314176469086e-05,
+      "loss": 3.0676,
+      "step": 16280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.947687909738435e-05,
+      "loss": 2.9107,
+      "step": 16285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.947061478525108e-05,
+      "loss": 2.8424,
+      "step": 16290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.946434882855203e-05,
+      "loss": 2.804,
+      "step": 16295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.945808122754828e-05,
+      "loss": 2.7644,
+      "step": 16300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.9451811982501e-05,
+      "loss": 2.8552,
+      "step": 16305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.944554109367138e-05,
+      "loss": 2.8331,
+      "step": 16310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.94392685613207e-05,
+      "loss": 2.9594,
+      "step": 16315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.943299438571034e-05,
+      "loss": 2.8306,
+      "step": 16320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.942671856710165e-05,
+      "loss": 2.8448,
+      "step": 16325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.942044110575616e-05,
+      "loss": 2.9006,
+      "step": 16330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.941416200193545e-05,
+      "loss": 2.9707,
+      "step": 16335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.940788125590108e-05,
+      "loss": 2.9215,
+      "step": 16340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.940159886791479e-05,
+      "loss": 2.8925,
+      "step": 16345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.939531483823832e-05,
+      "loss": 2.8906,
+      "step": 16350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.938902916713349e-05,
+      "loss": 2.9317,
+      "step": 16355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.938274185486221e-05,
+      "loss": 2.8505,
+      "step": 16360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.937645290168646e-05,
+      "loss": 2.7558,
+      "step": 16365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.937016230786823e-05,
+      "loss": 2.9905,
+      "step": 16370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.936387007366964e-05,
+      "loss": 2.8918,
+      "step": 16375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.935757619935287e-05,
+      "loss": 2.7478,
+      "step": 16380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.935128068518017e-05,
+      "loss": 2.8537,
+      "step": 16385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.93449835314138e-05,
+      "loss": 3.0259,
+      "step": 16390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.93386847383162e-05,
+      "loss": 2.8475,
+      "step": 16395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.933238430614975e-05,
+      "loss": 2.7053,
+      "step": 16400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.932608223517698e-05,
+      "loss": 2.8522,
+      "step": 16405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.931977852566049e-05,
+      "loss": 2.6832,
+      "step": 16410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.931347317786289e-05,
+      "loss": 2.6661,
+      "step": 16415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.930716619204693e-05,
+      "loss": 3.0132,
+      "step": 16420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.93008575684754e-05,
+      "loss": 2.9634,
+      "step": 16425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.929454730741112e-05,
+      "loss": 2.8223,
+      "step": 16430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.928823540911703e-05,
+      "loss": 2.7992,
+      "step": 16435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.928192187385612e-05,
+      "loss": 2.9791,
+      "step": 16440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.927560670189143e-05,
+      "loss": 2.8077,
+      "step": 16445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.926928989348612e-05,
+      "loss": 2.9276,
+      "step": 16450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.926297144890335e-05,
+      "loss": 3.0133,
+      "step": 16455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.925665136840638e-05,
+      "loss": 2.8208,
+      "step": 16460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.925032965225856e-05,
+      "loss": 2.9324,
+      "step": 16465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.924400630072329e-05,
+      "loss": 2.7052,
+      "step": 16470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.923768131406403e-05,
+      "loss": 2.818,
+      "step": 16475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.92313546925443e-05,
+      "loss": 3.0587,
+      "step": 16480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.922502643642771e-05,
+      "loss": 2.8372,
+      "step": 16485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.921869654597795e-05,
+      "loss": 2.9495,
+      "step": 16490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.921236502145872e-05,
+      "loss": 2.826,
+      "step": 16495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.920603186313385e-05,
+      "loss": 3.0891,
+      "step": 16500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.919969707126723e-05,
+      "loss": 2.8804,
+      "step": 16505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.919336064612279e-05,
+      "loss": 2.8476,
+      "step": 16510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.918702258796451e-05,
+      "loss": 2.8837,
+      "step": 16515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.918068289705651e-05,
+      "loss": 2.9902,
+      "step": 16520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.917434157366292e-05,
+      "loss": 2.7945,
+      "step": 16525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.916799861804796e-05,
+      "loss": 2.8832,
+      "step": 16530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.916165403047592e-05,
+      "loss": 2.9331,
+      "step": 16535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.915530781121111e-05,
+      "loss": 2.9591,
+      "step": 16540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.914895996051801e-05,
+      "loss": 2.98,
+      "step": 16545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.914261047866107e-05,
+      "loss": 2.939,
+      "step": 16550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.913625936590483e-05,
+      "loss": 2.7975,
+      "step": 16555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.912990662251397e-05,
+      "loss": 2.8329,
+      "step": 16560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.912355224875312e-05,
+      "loss": 2.9169,
+      "step": 16565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.911719624488707e-05,
+      "loss": 2.8691,
+      "step": 16570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.911083861118065e-05,
+      "loss": 2.7935,
+      "step": 16575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.910447934789874e-05,
+      "loss": 2.8689,
+      "step": 16580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.909811845530631e-05,
+      "loss": 2.7872,
+      "step": 16585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.909175593366837e-05,
+      "loss": 2.9389,
+      "step": 16590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.908539178325004e-05,
+      "loss": 2.9325,
+      "step": 16595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.907902600431649e-05,
+      "loss": 2.838,
+      "step": 16600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.907265859713294e-05,
+      "loss": 2.9514,
+      "step": 16605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.90662895619647e-05,
+      "loss": 2.8518,
+      "step": 16610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.905991889907714e-05,
+      "loss": 2.8455,
+      "step": 16615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.905354660873568e-05,
+      "loss": 2.8199,
+      "step": 16620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.904717269120585e-05,
+      "loss": 2.9202,
+      "step": 16625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.90407971467532e-05,
+      "loss": 2.8359,
+      "step": 16630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.903441997564339e-05,
+      "loss": 2.9073,
+      "step": 16635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.902804117814212e-05,
+      "loss": 2.9412,
+      "step": 16640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.902166075451516e-05,
+      "loss": 2.8896,
+      "step": 16645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.901527870502838e-05,
+      "loss": 2.9635,
+      "step": 16650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.900889502994767e-05,
+      "loss": 2.9268,
+      "step": 16655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.9002509729539e-05,
+      "loss": 2.8988,
+      "step": 16660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.899612280406844e-05,
+      "loss": 2.8825,
+      "step": 16665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.898973425380209e-05,
+      "loss": 2.9687,
+      "step": 16670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.898334407900613e-05,
+      "loss": 2.92,
+      "step": 16675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.897695227994683e-05,
+      "loss": 2.7781,
+      "step": 16680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.897055885689048e-05,
+      "loss": 2.8756,
+      "step": 16685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.896416381010349e-05,
+      "loss": 2.9502,
+      "step": 16690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.895776713985229e-05,
+      "loss": 2.8507,
+      "step": 16695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.895136884640342e-05,
+      "loss": 2.9723,
+      "step": 16700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.894496893002349e-05,
+      "loss": 2.8798,
+      "step": 16705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.89385673909791e-05,
+      "loss": 2.8736,
+      "step": 16710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.893216422953701e-05,
+      "loss": 2.9239,
+      "step": 16715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.892575944596402e-05,
+      "loss": 2.9681,
+      "step": 16720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.891935304052697e-05,
+      "loss": 2.908,
+      "step": 16725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.891294501349277e-05,
+      "loss": 2.9615,
+      "step": 16730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.890653536512846e-05,
+      "loss": 2.956,
+      "step": 16735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.890012409570106e-05,
+      "loss": 2.8714,
+      "step": 16740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.889371120547772e-05,
+      "loss": 2.8532,
+      "step": 16745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.888729669472564e-05,
+      "loss": 2.8386,
+      "step": 16750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.888088056371205e-05,
+      "loss": 2.8451,
+      "step": 16755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.887446281270433e-05,
+      "loss": 2.8293,
+      "step": 16760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.886804344196986e-05,
+      "loss": 2.8751,
+      "step": 16765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.886162245177609e-05,
+      "loss": 2.8371,
+      "step": 16770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.885519984239059e-05,
+      "loss": 3.1287,
+      "step": 16775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.884877561408092e-05,
+      "loss": 2.913,
+      "step": 16780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.884234976711478e-05,
+      "loss": 2.9555,
+      "step": 16785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.883592230175989e-05,
+      "loss": 2.9782,
+      "step": 16790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.882949321828407e-05,
+      "loss": 2.8989,
+      "step": 16795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.882306251695517e-05,
+      "loss": 2.9374,
+      "step": 16800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.881663019804114e-05,
+      "loss": 2.8806,
+      "step": 16805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.881019626181e-05,
+      "loss": 2.8253,
+      "step": 16810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.880376070852979e-05,
+      "loss": 2.9029,
+      "step": 16815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.879732353846869e-05,
+      "loss": 2.7704,
+      "step": 16820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.879088475189487e-05,
+      "loss": 2.95,
+      "step": 16825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.878444434907664e-05,
+      "loss": 2.8987,
+      "step": 16830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.877800233028233e-05,
+      "loss": 2.8977,
+      "step": 16835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.877155869578035e-05,
+      "loss": 2.8853,
+      "step": 16840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.876511344583917e-05,
+      "loss": 2.8497,
+      "step": 16845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.875866658072734e-05,
+      "loss": 2.8688,
+      "step": 16850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.875221810071348e-05,
+      "loss": 2.8553,
+      "step": 16855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.874576800606626e-05,
+      "loss": 2.9842,
+      "step": 16860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.873931629705443e-05,
+      "loss": 2.9105,
+      "step": 16865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.873286297394681e-05,
+      "loss": 2.9301,
+      "step": 16870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.872640803701226e-05,
+      "loss": 2.873,
+      "step": 16875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.871995148651976e-05,
+      "loss": 2.8362,
+      "step": 16880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.871349332273831e-05,
+      "loss": 2.8763,
+      "step": 16885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.870703354593699e-05,
+      "loss": 2.7661,
+      "step": 16890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.870057215638495e-05,
+      "loss": 3.0354,
+      "step": 16895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.869410915435141e-05,
+      "loss": 2.9536,
+      "step": 16900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.868764454010567e-05,
+      "loss": 2.9087,
+      "step": 16905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.868117831391704e-05,
+      "loss": 2.798,
+      "step": 16910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.867471047605501e-05,
+      "loss": 2.8638,
+      "step": 16915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8668241026789e-05,
+      "loss": 2.7787,
+      "step": 16920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.866176996638858e-05,
+      "loss": 2.959,
+      "step": 16925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.865529729512336e-05,
+      "loss": 2.7137,
+      "step": 16930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.864882301326307e-05,
+      "loss": 2.9541,
+      "step": 16935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.864234712107744e-05,
+      "loss": 2.9074,
+      "step": 16940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.863586961883629e-05,
+      "loss": 2.862,
+      "step": 16945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.86293905068095e-05,
+      "loss": 2.8597,
+      "step": 16950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.862290978526704e-05,
+      "loss": 2.7837,
+      "step": 16955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.861642745447894e-05,
+      "loss": 2.9916,
+      "step": 16960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.860994351471526e-05,
+      "loss": 2.9221,
+      "step": 16965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.860345796624619e-05,
+      "loss": 2.8978,
+      "step": 16970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.859697080934192e-05,
+      "loss": 2.859,
+      "step": 16975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.859048204427279e-05,
+      "loss": 2.8512,
+      "step": 16980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.858399167130909e-05,
+      "loss": 2.8224,
+      "step": 16985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.857749969072131e-05,
+      "loss": 2.9185,
+      "step": 16990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.85710061027799e-05,
+      "loss": 2.988,
+      "step": 16995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.856451090775546e-05,
+      "loss": 2.8352,
+      "step": 17000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.855801410591857e-05,
+      "loss": 2.9229,
+      "step": 17005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.855151569753997e-05,
+      "loss": 2.8027,
+      "step": 17010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.854501568289037e-05,
+      "loss": 3.0509,
+      "step": 17015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.853851406224064e-05,
+      "loss": 2.9355,
+      "step": 17020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.853201083586163e-05,
+      "loss": 2.9369,
+      "step": 17025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.852550600402434e-05,
+      "loss": 2.7855,
+      "step": 17030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.851899956699978e-05,
+      "loss": 3.0527,
+      "step": 17035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.851249152505905e-05,
+      "loss": 2.869,
+      "step": 17040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.85059818784733e-05,
+      "loss": 2.9918,
+      "step": 17045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.849947062751377e-05,
+      "loss": 2.8805,
+      "step": 17050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.849295777245174e-05,
+      "loss": 2.9256,
+      "step": 17055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.848644331355861e-05,
+      "loss": 2.9592,
+      "step": 17060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.847992725110576e-05,
+      "loss": 2.9143,
+      "step": 17065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.847340958536472e-05,
+      "loss": 2.9397,
+      "step": 17070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.846689031660705e-05,
+      "loss": 2.9005,
+      "step": 17075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.846036944510435e-05,
+      "loss": 2.8001,
+      "step": 17080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.845384697112834e-05,
+      "loss": 2.8021,
+      "step": 17085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.844732289495078e-05,
+      "loss": 2.8095,
+      "step": 17090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.84407972168435e-05,
+      "loss": 2.8114,
+      "step": 17095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.843426993707839e-05,
+      "loss": 2.8993,
+      "step": 17100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.842774105592741e-05,
+      "loss": 2.8481,
+      "step": 17105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.842121057366259e-05,
+      "loss": 2.825,
+      "step": 17110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.841467849055604e-05,
+      "loss": 2.8235,
+      "step": 17115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.840814480687992e-05,
+      "loss": 2.9125,
+      "step": 17120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.840160952290645e-05,
+      "loss": 3.0107,
+      "step": 17125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.839507263890793e-05,
+      "loss": 2.9077,
+      "step": 17130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.838853415515673e-05,
+      "loss": 2.7829,
+      "step": 17135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.838199407192527e-05,
+      "loss": 2.7941,
+      "step": 17140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.837545238948604e-05,
+      "loss": 2.805,
+      "step": 17145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.836890910811162e-05,
+      "loss": 2.8832,
+      "step": 17150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.836236422807463e-05,
+      "loss": 2.8324,
+      "step": 17155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.835581774964777e-05,
+      "loss": 2.8696,
+      "step": 17160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.83492696731038e-05,
+      "loss": 2.8016,
+      "step": 17165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.834271999871554e-05,
+      "loss": 2.9016,
+      "step": 17170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.833616872675592e-05,
+      "loss": 2.8227,
+      "step": 17175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.832961585749787e-05,
+      "loss": 2.8268,
+      "step": 17180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.832306139121442e-05,
+      "loss": 2.8977,
+      "step": 17185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.831650532817867e-05,
+      "loss": 2.7963,
+      "step": 17190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.830994766866381e-05,
+      "loss": 2.9186,
+      "step": 17195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.830338841294301e-05,
+      "loss": 2.8931,
+      "step": 17200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.829682756128963e-05,
+      "loss": 2.8895,
+      "step": 17205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.829026511397698e-05,
+      "loss": 2.8501,
+      "step": 17210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.828370107127853e-05,
+      "loss": 2.8494,
+      "step": 17215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.827713543346774e-05,
+      "loss": 2.9218,
+      "step": 17220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.827056820081818e-05,
+      "loss": 2.7861,
+      "step": 17225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.826399937360348e-05,
+      "loss": 2.9182,
+      "step": 17230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.825742895209736e-05,
+      "loss": 2.8202,
+      "step": 17235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.825085693657353e-05,
+      "loss": 2.8424,
+      "step": 17240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.824428332730585e-05,
+      "loss": 2.8548,
+      "step": 17245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.823770812456822e-05,
+      "loss": 3.0065,
+      "step": 17250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.823113132863457e-05,
+      "loss": 2.9638,
+      "step": 17255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.822455293977895e-05,
+      "loss": 2.8168,
+      "step": 17260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.821797295827546e-05,
+      "loss": 2.9583,
+      "step": 17265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.821139138439822e-05,
+      "loss": 2.7842,
+      "step": 17270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.820480821842148e-05,
+      "loss": 2.9012,
+      "step": 17275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.819822346061954e-05,
+      "loss": 2.8231,
+      "step": 17280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.819163711126675e-05,
+      "loss": 2.8972,
+      "step": 17285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.818504917063752e-05,
+      "loss": 2.7904,
+      "step": 17290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.817845963900635e-05,
+      "loss": 2.9186,
+      "step": 17295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.817186851664781e-05,
+      "loss": 2.7718,
+      "step": 17300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.81652758038365e-05,
+      "loss": 3.0046,
+      "step": 17305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.815868150084712e-05,
+      "loss": 2.8208,
+      "step": 17310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.815208560795443e-05,
+      "loss": 2.8402,
+      "step": 17315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.814548812543327e-05,
+      "loss": 2.9823,
+      "step": 17320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.813888905355848e-05,
+      "loss": 2.9105,
+      "step": 17325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.813228839260503e-05,
+      "loss": 2.9102,
+      "step": 17330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.812568614284797e-05,
+      "loss": 2.9017,
+      "step": 17335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.811908230456236e-05,
+      "loss": 2.9132,
+      "step": 17340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.811247687802336e-05,
+      "loss": 2.8966,
+      "step": 17345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.810586986350619e-05,
+      "loss": 2.7521,
+      "step": 17350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.809926126128611e-05,
+      "loss": 2.8688,
+      "step": 17355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.809265107163852e-05,
+      "loss": 2.8768,
+      "step": 17360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.808603929483881e-05,
+      "loss": 2.8705,
+      "step": 17365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.807942593116245e-05,
+      "loss": 2.8933,
+      "step": 17370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.807281098088502e-05,
+      "loss": 2.7228,
+      "step": 17375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.806619444428212e-05,
+      "loss": 2.9834,
+      "step": 17380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.805957632162942e-05,
+      "loss": 2.7948,
+      "step": 17385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.805295661320269e-05,
+      "loss": 2.8976,
+      "step": 17390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.804633531927773e-05,
+      "loss": 2.9996,
+      "step": 17395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.803971244013041e-05,
+      "loss": 2.8514,
+      "step": 17400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.803308797603671e-05,
+      "loss": 2.8535,
+      "step": 17405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.802646192727262e-05,
+      "loss": 2.8503,
+      "step": 17410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.801983429411421e-05,
+      "loss": 2.9074,
+      "step": 17415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.801320507683763e-05,
+      "loss": 2.8795,
+      "step": 17420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.800657427571913e-05,
+      "loss": 2.8703,
+      "step": 17425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.799994189103492e-05,
+      "loss": 2.8641,
+      "step": 17430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.799330792306137e-05,
+      "loss": 2.9932,
+      "step": 17435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.79866723720749e-05,
+      "loss": 2.8863,
+      "step": 17440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.798003523835197e-05,
+      "loss": 2.7035,
+      "step": 17445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.797339652216911e-05,
+      "loss": 2.7714,
+      "step": 17450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.796675622380297e-05,
+      "loss": 2.944,
+      "step": 17455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.796011434353017e-05,
+      "loss": 2.8594,
+      "step": 17460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.795347088162749e-05,
+      "loss": 2.9881,
+      "step": 17465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.794682583837169e-05,
+      "loss": 2.8614,
+      "step": 17470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.794017921403967e-05,
+      "loss": 2.9051,
+      "step": 17475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.793353100890838e-05,
+      "loss": 2.8608,
+      "step": 17480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.792688122325478e-05,
+      "loss": 2.9537,
+      "step": 17485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.792022985735595e-05,
+      "loss": 2.9788,
+      "step": 17490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.791357691148905e-05,
+      "loss": 2.7718,
+      "step": 17495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.790692238593126e-05,
+      "loss": 2.7893,
+      "step": 17500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.790026628095982e-05,
+      "loss": 2.9567,
+      "step": 17505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.78936085968521e-05,
+      "loss": 2.8126,
+      "step": 17510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.788694933388549e-05,
+      "loss": 2.8754,
+      "step": 17515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.788028849233742e-05,
+      "loss": 2.8215,
+      "step": 17520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.787362607248544e-05,
+      "loss": 2.8757,
+      "step": 17525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.786696207460716e-05,
+      "loss": 2.922,
+      "step": 17530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.786029649898022e-05,
+      "loss": 2.8743,
+      "step": 17535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.785362934588233e-05,
+      "loss": 2.8448,
+      "step": 17540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.784696061559132e-05,
+      "loss": 2.9094,
+      "step": 17545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.784029030838501e-05,
+      "loss": 2.7929,
+      "step": 17550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.783361842454135e-05,
+      "loss": 2.7802,
+      "step": 17555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.782694496433831e-05,
+      "loss": 2.9379,
+      "step": 17560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.782026992805394e-05,
+      "loss": 2.9581,
+      "step": 17565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.781359331596638e-05,
+      "loss": 2.8699,
+      "step": 17570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.78069151283538e-05,
+      "loss": 2.8462,
+      "step": 17575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.780023536549445e-05,
+      "loss": 2.92,
+      "step": 17580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.779355402766664e-05,
+      "loss": 2.9012,
+      "step": 17585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.778687111514878e-05,
+      "loss": 2.6992,
+      "step": 17590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.77801866282193e-05,
+      "loss": 2.8244,
+      "step": 17595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.777350056715668e-05,
+      "loss": 2.8787,
+      "step": 17600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.776681293223956e-05,
+      "loss": 2.9182,
+      "step": 17605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.776012372374654e-05,
+      "loss": 2.9053,
+      "step": 17610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.775343294195634e-05,
+      "loss": 2.8672,
+      "step": 17615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.774674058714776e-05,
+      "loss": 2.9544,
+      "step": 17620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.77400466595996e-05,
+      "loss": 2.8656,
+      "step": 17625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.773335115959081e-05,
+      "loss": 2.872,
+      "step": 17630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.772665408740031e-05,
+      "loss": 2.7518,
+      "step": 17635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.771995544330717e-05,
+      "loss": 2.9085,
+      "step": 17640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.77132552275905e-05,
+      "loss": 2.7133,
+      "step": 17645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.770655344052945e-05,
+      "loss": 2.9595,
+      "step": 17650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.769985008240326e-05,
+      "loss": 2.8754,
+      "step": 17655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.769314515349123e-05,
+      "loss": 2.8457,
+      "step": 17660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.768643865407273e-05,
+      "loss": 2.8195,
+      "step": 17665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.767973058442718e-05,
+      "loss": 2.9622,
+      "step": 17670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.767302094483407e-05,
+      "loss": 2.8263,
+      "step": 17675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.766630973557299e-05,
+      "loss": 2.8894,
+      "step": 17680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.765959695692355e-05,
+      "loss": 2.8949,
+      "step": 17685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.765288260916541e-05,
+      "loss": 2.8959,
+      "step": 17690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.764616669257837e-05,
+      "loss": 2.9476,
+      "step": 17695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.763944920744225e-05,
+      "loss": 2.766,
+      "step": 17700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.763273015403691e-05,
+      "loss": 2.8328,
+      "step": 17705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.762600953264233e-05,
+      "loss": 2.8405,
+      "step": 17710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.761928734353852e-05,
+      "loss": 2.891,
+      "step": 17715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.761256358700556e-05,
+      "loss": 2.768,
+      "step": 17720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.760583826332361e-05,
+      "loss": 2.7274,
+      "step": 17725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.759911137277286e-05,
+      "loss": 2.9139,
+      "step": 17730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.759238291563359e-05,
+      "loss": 2.8013,
+      "step": 17735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.758565289218619e-05,
+      "loss": 2.9251,
+      "step": 17740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.757892130271104e-05,
+      "loss": 2.9435,
+      "step": 17745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.75721881474886e-05,
+      "loss": 2.8887,
+      "step": 17750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.756545342679944e-05,
+      "loss": 2.7889,
+      "step": 17755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.755871714092415e-05,
+      "loss": 2.8617,
+      "step": 17760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.75519792901434e-05,
+      "loss": 2.8361,
+      "step": 17765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.754523987473793e-05,
+      "loss": 2.7966,
+      "step": 17770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.753849889498855e-05,
+      "loss": 2.7923,
+      "step": 17775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.753175635117612e-05,
+      "loss": 2.904,
+      "step": 17780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.752501224358156e-05,
+      "loss": 2.7448,
+      "step": 17785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.75182665724859e-05,
+      "loss": 2.9914,
+      "step": 17790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.751151933817016e-05,
+      "loss": 2.864,
+      "step": 17795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.750477054091548e-05,
+      "loss": 2.9476,
+      "step": 17800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.749802018100308e-05,
+      "loss": 2.9466,
+      "step": 17805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.74912682587142e-05,
+      "loss": 2.833,
+      "step": 17810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.748451477433016e-05,
+      "loss": 2.7821,
+      "step": 17815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.747775972813232e-05,
+      "loss": 2.8287,
+      "step": 17820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.74710031204022e-05,
+      "loss": 2.8858,
+      "step": 17825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.746424495142125e-05,
+      "loss": 2.7722,
+      "step": 17830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.745748522147109e-05,
+      "loss": 3.0612,
+      "step": 17835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.745072393083337e-05,
+      "loss": 2.8861,
+      "step": 17840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.744396107978978e-05,
+      "loss": 2.8426,
+      "step": 17845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.743719666862213e-05,
+      "loss": 3.0318,
+      "step": 17850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.743043069761223e-05,
+      "loss": 2.8448,
+      "step": 17855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.7423663167042e-05,
+      "loss": 2.8429,
+      "step": 17860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.741689407719343e-05,
+      "loss": 2.791,
+      "step": 17865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.741012342834853e-05,
+      "loss": 2.8711,
+      "step": 17870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.740335122078942e-05,
+      "loss": 2.8509,
+      "step": 17875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.739657745479825e-05,
+      "loss": 2.8755,
+      "step": 17880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.738980213065729e-05,
+      "loss": 2.9208,
+      "step": 17885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.738302524864881e-05,
+      "loss": 2.8934,
+      "step": 17890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.737624680905518e-05,
+      "loss": 2.8057,
+      "step": 17895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.73694668121588e-05,
+      "loss": 2.903,
+      "step": 17900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.736268525824222e-05,
+      "loss": 2.8029,
+      "step": 17905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.735590214758795e-05,
+      "loss": 2.8118,
+      "step": 17910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.734911748047864e-05,
+      "loss": 2.6674,
+      "step": 17915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.734233125719696e-05,
+      "loss": 2.9185,
+      "step": 17920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.733554347802563e-05,
+      "loss": 2.8648,
+      "step": 17925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.732875414324755e-05,
+      "loss": 2.9468,
+      "step": 17930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.732196325314554e-05,
+      "loss": 2.7982,
+      "step": 17935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.731517080800256e-05,
+      "loss": 2.8571,
+      "step": 17940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.730837680810162e-05,
+      "loss": 2.7621,
+      "step": 17945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.730158125372578e-05,
+      "loss": 2.8498,
+      "step": 17950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.729478414515824e-05,
+      "loss": 2.736,
+      "step": 17955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.728798548268213e-05,
+      "loss": 2.9376,
+      "step": 17960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.728118526658076e-05,
+      "loss": 2.9184,
+      "step": 17965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.727438349713745e-05,
+      "loss": 3.1001,
+      "step": 17970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.726758017463563e-05,
+      "loss": 2.7958,
+      "step": 17975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.726077529935873e-05,
+      "loss": 2.9208,
+      "step": 17980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.72539688715903e-05,
+      "loss": 2.9094,
+      "step": 17985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.724716089161392e-05,
+      "loss": 2.9469,
+      "step": 17990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.724035135971325e-05,
+      "loss": 2.8309,
+      "step": 17995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.723354027617204e-05,
+      "loss": 2.8828,
+      "step": 18000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.722672764127402e-05,
+      "loss": 2.9736,
+      "step": 18005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.721991345530309e-05,
+      "loss": 2.8772,
+      "step": 18010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.721309771854316e-05,
+      "loss": 2.8611,
+      "step": 18015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.72062804312782e-05,
+      "loss": 2.8988,
+      "step": 18020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.719946159379226e-05,
+      "loss": 2.7195,
+      "step": 18025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.719264120636947e-05,
+      "loss": 2.9263,
+      "step": 18030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.718581926929397e-05,
+      "loss": 2.8861,
+      "step": 18035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.717899578285003e-05,
+      "loss": 2.7388,
+      "step": 18040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.717217074732192e-05,
+      "loss": 2.7736,
+      "step": 18045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.716534416299404e-05,
+      "loss": 3.0213,
+      "step": 18050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.715851603015083e-05,
+      "loss": 2.9058,
+      "step": 18055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.715168634907675e-05,
+      "loss": 2.7826,
+      "step": 18060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.714485512005641e-05,
+      "loss": 2.892,
+      "step": 18065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.71380223433744e-05,
+      "loss": 2.9779,
+      "step": 18070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.713118801931542e-05,
+      "loss": 2.8471,
+      "step": 18075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.712435214816424e-05,
+      "loss": 2.9314,
+      "step": 18080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.711751473020567e-05,
+      "loss": 2.7877,
+      "step": 18085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.711067576572459e-05,
+      "loss": 2.9142,
+      "step": 18090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.710383525500594e-05,
+      "loss": 2.9099,
+      "step": 18095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.709699319833478e-05,
+      "loss": 2.8106,
+      "step": 18100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.709014959599613e-05,
+      "loss": 2.9472,
+      "step": 18105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.708330444827518e-05,
+      "loss": 2.8468,
+      "step": 18110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.707645775545709e-05,
+      "loss": 2.7534,
+      "step": 18115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.706960951782719e-05,
+      "loss": 2.8925,
+      "step": 18120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.706275973567076e-05,
+      "loss": 2.7874,
+      "step": 18125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.705590840927322e-05,
+      "loss": 2.9092,
+      "step": 18130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.704905553892006e-05,
+      "loss": 2.7108,
+      "step": 18135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.704220112489678e-05,
+      "loss": 2.8018,
+      "step": 18140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.703534516748897e-05,
+      "loss": 2.8577,
+      "step": 18145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.702848766698231e-05,
+      "loss": 2.8402,
+      "step": 18150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.70216286236625e-05,
+      "loss": 2.8267,
+      "step": 18155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.701476803781534e-05,
+      "loss": 2.8893,
+      "step": 18160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.700790590972667e-05,
+      "loss": 2.8933,
+      "step": 18165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.700104223968242e-05,
+      "loss": 2.6956,
+      "step": 18170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.699417702796855e-05,
+      "loss": 2.7781,
+      "step": 18175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.69873102748711e-05,
+      "loss": 2.9442,
+      "step": 18180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.69804419806762e-05,
+      "loss": 3.0235,
+      "step": 18185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.697357214567002e-05,
+      "loss": 2.8641,
+      "step": 18190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.696670077013877e-05,
+      "loss": 2.9167,
+      "step": 18195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.695982785436878e-05,
+      "loss": 2.9562,
+      "step": 18200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.695295339864638e-05,
+      "loss": 2.8636,
+      "step": 18205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.694607740325803e-05,
+      "loss": 2.9383,
+      "step": 18210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.693919986849021e-05,
+      "loss": 2.9283,
+      "step": 18215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.693232079462947e-05,
+      "loss": 2.8509,
+      "step": 18220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.692544018196245e-05,
+      "loss": 2.7805,
+      "step": 18225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.69185580307758e-05,
+      "loss": 2.8088,
+      "step": 18230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.691167434135632e-05,
+      "loss": 2.9662,
+      "step": 18235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.690478911399077e-05,
+      "loss": 2.8573,
+      "step": 18240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.689790234896606e-05,
+      "loss": 2.872,
+      "step": 18245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.689101404656912e-05,
+      "loss": 2.9229,
+      "step": 18250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.688412420708693e-05,
+      "loss": 2.7615,
+      "step": 18255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.687723283080662e-05,
+      "loss": 2.7674,
+      "step": 18260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.687033991801527e-05,
+      "loss": 2.9183,
+      "step": 18265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.686344546900012e-05,
+      "loss": 2.9113,
+      "step": 18270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.685654948404837e-05,
+      "loss": 2.8846,
+      "step": 18275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.68496519634474e-05,
+      "loss": 2.8419,
+      "step": 18280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.684275290748459e-05,
+      "loss": 2.9633,
+      "step": 18285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.683585231644738e-05,
+      "loss": 2.9457,
+      "step": 18290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.682895019062329e-05,
+      "loss": 2.938,
+      "step": 18295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.682204653029991e-05,
+      "loss": 2.8565,
+      "step": 18300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.681514133576484e-05,
+      "loss": 2.8675,
+      "step": 18305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.680823460730586e-05,
+      "loss": 2.935,
+      "step": 18310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.680132634521071e-05,
+      "loss": 2.8833,
+      "step": 18315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.679441654976721e-05,
+      "loss": 2.865,
+      "step": 18320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.678750522126328e-05,
+      "loss": 2.7942,
+      "step": 18325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.678059235998689e-05,
+      "loss": 2.7926,
+      "step": 18330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.677367796622605e-05,
+      "loss": 2.9002,
+      "step": 18335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.676676204026886e-05,
+      "loss": 2.8907,
+      "step": 18340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.67598445824035e-05,
+      "loss": 2.9296,
+      "step": 18345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.675292559291813e-05,
+      "loss": 2.95,
+      "step": 18350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.674600507210109e-05,
+      "loss": 2.7848,
+      "step": 18355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.67390830202407e-05,
+      "loss": 2.9131,
+      "step": 18360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.67321594376254e-05,
+      "loss": 2.8488,
+      "step": 18365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.672523432454362e-05,
+      "loss": 2.8858,
+      "step": 18370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.671830768128393e-05,
+      "loss": 2.9365,
+      "step": 18375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.671137950813492e-05,
+      "loss": 2.9213,
+      "step": 18380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.670444980538528e-05,
+      "loss": 2.8898,
+      "step": 18385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.669751857332371e-05,
+      "loss": 2.852,
+      "step": 18390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.669058581223902e-05,
+      "loss": 2.8886,
+      "step": 18395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.668365152242006e-05,
+      "loss": 2.859,
+      "step": 18400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.667671570415577e-05,
+      "loss": 2.7223,
+      "step": 18405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.66697783577351e-05,
+      "loss": 2.9159,
+      "step": 18410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.666283948344712e-05,
+      "loss": 2.9515,
+      "step": 18415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.665589908158094e-05,
+      "loss": 3.0843,
+      "step": 18420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.664895715242573e-05,
+      "loss": 2.796,
+      "step": 18425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.664201369627076e-05,
+      "loss": 2.8904,
+      "step": 18430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.663506871340529e-05,
+      "loss": 2.9482,
+      "step": 18435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.662812220411872e-05,
+      "loss": 2.882,
+      "step": 18440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.662117416870044e-05,
+      "loss": 2.9235,
+      "step": 18445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.661422460743998e-05,
+      "loss": 2.8706,
+      "step": 18450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.66072735206269e-05,
+      "loss": 2.8789,
+      "step": 18455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.660032090855078e-05,
+      "loss": 2.8496,
+      "step": 18460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.659336677150136e-05,
+      "loss": 2.7293,
+      "step": 18465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.658641110976834e-05,
+      "loss": 2.8107,
+      "step": 18470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.657945392364155e-05,
+      "loss": 2.975,
+      "step": 18475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.657249521341088e-05,
+      "loss": 2.9983,
+      "step": 18480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.656553497936626e-05,
+      "loss": 2.9698,
+      "step": 18485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.655857322179767e-05,
+      "loss": 2.9138,
+      "step": 18490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.65516099409952e-05,
+      "loss": 2.945,
+      "step": 18495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.654464513724899e-05,
+      "loss": 2.8063,
+      "step": 18500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.65376788108492e-05,
+      "loss": 2.7419,
+      "step": 18505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.653071096208609e-05,
+      "loss": 2.9927,
+      "step": 18510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.652374159125002e-05,
+      "loss": 2.7935,
+      "step": 18515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.651677069863132e-05,
+      "loss": 2.8172,
+      "step": 18520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.650979828452046e-05,
+      "loss": 2.8193,
+      "step": 18525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.650282434920796e-05,
+      "loss": 2.9025,
+      "step": 18530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.649584889298438e-05,
+      "loss": 2.804,
+      "step": 18535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.648887191614035e-05,
+      "loss": 2.8142,
+      "step": 18540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.648189341896659e-05,
+      "loss": 2.8761,
+      "step": 18545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.647491340175385e-05,
+      "loss": 2.8404,
+      "step": 18550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.646793186479294e-05,
+      "loss": 2.7878,
+      "step": 18555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.646094880837479e-05,
+      "loss": 2.9352,
+      "step": 18560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.645396423279031e-05,
+      "loss": 2.9624,
+      "step": 18565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.644697813833056e-05,
+      "loss": 2.8022,
+      "step": 18570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.643999052528658e-05,
+      "loss": 2.7738,
+      "step": 18575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.643300139394952e-05,
+      "loss": 2.8454,
+      "step": 18580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.642601074461062e-05,
+      "loss": 2.8477,
+      "step": 18585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.641901857756112e-05,
+      "loss": 2.8418,
+      "step": 18590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.641202489309235e-05,
+      "loss": 2.8401,
+      "step": 18595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.64050296914957e-05,
+      "loss": 2.7967,
+      "step": 18600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.639803297306266e-05,
+      "loss": 2.9659,
+      "step": 18605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.639103473808473e-05,
+      "loss": 2.7912,
+      "step": 18610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.638403498685351e-05,
+      "loss": 2.8383,
+      "step": 18615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.637703371966063e-05,
+      "loss": 2.8408,
+      "step": 18620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.63700309367978e-05,
+      "loss": 2.8776,
+      "step": 18625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.636302663855683e-05,
+      "loss": 2.9213,
+      "step": 18630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.635602082522952e-05,
+      "loss": 2.8225,
+      "step": 18635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.634901349710778e-05,
+      "loss": 2.7134,
+      "step": 18640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.63420046544836e-05,
+      "loss": 2.8485,
+      "step": 18645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.633499429764897e-05,
+      "loss": 2.7888,
+      "step": 18650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.6327982426896e-05,
+      "loss": 2.874,
+      "step": 18655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.632096904251683e-05,
+      "loss": 2.8323,
+      "step": 18660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.63139541448037e-05,
+      "loss": 2.8621,
+      "step": 18665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.630693773404888e-05,
+      "loss": 2.874,
+      "step": 18670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.62999198105447e-05,
+      "loss": 2.8403,
+      "step": 18675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.629290037458359e-05,
+      "loss": 2.8503,
+      "step": 18680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.628587942645798e-05,
+      "loss": 2.843,
+      "step": 18685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.627885696646046e-05,
+      "loss": 2.9781,
+      "step": 18690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.627183299488358e-05,
+      "loss": 2.814,
+      "step": 18695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.626480751201999e-05,
+      "loss": 2.8935,
+      "step": 18700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.625778051816244e-05,
+      "loss": 2.9003,
+      "step": 18705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.625075201360371e-05,
+      "loss": 2.7475,
+      "step": 18710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.624372199863664e-05,
+      "loss": 2.8108,
+      "step": 18715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.623669047355414e-05,
+      "loss": 2.7636,
+      "step": 18720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.622965743864917e-05,
+      "loss": 2.8738,
+      "step": 18725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.62226228942148e-05,
+      "loss": 2.8207,
+      "step": 18730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.621558684054408e-05,
+      "loss": 2.8541,
+      "step": 18735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.620854927793021e-05,
+      "loss": 2.9069,
+      "step": 18740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.620151020666641e-05,
+      "loss": 2.7861,
+      "step": 18745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.619446962704595e-05,
+      "loss": 2.9401,
+      "step": 18750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.618742753936218e-05,
+      "loss": 2.6734,
+      "step": 18755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.618038394390855e-05,
+      "loss": 2.8508,
+      "step": 18760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.617333884097847e-05,
+      "loss": 2.7391,
+      "step": 18765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.616629223086554e-05,
+      "loss": 2.7186,
+      "step": 18770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.615924411386332e-05,
+      "loss": 2.9019,
+      "step": 18775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.61521944902655e-05,
+      "loss": 2.9219,
+      "step": 18780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.614514336036578e-05,
+      "loss": 2.7895,
+      "step": 18785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.613809072445798e-05,
+      "loss": 2.791,
+      "step": 18790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.613103658283594e-05,
+      "loss": 2.9076,
+      "step": 18795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.612398093579357e-05,
+      "loss": 2.9213,
+      "step": 18800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.611692378362483e-05,
+      "loss": 2.9449,
+      "step": 18805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.610986512662381e-05,
+      "loss": 2.8225,
+      "step": 18810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.610280496508456e-05,
+      "loss": 2.7793,
+      "step": 18815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.609574329930128e-05,
+      "loss": 2.817,
+      "step": 18820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.608868012956819e-05,
+      "loss": 2.8681,
+      "step": 18825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.608161545617957e-05,
+      "loss": 2.8543,
+      "step": 18830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.607454927942979e-05,
+      "loss": 2.806,
+      "step": 18835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.606748159961326e-05,
+      "loss": 2.7943,
+      "step": 18840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.606041241702445e-05,
+      "loss": 2.7071,
+      "step": 18845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.60533417319579e-05,
+      "loss": 2.8859,
+      "step": 18850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.604626954470823e-05,
+      "loss": 2.8721,
+      "step": 18855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.60391958555701e-05,
+      "loss": 2.8753,
+      "step": 18860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.603212066483825e-05,
+      "loss": 2.8492,
+      "step": 18865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.602504397280745e-05,
+      "loss": 2.8303,
+      "step": 18870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.601796577977257e-05,
+      "loss": 2.8135,
+      "step": 18875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.601088608602852e-05,
+      "loss": 2.8347,
+      "step": 18880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.600380489187027e-05,
+      "loss": 2.8233,
+      "step": 18885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.599672219759289e-05,
+      "loss": 2.731,
+      "step": 18890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.598963800349146e-05,
+      "loss": 2.774,
+      "step": 18895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.598255230986115e-05,
+      "loss": 2.9218,
+      "step": 18900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.59754651169972e-05,
+      "loss": 2.8293,
+      "step": 18905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.59683764251949e-05,
+      "loss": 2.7581,
+      "step": 18910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.596128623474959e-05,
+      "loss": 2.8178,
+      "step": 18915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.595419454595671e-05,
+      "loss": 2.8358,
+      "step": 18920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.594710135911173e-05,
+      "loss": 2.9374,
+      "step": 18925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.594000667451018e-05,
+      "loss": 2.7811,
+      "step": 18930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.593291049244768e-05,
+      "loss": 2.755,
+      "step": 18935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.592581281321988e-05,
+      "loss": 2.9588,
+      "step": 18940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.591871363712251e-05,
+      "loss": 2.8899,
+      "step": 18945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.591161296445138e-05,
+      "loss": 2.8121,
+      "step": 18950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.590451079550234e-05,
+      "loss": 2.9254,
+      "step": 18955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.589740713057129e-05,
+      "loss": 2.8473,
+      "step": 18960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.589030196995423e-05,
+      "loss": 2.8821,
+      "step": 18965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.588319531394718e-05,
+      "loss": 2.8777,
+      "step": 18970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.587608716284624e-05,
+      "loss": 2.9024,
+      "step": 18975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.586897751694761e-05,
+      "loss": 2.6504,
+      "step": 18980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.586186637654747e-05,
+      "loss": 2.9306,
+      "step": 18985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.585475374194213e-05,
+      "loss": 2.8189,
+      "step": 18990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.584763961342797e-05,
+      "loss": 2.8515,
+      "step": 18995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.584052399130136e-05,
+      "loss": 3.0231,
+      "step": 19000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.583340687585879e-05,
+      "loss": 2.8529,
+      "step": 19005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.582628826739682e-05,
+      "loss": 2.8419,
+      "step": 19010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.581916816621202e-05,
+      "loss": 2.8413,
+      "step": 19015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.581204657260109e-05,
+      "loss": 2.8844,
+      "step": 19020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.580492348686072e-05,
+      "loss": 2.8395,
+      "step": 19025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.579779890928772e-05,
+      "loss": 2.8629,
+      "step": 19030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.579067284017892e-05,
+      "loss": 2.8853,
+      "step": 19035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.578354527983126e-05,
+      "loss": 2.7729,
+      "step": 19040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.577641622854168e-05,
+      "loss": 2.8213,
+      "step": 19045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.576928568660722e-05,
+      "loss": 2.9456,
+      "step": 19050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.576215365432502e-05,
+      "loss": 2.9241,
+      "step": 19055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.575502013199219e-05,
+      "loss": 2.8839,
+      "step": 19060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.574788511990599e-05,
+      "loss": 2.7873,
+      "step": 19065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.574074861836367e-05,
+      "loss": 2.9026,
+      "step": 19070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.573361062766261e-05,
+      "loss": 2.7978,
+      "step": 19075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.57264711481002e-05,
+      "loss": 2.8947,
+      "step": 19080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.57193301799739e-05,
+      "loss": 2.8082,
+      "step": 19085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.571218772358126e-05,
+      "loss": 2.8175,
+      "step": 19090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.570504377921987e-05,
+      "loss": 2.7482,
+      "step": 19095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.569789834718739e-05,
+      "loss": 2.9652,
+      "step": 19100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.569075142778153e-05,
+      "loss": 2.8325,
+      "step": 19105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.568360302130008e-05,
+      "loss": 2.8001,
+      "step": 19110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.567645312804088e-05,
+      "loss": 2.9401,
+      "step": 19115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.566930174830182e-05,
+      "loss": 2.7833,
+      "step": 19120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.566214888238089e-05,
+      "loss": 2.8706,
+      "step": 19125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.56549945305761e-05,
+      "loss": 2.9645,
+      "step": 19130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.564783869318555e-05,
+      "loss": 2.889,
+      "step": 19135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.56406813705074e-05,
+      "loss": 2.8149,
+      "step": 19140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.563352256283984e-05,
+      "loss": 2.7542,
+      "step": 19145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.562636227048115e-05,
+      "loss": 2.9655,
+      "step": 19150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.56192004937297e-05,
+      "loss": 2.8608,
+      "step": 19155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.561203723288384e-05,
+      "loss": 2.7727,
+      "step": 19160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.560487248824208e-05,
+      "loss": 2.7109,
+      "step": 19165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.559770626010292e-05,
+      "loss": 2.8621,
+      "step": 19170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.559053854876494e-05,
+      "loss": 2.7917,
+      "step": 19175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.558336935452678e-05,
+      "loss": 2.9016,
+      "step": 19180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.557619867768719e-05,
+      "loss": 2.7516,
+      "step": 19185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.556902651854488e-05,
+      "loss": 2.9169,
+      "step": 19190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.556185287739871e-05,
+      "loss": 2.9911,
+      "step": 19195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.555467775454761e-05,
+      "loss": 2.8956,
+      "step": 19200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.554750115029048e-05,
+      "loss": 2.9037,
+      "step": 19205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.554032306492637e-05,
+      "loss": 2.8272,
+      "step": 19210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.553314349875434e-05,
+      "loss": 2.9125,
+      "step": 19215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.552596245207354e-05,
+      "loss": 2.7439,
+      "step": 19220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.551877992518318e-05,
+      "loss": 2.8069,
+      "step": 19225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.551159591838251e-05,
+      "loss": 2.946,
+      "step": 19230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.550441043197088e-05,
+      "loss": 2.802,
+      "step": 19235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.549722346624765e-05,
+      "loss": 2.8572,
+      "step": 19240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.549003502151227e-05,
+      "loss": 2.7686,
+      "step": 19245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.548284509806426e-05,
+      "loss": 2.8368,
+      "step": 19250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.54756536962032e-05,
+      "loss": 2.8511,
+      "step": 19255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.54684608162287e-05,
+      "loss": 2.8298,
+      "step": 19260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.546126645844049e-05,
+      "loss": 2.9342,
+      "step": 19265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.54540706231383e-05,
+      "loss": 2.7552,
+      "step": 19270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.544687331062196e-05,
+      "loss": 2.8036,
+      "step": 19275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.543967452119135e-05,
+      "loss": 2.8513,
+      "step": 19280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.54324742551464e-05,
+      "loss": 2.8121,
+      "step": 19285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.542527251278712e-05,
+      "loss": 2.7916,
+      "step": 19290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.541806929441358e-05,
+      "loss": 2.7968,
+      "step": 19295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.54108646003259e-05,
+      "loss": 2.8723,
+      "step": 19300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.540365843082428e-05,
+      "loss": 2.857,
+      "step": 19305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.539645078620895e-05,
+      "loss": 2.8262,
+      "step": 19310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.538924166678023e-05,
+      "loss": 2.8006,
+      "step": 19315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.538203107283849e-05,
+      "loss": 2.7803,
+      "step": 19320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.537481900468418e-05,
+      "loss": 2.7691,
+      "step": 19325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.536760546261777e-05,
+      "loss": 2.8266,
+      "step": 19330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.536039044693982e-05,
+      "loss": 2.8939,
+      "step": 19335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.535317395795097e-05,
+      "loss": 2.8126,
+      "step": 19340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.534595599595188e-05,
+      "loss": 2.7797,
+      "step": 19345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.533873656124329e-05,
+      "loss": 2.9304,
+      "step": 19350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.533151565412601e-05,
+      "loss": 2.9682,
+      "step": 19355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.53242932749009e-05,
+      "loss": 2.7249,
+      "step": 19360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.531706942386887e-05,
+      "loss": 2.9039,
+      "step": 19365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.530984410133094e-05,
+      "loss": 2.7904,
+      "step": 19370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.530261730758813e-05,
+      "loss": 2.7664,
+      "step": 19375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.529538904294154e-05,
+      "loss": 2.8181,
+      "step": 19380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.528815930769238e-05,
+      "loss": 2.7747,
+      "step": 19385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.528092810214185e-05,
+      "loss": 2.8756,
+      "step": 19390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.527369542659125e-05,
+      "loss": 2.9055,
+      "step": 19395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.526646128134192e-05,
+      "loss": 2.8354,
+      "step": 19400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.525922566669528e-05,
+      "loss": 2.9435,
+      "step": 19405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.525198858295285e-05,
+      "loss": 2.8707,
+      "step": 19410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.52447500304161e-05,
+      "loss": 2.7989,
+      "step": 19415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.523751000938666e-05,
+      "loss": 2.8743,
+      "step": 19420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.523026852016621e-05,
+      "loss": 2.7509,
+      "step": 19425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.522302556305644e-05,
+      "loss": 2.9532,
+      "step": 19430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.521578113835913e-05,
+      "loss": 2.8704,
+      "step": 19435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.520853524637614e-05,
+      "loss": 2.8874,
+      "step": 19440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.520128788740937e-05,
+      "loss": 2.7929,
+      "step": 19445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.51940390617608e-05,
+      "loss": 2.7844,
+      "step": 19450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.518678876973241e-05,
+      "loss": 2.7533,
+      "step": 19455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.517953701162635e-05,
+      "loss": 2.748,
+      "step": 19460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.51722837877447e-05,
+      "loss": 2.9059,
+      "step": 19465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.516502909838974e-05,
+      "loss": 2.8125,
+      "step": 19470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.515777294386368e-05,
+      "loss": 3.0028,
+      "step": 19475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.51505153244689e-05,
+      "loss": 2.8751,
+      "step": 19480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.514325624050776e-05,
+      "loss": 2.8086,
+      "step": 19485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.513599569228274e-05,
+      "loss": 2.7569,
+      "step": 19490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.512873368009633e-05,
+      "loss": 2.7301,
+      "step": 19495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.512147020425112e-05,
+      "loss": 2.8867,
+      "step": 19500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.511420526504973e-05,
+      "loss": 2.8611,
+      "step": 19505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.510693886279489e-05,
+      "loss": 2.7505,
+      "step": 19510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.509967099778933e-05,
+      "loss": 2.9556,
+      "step": 19515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.50924016703359e-05,
+      "loss": 2.9489,
+      "step": 19520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.508513088073744e-05,
+      "loss": 2.7996,
+      "step": 19525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.507785862929692e-05,
+      "loss": 2.7675,
+      "step": 19530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.507058491631732e-05,
+      "loss": 2.7822,
+      "step": 19535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.506330974210173e-05,
+      "loss": 2.9596,
+      "step": 19540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.505603310695328e-05,
+      "loss": 2.7663,
+      "step": 19545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.504875501117511e-05,
+      "loss": 2.8038,
+      "step": 19550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.504147545507048e-05,
+      "loss": 2.8452,
+      "step": 19555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.503419443894274e-05,
+      "loss": 2.7275,
+      "step": 19560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.502691196309522e-05,
+      "loss": 2.8704,
+      "step": 19565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.501962802783137e-05,
+      "loss": 2.9096,
+      "step": 19570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.501234263345465e-05,
+      "loss": 2.7726,
+      "step": 19575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.500505578026863e-05,
+      "loss": 2.8224,
+      "step": 19580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.499776746857692e-05,
+      "loss": 2.8599,
+      "step": 19585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.499047769868318e-05,
+      "loss": 2.8255,
+      "step": 19590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.498318647089116e-05,
+      "loss": 2.8383,
+      "step": 19595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.497589378550465e-05,
+      "loss": 2.7971,
+      "step": 19600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.496859964282749e-05,
+      "loss": 2.8848,
+      "step": 19605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.49613040431636e-05,
+      "loss": 2.7548,
+      "step": 19610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.495400698681697e-05,
+      "loss": 2.8601,
+      "step": 19615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.49467084740916e-05,
+      "loss": 2.8535,
+      "step": 19620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.493940850529163e-05,
+      "loss": 3.0614,
+      "step": 19625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.493210708072121e-05,
+      "loss": 2.8543,
+      "step": 19630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.492480420068452e-05,
+      "loss": 2.89,
+      "step": 19635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.491749986548589e-05,
+      "loss": 2.9022,
+      "step": 19640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.491019407542962e-05,
+      "loss": 2.8318,
+      "step": 19645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.490288683082013e-05,
+      "loss": 2.8161,
+      "step": 19650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.489557813196187e-05,
+      "loss": 2.8039,
+      "step": 19655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.488826797915936e-05,
+      "loss": 2.8246,
+      "step": 19660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.48809563727172e-05,
+      "loss": 2.8961,
+      "step": 19665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.487364331294001e-05,
+      "loss": 2.9096,
+      "step": 19670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.486632880013252e-05,
+      "loss": 2.7734,
+      "step": 19675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.485901283459945e-05,
+      "loss": 2.8698,
+      "step": 19680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.485169541664566e-05,
+      "loss": 2.8548,
+      "step": 19685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.484437654657604e-05,
+      "loss": 2.684,
+      "step": 19690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.483705622469551e-05,
+      "loss": 2.8416,
+      "step": 19695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.482973445130908e-05,
+      "loss": 2.8143,
+      "step": 19700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.482241122672183e-05,
+      "loss": 2.8121,
+      "step": 19705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.481508655123887e-05,
+      "loss": 2.948,
+      "step": 19710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.48077604251654e-05,
+      "loss": 2.8541,
+      "step": 19715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.480043284880666e-05,
+      "loss": 2.8722,
+      "step": 19720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.479310382246794e-05,
+      "loss": 2.9444,
+      "step": 19725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.478577334645465e-05,
+      "loss": 2.8622,
+      "step": 19730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.477844142107219e-05,
+      "loss": 2.7768,
+      "step": 19735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.477110804662604e-05,
+      "loss": 2.693,
+      "step": 19740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.476377322342177e-05,
+      "loss": 2.8683,
+      "step": 19745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.4756436951765e-05,
+      "loss": 2.8073,
+      "step": 19750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.474909923196138e-05,
+      "loss": 2.7662,
+      "step": 19755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.474176006431663e-05,
+      "loss": 2.7306,
+      "step": 19760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.473441944913655e-05,
+      "loss": 2.9108,
+      "step": 19765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.472707738672701e-05,
+      "loss": 2.8736,
+      "step": 19770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.471973387739389e-05,
+      "loss": 2.7241,
+      "step": 19775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.471238892144319e-05,
+      "loss": 2.8221,
+      "step": 19780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.470504251918092e-05,
+      "loss": 2.8808,
+      "step": 19785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.469769467091319e-05,
+      "loss": 2.874,
+      "step": 19790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.469034537694615e-05,
+      "loss": 2.8774,
+      "step": 19795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.468299463758597e-05,
+      "loss": 2.8851,
+      "step": 19800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.4675642453139e-05,
+      "loss": 2.6736,
+      "step": 19805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.466828882391152e-05,
+      "loss": 2.7563,
+      "step": 19810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.466093375020993e-05,
+      "loss": 2.9171,
+      "step": 19815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.46535772323407e-05,
+      "loss": 2.9152,
+      "step": 19820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.464621927061032e-05,
+      "loss": 2.76,
+      "step": 19825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.463885986532538e-05,
+      "loss": 2.9569,
+      "step": 19830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.463149901679251e-05,
+      "loss": 2.929,
+      "step": 19835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.462413672531841e-05,
+      "loss": 2.6674,
+      "step": 19840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.461677299120982e-05,
+      "loss": 2.837,
+      "step": 19845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.460940781477357e-05,
+      "loss": 2.9678,
+      "step": 19850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.460204119631651e-05,
+      "loss": 2.7114,
+      "step": 19855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.45946731361456e-05,
+      "loss": 2.8295,
+      "step": 19860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.458730363456783e-05,
+      "loss": 2.9392,
+      "step": 19865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.457993269189024e-05,
+      "loss": 2.9719,
+      "step": 19870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.457256030841996e-05,
+      "loss": 2.8661,
+      "step": 19875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.456518648446417e-05,
+      "loss": 2.8328,
+      "step": 19880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.455781122033007e-05,
+      "loss": 2.845,
+      "step": 19885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.455043451632499e-05,
+      "loss": 2.9318,
+      "step": 19890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.454305637275626e-05,
+      "loss": 2.9513,
+      "step": 19895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.453567678993132e-05,
+      "loss": 2.7697,
+      "step": 19900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.452829576815763e-05,
+      "loss": 2.8385,
+      "step": 19905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.452091330774272e-05,
+      "loss": 2.7998,
+      "step": 19910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.45135294089942e-05,
+      "loss": 2.7164,
+      "step": 19915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.450614407221972e-05,
+      "loss": 2.8629,
+      "step": 19920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.449875729772697e-05,
+      "loss": 2.8403,
+      "step": 19925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.449136908582375e-05,
+      "loss": 2.7757,
+      "step": 19930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.44839794368179e-05,
+      "loss": 2.8392,
+      "step": 19935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.447658835101729e-05,
+      "loss": 2.8059,
+      "step": 19940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.44691958287299e-05,
+      "loss": 2.8333,
+      "step": 19945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.44618018702637e-05,
+      "loss": 2.835,
+      "step": 19950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.445440647592681e-05,
+      "loss": 2.7274,
+      "step": 19955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.444700964602734e-05,
+      "loss": 2.8354,
+      "step": 19960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.443961138087351e-05,
+      "loss": 2.8487,
+      "step": 19965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.443221168077352e-05,
+      "loss": 2.7877,
+      "step": 19970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.442481054603573e-05,
+      "loss": 2.7952,
+      "step": 19975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.441740797696851e-05,
+      "loss": 2.9898,
+      "step": 19980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.441000397388025e-05,
+      "loss": 2.8594,
+      "step": 19985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.440259853707949e-05,
+      "loss": 2.887,
+      "step": 19990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.439519166687476e-05,
+      "loss": 2.8027,
+      "step": 19995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.438778336357467e-05,
+      "loss": 2.7233,
+      "step": 20000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.43803736274879e-05,
+      "loss": 2.8225,
+      "step": 20005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.437296245892319e-05,
+      "loss": 2.8856,
+      "step": 20010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.436554985818929e-05,
+      "loss": 2.8254,
+      "step": 20015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.43581358255951e-05,
+      "loss": 2.8578,
+      "step": 20020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.43507203614495e-05,
+      "loss": 2.8947,
+      "step": 20025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.434330346606147e-05,
+      "loss": 2.8401,
+      "step": 20030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.433588513974004e-05,
+      "loss": 2.944,
+      "step": 20035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.432846538279428e-05,
+      "loss": 2.7834,
+      "step": 20040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.432104419553336e-05,
+      "loss": 2.7319,
+      "step": 20045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.431362157826648e-05,
+      "loss": 2.7808,
+      "step": 20050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.430619753130291e-05,
+      "loss": 2.8944,
+      "step": 20055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.429877205495198e-05,
+      "loss": 2.8284,
+      "step": 20060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.429134514952308e-05,
+      "loss": 2.7867,
+      "step": 20065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.428391681532564e-05,
+      "loss": 2.814,
+      "step": 20070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.427648705266917e-05,
+      "loss": 2.9078,
+      "step": 20075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.426905586186325e-05,
+      "loss": 2.8382,
+      "step": 20080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.426162324321749e-05,
+      "loss": 2.8336,
+      "step": 20085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.425418919704158e-05,
+      "loss": 2.8863,
+      "step": 20090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.424675372364527e-05,
+      "loss": 2.8295,
+      "step": 20095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.423931682333835e-05,
+      "loss": 2.6698,
+      "step": 20100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.42318784964307e-05,
+      "loss": 2.9428,
+      "step": 20105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.42244387432322e-05,
+      "loss": 2.8183,
+      "step": 20110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.42169975640529e-05,
+      "loss": 2.8815,
+      "step": 20115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.420955495920277e-05,
+      "loss": 2.9489,
+      "step": 20120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.420211092899198e-05,
+      "loss": 2.8695,
+      "step": 20125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.419466547373063e-05,
+      "loss": 2.7612,
+      "step": 20130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.418721859372897e-05,
+      "loss": 2.8362,
+      "step": 20135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.417977028929727e-05,
+      "loss": 2.89,
+      "step": 20140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.417232056074587e-05,
+      "loss": 2.8991,
+      "step": 20145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.416486940838517e-05,
+      "loss": 2.8725,
+      "step": 20150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.415741683252562e-05,
+      "loss": 2.8497,
+      "step": 20155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.414996283347775e-05,
+      "loss": 2.7205,
+      "step": 20160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.414250741155211e-05,
+      "loss": 2.9345,
+      "step": 20165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.413505056705935e-05,
+      "loss": 2.8616,
+      "step": 20170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.412759230031017e-05,
+      "loss": 2.8277,
+      "step": 20175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.412013261161531e-05,
+      "loss": 2.8617,
+      "step": 20180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.41126715012856e-05,
+      "loss": 2.7756,
+      "step": 20185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.410520896963189e-05,
+      "loss": 2.9045,
+      "step": 20190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.409774501696512e-05,
+      "loss": 2.8965,
+      "step": 20195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.409027964359627e-05,
+      "loss": 2.8003,
+      "step": 20200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.408281284983642e-05,
+      "loss": 2.9432,
+      "step": 20205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.407534463599663e-05,
+      "loss": 2.7341,
+      "step": 20210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.406787500238812e-05,
+      "loss": 2.9937,
+      "step": 20215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.406040394932208e-05,
+      "loss": 2.9055,
+      "step": 20220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.40529314771098e-05,
+      "loss": 2.842,
+      "step": 20225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.404545758606264e-05,
+      "loss": 2.8992,
+      "step": 20230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.403798227649199e-05,
+      "loss": 2.6739,
+      "step": 20235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.403050554870932e-05,
+      "loss": 2.8158,
+      "step": 20240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.402302740302616e-05,
+      "loss": 2.865,
+      "step": 20245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.401554783975407e-05,
+      "loss": 2.6896,
+      "step": 20250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.400806685920471e-05,
+      "loss": 2.8107,
+      "step": 20255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.400058446168976e-05,
+      "loss": 2.8634,
+      "step": 20260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.399310064752099e-05,
+      "loss": 2.6625,
+      "step": 20265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.398561541701023e-05,
+      "loss": 2.9488,
+      "step": 20270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.397812877046933e-05,
+      "loss": 2.7773,
+      "step": 20275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.397064070821025e-05,
+      "loss": 2.7508,
+      "step": 20280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.396315123054498e-05,
+      "loss": 2.8387,
+      "step": 20285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.395566033778556e-05,
+      "loss": 2.7477,
+      "step": 20290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.39481680302441e-05,
+      "loss": 2.6794,
+      "step": 20295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.394067430823279e-05,
+      "loss": 2.6315,
+      "step": 20300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.393317917206385e-05,
+      "loss": 2.8323,
+      "step": 20305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.392568262204956e-05,
+      "loss": 2.8156,
+      "step": 20310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.39181846585023e-05,
+      "loss": 2.8022,
+      "step": 20315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.391068528173442e-05,
+      "loss": 2.8929,
+      "step": 20320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.390318449205846e-05,
+      "loss": 2.8446,
+      "step": 20325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.389568228978688e-05,
+      "loss": 2.833,
+      "step": 20330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.388817867523229e-05,
+      "loss": 2.8579,
+      "step": 20335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.388067364870734e-05,
+      "loss": 2.8883,
+      "step": 20340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.387316721052471e-05,
+      "loss": 2.9152,
+      "step": 20345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.386565936099718e-05,
+      "loss": 2.7823,
+      "step": 20350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.385815010043755e-05,
+      "loss": 2.8392,
+      "step": 20355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.385063942915874e-05,
+      "loss": 2.9732,
+      "step": 20360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.384312734747363e-05,
+      "loss": 2.7732,
+      "step": 20365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.383561385569525e-05,
+      "loss": 2.7355,
+      "step": 20370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.382809895413664e-05,
+      "loss": 2.8445,
+      "step": 20375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.382058264311093e-05,
+      "loss": 2.7947,
+      "step": 20380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.381306492293127e-05,
+      "loss": 2.843,
+      "step": 20385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.38055457939109e-05,
+      "loss": 2.8457,
+      "step": 20390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.37980252563631e-05,
+      "loss": 2.8786,
+      "step": 20395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.379050331060123e-05,
+      "loss": 2.8437,
+      "step": 20400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.37829799569387e-05,
+      "loss": 2.8227,
+      "step": 20405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.377545519568894e-05,
+      "loss": 2.7379,
+      "step": 20410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.37679290271655e-05,
+      "loss": 2.8368,
+      "step": 20415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.3760401451682e-05,
+      "loss": 2.9334,
+      "step": 20420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.375287246955202e-05,
+      "loss": 2.7619,
+      "step": 20425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.374534208108928e-05,
+      "loss": 2.877,
+      "step": 20430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.373781028660755e-05,
+      "loss": 2.778,
+      "step": 20435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.373027708642062e-05,
+      "loss": 2.7785,
+      "step": 20440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.372274248084237e-05,
+      "loss": 2.8401,
+      "step": 20445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.371520647018678e-05,
+      "loss": 2.851,
+      "step": 20450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.37076690547678e-05,
+      "loss": 2.8422,
+      "step": 20455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.370013023489946e-05,
+      "loss": 2.6805,
+      "step": 20460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.369259001089593e-05,
+      "loss": 2.8858,
+      "step": 20465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.368504838307134e-05,
+      "loss": 2.8582,
+      "step": 20470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.36775053517399e-05,
+      "loss": 2.7704,
+      "step": 20475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.366996091721593e-05,
+      "loss": 2.7856,
+      "step": 20480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.366241507981375e-05,
+      "loss": 2.7211,
+      "step": 20485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.365486783984779e-05,
+      "loss": 2.8252,
+      "step": 20490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.364731919763249e-05,
+      "loss": 2.9665,
+      "step": 20495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.363976915348235e-05,
+      "loss": 2.8827,
+      "step": 20500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.363221770771197e-05,
+      "loss": 2.9032,
+      "step": 20505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.362466486063599e-05,
+      "loss": 2.8867,
+      "step": 20510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.36171106125691e-05,
+      "loss": 2.8517,
+      "step": 20515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.360955496382602e-05,
+      "loss": 2.8712,
+      "step": 20520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.36019979147216e-05,
+      "loss": 2.8043,
+      "step": 20525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.359443946557072e-05,
+      "loss": 2.8581,
+      "step": 20530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.358687961668828e-05,
+      "loss": 2.7953,
+      "step": 20535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.357931836838924e-05,
+      "loss": 2.7394,
+      "step": 20540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.35717557209887e-05,
+      "loss": 2.7699,
+      "step": 20545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.356419167480172e-05,
+      "loss": 2.8497,
+      "step": 20550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.355662623014346e-05,
+      "loss": 2.8918,
+      "step": 20555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.354905938732918e-05,
+      "loss": 2.7631,
+      "step": 20560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.354149114667413e-05,
+      "loss": 2.7701,
+      "step": 20565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.353392150849365e-05,
+      "loss": 2.7613,
+      "step": 20570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.35263504731031e-05,
+      "loss": 2.7884,
+      "step": 20575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.351877804081798e-05,
+      "loss": 2.8179,
+      "step": 20580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.351120421195378e-05,
+      "loss": 2.8753,
+      "step": 20585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.350362898682606e-05,
+      "loss": 2.8568,
+      "step": 20590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.349605236575046e-05,
+      "loss": 2.9061,
+      "step": 20595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.348847434904267e-05,
+      "loss": 2.9413,
+      "step": 20600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.34808949370184e-05,
+      "loss": 2.8998,
+      "step": 20605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.347331412999348e-05,
+      "loss": 2.8282,
+      "step": 20610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.346573192828377e-05,
+      "loss": 2.8102,
+      "step": 20615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.345814833220516e-05,
+      "loss": 2.9363,
+      "step": 20620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.345056334207367e-05,
+      "loss": 2.842,
+      "step": 20625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.344297695820527e-05,
+      "loss": 2.9183,
+      "step": 20630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.343538918091611e-05,
+      "loss": 2.8037,
+      "step": 20635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.342780001052229e-05,
+      "loss": 2.805,
+      "step": 20640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.342020944734006e-05,
+      "loss": 2.8396,
+      "step": 20645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.341261749168568e-05,
+      "loss": 2.829,
+      "step": 20650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.340502414387544e-05,
+      "loss": 2.8368,
+      "step": 20655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.339742940422573e-05,
+      "loss": 2.7588,
+      "step": 20660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.338983327305303e-05,
+      "loss": 2.725,
+      "step": 20665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.338223575067379e-05,
+      "loss": 2.7922,
+      "step": 20670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.337463683740458e-05,
+      "loss": 2.8254,
+      "step": 20675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.336703653356204e-05,
+      "loss": 2.8429,
+      "step": 20680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.33594348394628e-05,
+      "loss": 2.9581,
+      "step": 20685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.335183175542359e-05,
+      "loss": 2.7779,
+      "step": 20690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.334422728176124e-05,
+      "loss": 2.8916,
+      "step": 20695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.333662141879255e-05,
+      "loss": 2.7971,
+      "step": 20700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.332901416683445e-05,
+      "loss": 2.9633,
+      "step": 20705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.332140552620388e-05,
+      "loss": 3.0269,
+      "step": 20710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.33137954972179e-05,
+      "loss": 2.9763,
+      "step": 20715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.330618408019353e-05,
+      "loss": 2.7126,
+      "step": 20720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.329857127544791e-05,
+      "loss": 2.8417,
+      "step": 20725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.329095708329829e-05,
+      "loss": 2.8571,
+      "step": 20730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.328334150406185e-05,
+      "loss": 2.7943,
+      "step": 20735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.327572453805594e-05,
+      "loss": 2.9583,
+      "step": 20740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.326810618559793e-05,
+      "loss": 2.8351,
+      "step": 20745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.326048644700521e-05,
+      "loss": 2.9569,
+      "step": 20750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.325286532259526e-05,
+      "loss": 2.8538,
+      "step": 20755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.324524281268567e-05,
+      "loss": 2.8286,
+      "step": 20760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.323761891759399e-05,
+      "loss": 2.8784,
+      "step": 20765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.322999363763788e-05,
+      "loss": 2.7322,
+      "step": 20770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.322236697313506e-05,
+      "loss": 2.8557,
+      "step": 20775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.321473892440331e-05,
+      "loss": 2.8463,
+      "step": 20780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.320710949176042e-05,
+      "loss": 2.9519,
+      "step": 20785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.319947867552431e-05,
+      "loss": 2.8649,
+      "step": 20790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.31918464760129e-05,
+      "loss": 2.9379,
+      "step": 20795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.318421289354422e-05,
+      "loss": 2.7339,
+      "step": 20800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.31765779284363e-05,
+      "loss": 2.9528,
+      "step": 20805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.316894158100727e-05,
+      "loss": 2.9434,
+      "step": 20810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.316130385157527e-05,
+      "loss": 2.8801,
+      "step": 20815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.315366474045861e-05,
+      "loss": 2.9678,
+      "step": 20820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.314602424797549e-05,
+      "loss": 2.8625,
+      "step": 20825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.313838237444428e-05,
+      "loss": 2.7811,
+      "step": 20830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.31307391201834e-05,
+      "loss": 2.7991,
+      "step": 20835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.312309448551132e-05,
+      "loss": 2.8105,
+      "step": 20840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.311544847074653e-05,
+      "loss": 2.9287,
+      "step": 20845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.310780107620764e-05,
+      "loss": 2.8664,
+      "step": 20850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.310015230221324e-05,
+      "loss": 2.7463,
+      "step": 20855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.309250214908204e-05,
+      "loss": 2.7332,
+      "step": 20860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.30848506171328e-05,
+      "loss": 2.8445,
+      "step": 20865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.30771977066843e-05,
+      "loss": 2.71,
+      "step": 20870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.306954341805541e-05,
+      "loss": 2.8463,
+      "step": 20875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.306188775156509e-05,
+      "loss": 2.8218,
+      "step": 20880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.305423070753227e-05,
+      "loss": 2.9364,
+      "step": 20885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.304657228627599e-05,
+      "loss": 2.7102,
+      "step": 20890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.303891248811537e-05,
+      "loss": 2.8664,
+      "step": 20895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.303125131336954e-05,
+      "loss": 2.7817,
+      "step": 20900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.30235887623577e-05,
+      "loss": 2.8174,
+      "step": 20905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.301592483539913e-05,
+      "loss": 2.8293,
+      "step": 20910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.300825953281314e-05,
+      "loss": 2.916,
+      "step": 20915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.300059285491914e-05,
+      "loss": 2.8362,
+      "step": 20920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.299292480203653e-05,
+      "loss": 2.9084,
+      "step": 20925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.298525537448482e-05,
+      "loss": 2.9012,
+      "step": 20930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.297758457258354e-05,
+      "loss": 2.8366,
+      "step": 20935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.296991239665233e-05,
+      "loss": 2.8641,
+      "step": 20940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.296223884701084e-05,
+      "loss": 2.8076,
+      "step": 20945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.29545639239788e-05,
+      "loss": 2.8787,
+      "step": 20950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.294688762787597e-05,
+      "loss": 2.878,
+      "step": 20955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.293920995902221e-05,
+      "loss": 2.8235,
+      "step": 20960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.293153091773741e-05,
+      "loss": 2.8292,
+      "step": 20965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.29238505043415e-05,
+      "loss": 2.8607,
+      "step": 20970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.291616871915454e-05,
+      "loss": 2.7997,
+      "step": 20975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.290848556249653e-05,
+      "loss": 2.8558,
+      "step": 20980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.290080103468765e-05,
+      "loss": 2.8441,
+      "step": 20985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.289311513604806e-05,
+      "loss": 2.93,
+      "step": 20990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.288542786689796e-05,
+      "loss": 2.9643,
+      "step": 20995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.287773922755769e-05,
+      "loss": 2.8265,
+      "step": 21000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.28700492183476e-05,
+      "loss": 2.8975,
+      "step": 21005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.286235783958809e-05,
+      "loss": 2.7502,
+      "step": 21010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.285466509159961e-05,
+      "loss": 2.7421,
+      "step": 21015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.284697097470271e-05,
+      "loss": 2.9435,
+      "step": 21020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.283927548921794e-05,
+      "loss": 2.9232,
+      "step": 21025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.283157863546597e-05,
+      "loss": 2.8127,
+      "step": 21030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.282388041376747e-05,
+      "loss": 2.8794,
+      "step": 21035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.28161808244432e-05,
+      "loss": 2.8392,
+      "step": 21040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.280847986781397e-05,
+      "loss": 2.7405,
+      "step": 21045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.280077754420062e-05,
+      "loss": 2.7727,
+      "step": 21050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.279307385392412e-05,
+      "loss": 2.8539,
+      "step": 21055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.27853687973054e-05,
+      "loss": 2.9515,
+      "step": 21060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.277766237466552e-05,
+      "loss": 2.625,
+      "step": 21065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.276995458632557e-05,
+      "loss": 2.8236,
+      "step": 21070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.276224543260671e-05,
+      "loss": 2.7142,
+      "step": 21075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.275453491383012e-05,
+      "loss": 2.9273,
+      "step": 21080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.274682303031709e-05,
+      "loss": 2.8074,
+      "step": 21085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.27391097823889e-05,
+      "loss": 2.8061,
+      "step": 21090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.273139517036699e-05,
+      "loss": 2.9422,
+      "step": 21095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.272367919457274e-05,
+      "loss": 2.8016,
+      "step": 21100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.271596185532766e-05,
+      "loss": 2.8473,
+      "step": 21105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.270824315295329e-05,
+      "loss": 2.8043,
+      "step": 21110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.270052308777125e-05,
+      "loss": 2.7369,
+      "step": 21115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.26928016601032e-05,
+      "loss": 2.8868,
+      "step": 21120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.268507887027084e-05,
+      "loss": 2.746,
+      "step": 21125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.267735471859595e-05,
+      "loss": 2.7271,
+      "step": 21130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.266962920540035e-05,
+      "loss": 2.8576,
+      "step": 21135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.266190233100598e-05,
+      "loss": 2.7357,
+      "step": 21140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.265417409573472e-05,
+      "loss": 2.7607,
+      "step": 21145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.264644449990861e-05,
+      "loss": 2.7621,
+      "step": 21150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.263871354384969e-05,
+      "loss": 2.9733,
+      "step": 21155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.263098122788008e-05,
+      "loss": 2.908,
+      "step": 21160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.262324755232196e-05,
+      "loss": 2.9622,
+      "step": 21165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.261551251749755e-05,
+      "loss": 2.7262,
+      "step": 21170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.260777612372913e-05,
+      "loss": 2.855,
+      "step": 21175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.260003837133905e-05,
+      "loss": 2.7786,
+      "step": 21180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.259229926064969e-05,
+      "loss": 2.7349,
+      "step": 21185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.258455879198354e-05,
+      "loss": 2.8526,
+      "step": 21190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.257681696566308e-05,
+      "loss": 2.9591,
+      "step": 21195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.256907378201091e-05,
+      "loss": 2.8291,
+      "step": 21200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.256132924134961e-05,
+      "loss": 2.7681,
+      "step": 21205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.255358334400187e-05,
+      "loss": 2.8973,
+      "step": 21210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.254583609029046e-05,
+      "loss": 2.7973,
+      "step": 21215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.253808748053816e-05,
+      "loss": 2.8285,
+      "step": 21220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.25303375150678e-05,
+      "loss": 2.7161,
+      "step": 21225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.25225861942023e-05,
+      "loss": 2.8791,
+      "step": 21230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.251483351826464e-05,
+      "loss": 2.9032,
+      "step": 21235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.25070794875778e-05,
+      "loss": 2.8012,
+      "step": 21240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.249932410246487e-05,
+      "loss": 2.7946,
+      "step": 21245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.249156736324904e-05,
+      "loss": 2.7523,
+      "step": 21250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.24838092702534e-05,
+      "loss": 2.9288,
+      "step": 21255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.247604982380128e-05,
+      "loss": 2.7648,
+      "step": 21260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.246828902421595e-05,
+      "loss": 2.8231,
+      "step": 21265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.246052687182076e-05,
+      "loss": 2.7702,
+      "step": 21270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.245276336693912e-05,
+      "loss": 2.8436,
+      "step": 21275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.244499850989452e-05,
+      "loss": 2.9064,
+      "step": 21280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.243723230101048e-05,
+      "loss": 2.8461,
+      "step": 21285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.242946474061058e-05,
+      "loss": 2.7837,
+      "step": 21290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.242169582901846e-05,
+      "loss": 2.7392,
+      "step": 21295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.241392556655783e-05,
+      "loss": 2.7482,
+      "step": 21300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.240615395355242e-05,
+      "loss": 2.9085,
+      "step": 21305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.239838099032606e-05,
+      "loss": 2.7913,
+      "step": 21310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.23906066772026e-05,
+      "loss": 2.7928,
+      "step": 21315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.238283101450597e-05,
+      "loss": 2.9063,
+      "step": 21320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.237505400256014e-05,
+      "loss": 2.7684,
+      "step": 21325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.236727564168914e-05,
+      "loss": 2.9124,
+      "step": 21330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.235949593221706e-05,
+      "loss": 2.9235,
+      "step": 21335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.235171487446806e-05,
+      "loss": 2.732,
+      "step": 21340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.234393246876635e-05,
+      "loss": 2.6633,
+      "step": 21345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.233614871543617e-05,
+      "loss": 2.8382,
+      "step": 21350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.232836361480183e-05,
+      "loss": 2.886,
+      "step": 21355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.232057716718771e-05,
+      "loss": 2.7865,
+      "step": 21360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.231278937291825e-05,
+      "loss": 2.742,
+      "step": 21365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.230500023231789e-05,
+      "loss": 2.8305,
+      "step": 21370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.229720974571123e-05,
+      "loss": 2.8583,
+      "step": 21375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.228941791342279e-05,
+      "loss": 2.8363,
+      "step": 21380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.22816247357773e-05,
+      "loss": 2.8377,
+      "step": 21385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.227383021309943e-05,
+      "loss": 2.8664,
+      "step": 21390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.226603434571394e-05,
+      "loss": 2.9228,
+      "step": 21395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.225823713394564e-05,
+      "loss": 2.7442,
+      "step": 21400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.225043857811943e-05,
+      "loss": 2.8426,
+      "step": 21405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.224263867856024e-05,
+      "loss": 2.8851,
+      "step": 21410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.223483743559301e-05,
+      "loss": 2.7264,
+      "step": 21415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.222703484954286e-05,
+      "loss": 2.8276,
+      "step": 21420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.221923092073484e-05,
+      "loss": 2.6728,
+      "step": 21425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.22114256494941e-05,
+      "loss": 2.8755,
+      "step": 21430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.220361903614588e-05,
+      "loss": 2.8443,
+      "step": 21435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.219581108101544e-05,
+      "loss": 2.8347,
+      "step": 21440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.218800178442807e-05,
+      "loss": 2.8231,
+      "step": 21445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.218019114670921e-05,
+      "loss": 2.7984,
+      "step": 21450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.217237916818423e-05,
+      "loss": 2.8748,
+      "step": 21455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.216456584917866e-05,
+      "loss": 2.7867,
+      "step": 21460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.215675119001804e-05,
+      "loss": 2.9212,
+      "step": 21465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.214893519102797e-05,
+      "loss": 2.8648,
+      "step": 21470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.21411178525341e-05,
+      "loss": 2.7815,
+      "step": 21475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.213329917486215e-05,
+      "loss": 2.7196,
+      "step": 21480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.21254791583379e-05,
+      "loss": 2.6825,
+      "step": 21485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.211765780328716e-05,
+      "loss": 2.7757,
+      "step": 21490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.210983511003582e-05,
+      "loss": 2.8593,
+      "step": 21495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.210201107890981e-05,
+      "loss": 2.8535,
+      "step": 21500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.209418571023513e-05,
+      "loss": 2.6497,
+      "step": 21505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.208635900433783e-05,
+      "loss": 2.8243,
+      "step": 21510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.207853096154401e-05,
+      "loss": 2.8927,
+      "step": 21515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.207070158217981e-05,
+      "loss": 2.6815,
+      "step": 21520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.206287086657147e-05,
+      "loss": 2.7605,
+      "step": 21525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.205503881504528e-05,
+      "loss": 2.7835,
+      "step": 21530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.204720542792753e-05,
+      "loss": 2.7539,
+      "step": 21535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.203937070554461e-05,
+      "loss": 2.8604,
+      "step": 21540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.203153464822298e-05,
+      "loss": 2.7754,
+      "step": 21545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.20236972562891e-05,
+      "loss": 2.8258,
+      "step": 21550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.201585853006953e-05,
+      "loss": 2.7878,
+      "step": 21555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.20080184698909e-05,
+      "loss": 2.8529,
+      "step": 21560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.200017707607985e-05,
+      "loss": 2.7139,
+      "step": 21565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.199233434896309e-05,
+      "loss": 2.6667,
+      "step": 21570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.198449028886741e-05,
+      "loss": 2.8916,
+      "step": 21575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.197664489611962e-05,
+      "loss": 2.8631,
+      "step": 21580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.196879817104662e-05,
+      "loss": 2.8112,
+      "step": 21585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.196095011397534e-05,
+      "loss": 2.822,
+      "step": 21590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.195310072523277e-05,
+      "loss": 2.9301,
+      "step": 21595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.194525000514594e-05,
+      "loss": 2.8553,
+      "step": 21600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.1937397954042e-05,
+      "loss": 2.9168,
+      "step": 21605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.192954457224809e-05,
+      "loss": 2.8189,
+      "step": 21610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.19216898600914e-05,
+      "loss": 2.7265,
+      "step": 21615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.191383381789926e-05,
+      "loss": 2.9251,
+      "step": 21620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.190597644599894e-05,
+      "loss": 2.8298,
+      "step": 21625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.189811774471783e-05,
+      "loss": 2.8901,
+      "step": 21630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.18902577143834e-05,
+      "loss": 2.7688,
+      "step": 21635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.188239635532311e-05,
+      "loss": 2.8476,
+      "step": 21640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.187453366786453e-05,
+      "loss": 2.8353,
+      "step": 21645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.186666965233524e-05,
+      "loss": 2.8224,
+      "step": 21650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.185880430906292e-05,
+      "loss": 2.88,
+      "step": 21655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.185093763837528e-05,
+      "loss": 2.8702,
+      "step": 21660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.184306964060009e-05,
+      "loss": 2.7859,
+      "step": 21665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.183520031606516e-05,
+      "loss": 2.7979,
+      "step": 21670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.182732966509838e-05,
+      "loss": 2.7278,
+      "step": 21675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.181945768802768e-05,
+      "loss": 2.7511,
+      "step": 21680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.181158438518107e-05,
+      "loss": 2.7425,
+      "step": 21685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.180370975688657e-05,
+      "loss": 2.8466,
+      "step": 21690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.179583380347228e-05,
+      "loss": 2.858,
+      "step": 21695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.17879565252664e-05,
+      "loss": 2.7343,
+      "step": 21700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.178007792259709e-05,
+      "loss": 2.8415,
+      "step": 21705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.177219799579264e-05,
+      "loss": 2.8618,
+      "step": 21710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.176431674518137e-05,
+      "loss": 2.8291,
+      "step": 21715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.175643417109165e-05,
+      "loss": 2.9127,
+      "step": 21720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.174855027385193e-05,
+      "loss": 2.7838,
+      "step": 21725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.174066505379067e-05,
+      "loss": 2.7389,
+      "step": 21730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.173277851123643e-05,
+      "loss": 2.8205,
+      "step": 21735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.17248906465178e-05,
+      "loss": 2.7177,
+      "step": 21740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.171700145996343e-05,
+      "loss": 2.895,
+      "step": 21745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.170911095190206e-05,
+      "loss": 2.8173,
+      "step": 21750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.17012191226624e-05,
+      "loss": 2.747,
+      "step": 21755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.169332597257331e-05,
+      "loss": 2.8114,
+      "step": 21760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.168543150196361e-05,
+      "loss": 2.9443,
+      "step": 21765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.16775357111623e-05,
+      "loss": 2.8381,
+      "step": 21770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.166963860049832e-05,
+      "loss": 2.7811,
+      "step": 21775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.166174017030072e-05,
+      "loss": 2.7258,
+      "step": 21780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.165384042089858e-05,
+      "loss": 2.7681,
+      "step": 21785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.164593935262103e-05,
+      "loss": 2.81,
+      "step": 21790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.163803696579733e-05,
+      "loss": 2.8644,
+      "step": 21795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.163013326075668e-05,
+      "loss": 2.7026,
+      "step": 21800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.162222823782843e-05,
+      "loss": 2.8335,
+      "step": 21805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.161432189734193e-05,
+      "loss": 2.7858,
+      "step": 21810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.160641423962661e-05,
+      "loss": 2.8263,
+      "step": 21815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.159850526501195e-05,
+      "loss": 2.7086,
+      "step": 21820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.159059497382746e-05,
+      "loss": 2.7539,
+      "step": 21825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.158268336640275e-05,
+      "loss": 2.8215,
+      "step": 21830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.157477044306745e-05,
+      "loss": 2.8676,
+      "step": 21835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.156685620415125e-05,
+      "loss": 2.9289,
+      "step": 21840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.155894064998392e-05,
+      "loss": 2.7548,
+      "step": 21845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.155102378089527e-05,
+      "loss": 2.8086,
+      "step": 21850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.154310559721514e-05,
+      "loss": 2.7916,
+      "step": 21855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.153518609927344e-05,
+      "loss": 2.7206,
+      "step": 21860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.152726528740017e-05,
+      "loss": 2.8006,
+      "step": 21865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.151934316192534e-05,
+      "loss": 2.802,
+      "step": 21870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.151141972317902e-05,
+      "loss": 2.8197,
+      "step": 21875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.150349497149136e-05,
+      "loss": 2.8815,
+      "step": 21880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.149556890719253e-05,
+      "loss": 2.701,
+      "step": 21885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.148764153061279e-05,
+      "loss": 2.8184,
+      "step": 21890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.147971284208245e-05,
+      "loss": 2.8122,
+      "step": 21895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.147178284193184e-05,
+      "loss": 2.7871,
+      "step": 21900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.146385153049139e-05,
+      "loss": 2.8076,
+      "step": 21905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.145591890809154e-05,
+      "loss": 2.9428,
+      "step": 21910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.144798497506282e-05,
+      "loss": 2.8419,
+      "step": 21915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.144004973173578e-05,
+      "loss": 2.7597,
+      "step": 21920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.14321131784411e-05,
+      "loss": 2.6668,
+      "step": 21925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.142417531550941e-05,
+      "loss": 2.8297,
+      "step": 21930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.141623614327145e-05,
+      "loss": 2.8956,
+      "step": 21935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.140829566205803e-05,
+      "loss": 2.7861,
+      "step": 21940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.14003538722e-05,
+      "loss": 2.7351,
+      "step": 21945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.139241077402823e-05,
+      "loss": 2.7676,
+      "step": 21950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.138446636787369e-05,
+      "loss": 2.8807,
+      "step": 21955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.13765206540674e-05,
+      "loss": 2.9202,
+      "step": 21960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.136857363294041e-05,
+      "loss": 2.8387,
+      "step": 21965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.136062530482382e-05,
+      "loss": 2.9035,
+      "step": 21970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.135267567004884e-05,
+      "loss": 2.8192,
+      "step": 21975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.134472472894668e-05,
+      "loss": 2.9353,
+      "step": 21980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.133677248184858e-05,
+      "loss": 2.8325,
+      "step": 21985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.132881892908595e-05,
+      "loss": 2.8135,
+      "step": 21990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.132086407099013e-05,
+      "loss": 2.7453,
+      "step": 21995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.131290790789257e-05,
+      "loss": 2.8583,
+      "step": 22000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.130495044012478e-05,
+      "loss": 2.8046,
+      "step": 22005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.12969916680183e-05,
+      "loss": 2.7617,
+      "step": 22010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.128903159190472e-05,
+      "loss": 2.7107,
+      "step": 22015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.128107021211575e-05,
+      "loss": 2.8925,
+      "step": 22020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.127310752898308e-05,
+      "loss": 2.9172,
+      "step": 22025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.126514354283845e-05,
+      "loss": 2.8172,
+      "step": 22030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.125717825401375e-05,
+      "loss": 2.9165,
+      "step": 22035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.12492116628408e-05,
+      "loss": 2.7359,
+      "step": 22040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.124124376965155e-05,
+      "loss": 2.8562,
+      "step": 22045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.123327457477798e-05,
+      "loss": 2.8355,
+      "step": 22050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.122530407855216e-05,
+      "loss": 2.7934,
+      "step": 22055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.121733228130617e-05,
+      "loss": 2.806,
+      "step": 22060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.120935918337213e-05,
+      "loss": 2.8267,
+      "step": 22065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.120138478508231e-05,
+      "loss": 2.8957,
+      "step": 22070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.119340908676889e-05,
+      "loss": 2.77,
+      "step": 22075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.118543208876423e-05,
+      "loss": 2.9082,
+      "step": 22080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.117745379140068e-05,
+      "loss": 2.9228,
+      "step": 22085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.116947419501069e-05,
+      "loss": 2.9314,
+      "step": 22090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.116149329992668e-05,
+      "loss": 2.8198,
+      "step": 22095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.115351110648122e-05,
+      "loss": 2.7416,
+      "step": 22100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.114552761500688e-05,
+      "loss": 2.8475,
+      "step": 22105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.11375428258363e-05,
+      "loss": 2.7781,
+      "step": 22110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.112955673930218e-05,
+      "loss": 2.7716,
+      "step": 22115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.112156935573724e-05,
+      "loss": 2.8906,
+      "step": 22120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.11135806754743e-05,
+      "loss": 2.9252,
+      "step": 22125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.11055906988462e-05,
+      "loss": 2.8436,
+      "step": 22130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.109759942618584e-05,
+      "loss": 2.8044,
+      "step": 22135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.108960685782622e-05,
+      "loss": 2.8847,
+      "step": 22140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.108161299410031e-05,
+      "loss": 2.7569,
+      "step": 22145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.107361783534121e-05,
+      "loss": 2.8493,
+      "step": 22150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.106562138188204e-05,
+      "loss": 2.7835,
+      "step": 22155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.105762363405593e-05,
+      "loss": 2.8519,
+      "step": 22160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.104962459219616e-05,
+      "loss": 2.9695,
+      "step": 22165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.104162425663602e-05,
+      "loss": 2.6639,
+      "step": 22170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.103362262770881e-05,
+      "loss": 2.8346,
+      "step": 22175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.102561970574798e-05,
+      "loss": 2.9066,
+      "step": 22180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.10176154910869e-05,
+      "loss": 2.8312,
+      "step": 22185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.100960998405914e-05,
+      "loss": 2.8645,
+      "step": 22190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.10016031849982e-05,
+      "loss": 2.7821,
+      "step": 22195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.099359509423772e-05,
+      "loss": 2.7629,
+      "step": 22200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.098558571211135e-05,
+      "loss": 2.801,
+      "step": 22205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.097757503895282e-05,
+      "loss": 2.9012,
+      "step": 22210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.096956307509588e-05,
+      "loss": 2.7833,
+      "step": 22215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.096154982087438e-05,
+      "loss": 2.6463,
+      "step": 22220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.095353527662215e-05,
+      "loss": 2.7943,
+      "step": 22225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.094551944267317e-05,
+      "loss": 2.8002,
+      "step": 22230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.093750231936139e-05,
+      "loss": 2.6841,
+      "step": 22235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.092948390702087e-05,
+      "loss": 2.75,
+      "step": 22240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.092146420598568e-05,
+      "loss": 2.8198,
+      "step": 22245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.091344321658999e-05,
+      "loss": 2.6981,
+      "step": 22250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.090542093916799e-05,
+      "loss": 2.8914,
+      "step": 22255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.089739737405392e-05,
+      "loss": 2.7466,
+      "step": 22260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.088937252158211e-05,
+      "loss": 2.8532,
+      "step": 22265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.088134638208689e-05,
+      "loss": 2.7974,
+      "step": 22270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.08733189559027e-05,
+      "loss": 2.7925,
+      "step": 22275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.0865290243364e-05,
+      "loss": 2.7893,
+      "step": 22280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.08572602448053e-05,
+      "loss": 2.8797,
+      "step": 22285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.084922896056117e-05,
+      "loss": 2.7457,
+      "step": 22290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.084119639096628e-05,
+      "loss": 2.742,
+      "step": 22295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.083316253635527e-05,
+      "loss": 2.858,
+      "step": 22300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.08251273970629e-05,
+      "loss": 2.8365,
+      "step": 22305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.081709097342394e-05,
+      "loss": 3.0244,
+      "step": 22310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.080905326577324e-05,
+      "loss": 2.8499,
+      "step": 22315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.080101427444568e-05,
+      "loss": 2.709,
+      "step": 22320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.079297399977624e-05,
+      "loss": 2.7955,
+      "step": 22325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.07849324420999e-05,
+      "loss": 2.6179,
+      "step": 22330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.077688960175172e-05,
+      "loss": 2.7663,
+      "step": 22335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.076884547906681e-05,
+      "loss": 2.8792,
+      "step": 22340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.076080007438034e-05,
+      "loss": 2.788,
+      "step": 22345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.075275338802752e-05,
+      "loss": 2.815,
+      "step": 22350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.074470542034362e-05,
+      "loss": 2.8925,
+      "step": 22355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.073665617166397e-05,
+      "loss": 2.8252,
+      "step": 22360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.072860564232393e-05,
+      "loss": 2.8465,
+      "step": 22365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.072055383265895e-05,
+      "loss": 2.6822,
+      "step": 22370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.071250074300451e-05,
+      "loss": 2.7763,
+      "step": 22375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.070444637369612e-05,
+      "loss": 2.6849,
+      "step": 22380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.06963907250694e-05,
+      "loss": 2.765,
+      "step": 22385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.068833379746e-05,
+      "loss": 2.8557,
+      "step": 22390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.068027559120359e-05,
+      "loss": 2.7163,
+      "step": 22395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.067221610663594e-05,
+      "loss": 2.6849,
+      "step": 22400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.066415534409281e-05,
+      "loss": 2.7577,
+      "step": 22405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.065609330391013e-05,
+      "loss": 2.9535,
+      "step": 22410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.064802998642374e-05,
+      "loss": 2.6609,
+      "step": 22415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.063996539196962e-05,
+      "loss": 2.7703,
+      "step": 22420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.063189952088384e-05,
+      "loss": 2.8175,
+      "step": 22425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.062383237350239e-05,
+      "loss": 2.7152,
+      "step": 22430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.061576395016146e-05,
+      "loss": 2.8627,
+      "step": 22435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.060769425119718e-05,
+      "loss": 2.6993,
+      "step": 22440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.05996232769458e-05,
+      "loss": 2.815,
+      "step": 22445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.059155102774358e-05,
+      "loss": 2.7816,
+      "step": 22450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.058347750392687e-05,
+      "loss": 2.7424,
+      "step": 22455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.057540270583205e-05,
+      "loss": 2.7043,
+      "step": 22460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.05673266337956e-05,
+      "loss": 2.8059,
+      "step": 22465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.055924928815397e-05,
+      "loss": 2.9364,
+      "step": 22470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.055117066924372e-05,
+      "loss": 2.9591,
+      "step": 22475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.054309077740143e-05,
+      "loss": 2.8167,
+      "step": 22480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.053500961296381e-05,
+      "loss": 2.7491,
+      "step": 22485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.05269271762675e-05,
+      "loss": 2.8189,
+      "step": 22490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.051884346764931e-05,
+      "loss": 2.9975,
+      "step": 22495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.051075848744603e-05,
+      "loss": 2.8249,
+      "step": 22500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.05026722359945e-05,
+      "loss": 2.8814,
+      "step": 22505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.04945847136317e-05,
+      "loss": 2.8387,
+      "step": 22510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.048649592069455e-05,
+      "loss": 2.779,
+      "step": 22515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.047840585752009e-05,
+      "loss": 2.6404,
+      "step": 22520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.04703145244454e-05,
+      "loss": 2.9152,
+      "step": 22525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.046222192180761e-05,
+      "loss": 2.7803,
+      "step": 22530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.04541280499439e-05,
+      "loss": 2.8235,
+      "step": 22535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.044603290919151e-05,
+      "loss": 2.7455,
+      "step": 22540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.043793649988773e-05,
+      "loss": 2.882,
+      "step": 22545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.04298388223699e-05,
+      "loss": 2.8881,
+      "step": 22550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.042173987697538e-05,
+      "loss": 2.8281,
+      "step": 22555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.041363966404168e-05,
+      "loss": 2.7427,
+      "step": 22560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.040553818390625e-05,
+      "loss": 2.7333,
+      "step": 22565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.039743543690667e-05,
+      "loss": 2.7118,
+      "step": 22570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.038933142338054e-05,
+      "loss": 2.771,
+      "step": 22575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.038122614366552e-05,
+      "loss": 2.7128,
+      "step": 22580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.037311959809933e-05,
+      "loss": 2.9397,
+      "step": 22585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.036501178701969e-05,
+      "loss": 2.694,
+      "step": 22590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.035690271076444e-05,
+      "loss": 2.8001,
+      "step": 22595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.034879236967149e-05,
+      "loss": 2.8264,
+      "step": 22600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.03406807640787e-05,
+      "loss": 2.9224,
+      "step": 22605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.03325678943241e-05,
+      "loss": 2.6841,
+      "step": 22610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.032445376074566e-05,
+      "loss": 2.897,
+      "step": 22615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.031633836368152e-05,
+      "loss": 2.8445,
+      "step": 22620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.030822170346977e-05,
+      "loss": 2.7167,
+      "step": 22625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.03001037804486e-05,
+      "loss": 2.8169,
+      "step": 22630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.029198459495626e-05,
+      "loss": 2.8541,
+      "step": 22635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.028386414733105e-05,
+      "loss": 2.7048,
+      "step": 22640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.027574243791129e-05,
+      "loss": 2.8219,
+      "step": 22645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.026761946703538e-05,
+      "loss": 2.8432,
+      "step": 22650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.025949523504177e-05,
+      "loss": 2.9343,
+      "step": 22655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.025136974226899e-05,
+      "loss": 2.7011,
+      "step": 22660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.024324298905555e-05,
+      "loss": 2.8329,
+      "step": 22665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.023511497574008e-05,
+      "loss": 2.8246,
+      "step": 22670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.022698570266123e-05,
+      "loss": 2.8142,
+      "step": 22675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.02188551701577e-05,
+      "loss": 2.8423,
+      "step": 22680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.021072337856827e-05,
+      "loss": 2.8238,
+      "step": 22685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.020259032823176e-05,
+      "loss": 2.8407,
+      "step": 22690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.019445601948704e-05,
+      "loss": 2.8365,
+      "step": 22695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.0186320452673e-05,
+      "loss": 2.7053,
+      "step": 22700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.017818362812866e-05,
+      "loss": 2.9347,
+      "step": 22705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.017004554619298e-05,
+      "loss": 2.6307,
+      "step": 22710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.016190620720512e-05,
+      "loss": 2.6486,
+      "step": 22715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.015376561150414e-05,
+      "loss": 2.8981,
+      "step": 22720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.014562375942925e-05,
+      "loss": 2.6767,
+      "step": 22725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.01374806513197e-05,
+      "loss": 2.744,
+      "step": 22730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.012933628751475e-05,
+      "loss": 2.7597,
+      "step": 22735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.012119066835375e-05,
+      "loss": 2.8143,
+      "step": 22740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.01130437941761e-05,
+      "loss": 2.7545,
+      "step": 22745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.010489566532124e-05,
+      "loss": 2.7693,
+      "step": 22750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.009674628212866e-05,
+      "loss": 2.8715,
+      "step": 22755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.008859564493791e-05,
+      "loss": 2.798,
+      "step": 22760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.00804437540886e-05,
+      "loss": 2.9693,
+      "step": 22765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.007229060992038e-05,
+      "loss": 2.8196,
+      "step": 22770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.006413621277294e-05,
+      "loss": 2.7598,
+      "step": 22775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.005598056298605e-05,
+      "loss": 2.7974,
+      "step": 22780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.004782366089954e-05,
+      "loss": 2.8276,
+      "step": 22785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.003966550685324e-05,
+      "loss": 2.8845,
+      "step": 22790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.003150610118708e-05,
+      "loss": 2.7847,
+      "step": 22795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.002334544424102e-05,
+      "loss": 2.6422,
+      "step": 22800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.001518353635507e-05,
+      "loss": 2.8828,
+      "step": 22805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.000702037786931e-05,
+      "loss": 2.7558,
+      "step": 22810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.999885596912387e-05,
+      "loss": 2.7218,
+      "step": 22815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.999069031045894e-05,
+      "loss": 2.7208,
+      "step": 22820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.99825234022147e-05,
+      "loss": 2.9703,
+      "step": 22825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.997435524473148e-05,
+      "loss": 2.8839,
+      "step": 22830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.996618583834955e-05,
+      "loss": 2.7748,
+      "step": 22835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.995801518340934e-05,
+      "loss": 2.7642,
+      "step": 22840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.994984328025128e-05,
+      "loss": 2.7723,
+      "step": 22845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.994167012921586e-05,
+      "loss": 2.7882,
+      "step": 22850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.99334957306436e-05,
+      "loss": 2.8683,
+      "step": 22855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.99253200848751e-05,
+      "loss": 2.6766,
+      "step": 22860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.991714319225101e-05,
+      "loss": 2.7096,
+      "step": 22865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.9908965053112e-05,
+      "loss": 2.8139,
+      "step": 22870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.990078566779885e-05,
+      "loss": 2.9626,
+      "step": 22875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.989260503665234e-05,
+      "loss": 2.796,
+      "step": 22880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.988442316001332e-05,
+      "loss": 2.7551,
+      "step": 22885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.987624003822272e-05,
+      "loss": 2.9679,
+      "step": 22890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.986805567162146e-05,
+      "loss": 2.9264,
+      "step": 22895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.985987006055055e-05,
+      "loss": 2.7503,
+      "step": 22900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.985168320535106e-05,
+      "loss": 2.7823,
+      "step": 22905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.984349510636408e-05,
+      "loss": 2.683,
+      "step": 22910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.983530576393081e-05,
+      "loss": 2.7531,
+      "step": 22915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.982711517839244e-05,
+      "loss": 2.7868,
+      "step": 22920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.981892335009025e-05,
+      "loss": 2.9226,
+      "step": 22925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.981073027936551e-05,
+      "loss": 2.8978,
+      "step": 22930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.980253596655965e-05,
+      "loss": 2.8071,
+      "step": 22935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.979434041201406e-05,
+      "loss": 2.8463,
+      "step": 22940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.978614361607019e-05,
+      "loss": 2.7753,
+      "step": 22945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.977794557906963e-05,
+      "loss": 2.7814,
+      "step": 22950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.97697463013539e-05,
+      "loss": 2.8388,
+      "step": 22955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.976154578326465e-05,
+      "loss": 2.9183,
+      "step": 22960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.975334402514354e-05,
+      "loss": 2.8926,
+      "step": 22965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.974514102733233e-05,
+      "loss": 2.7444,
+      "step": 22970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.973693679017277e-05,
+      "loss": 2.642,
+      "step": 22975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.972873131400673e-05,
+      "loss": 2.7875,
+      "step": 22980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.972052459917606e-05,
+      "loss": 2.768,
+      "step": 22985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.971231664602273e-05,
+      "loss": 2.716,
+      "step": 22990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.97041074548887e-05,
+      "loss": 2.8894,
+      "step": 22995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.969589702611603e-05,
+      "loss": 2.7682,
+      "step": 23000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.96876853600468e-05,
+      "loss": 2.7515,
+      "step": 23005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.967947245702319e-05,
+      "loss": 2.6901,
+      "step": 23010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.967125831738733e-05,
+      "loss": 2.9775,
+      "step": 23015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.966304294148152e-05,
+      "loss": 2.7479,
+      "step": 23020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.965482632964802e-05,
+      "loss": 2.9838,
+      "step": 23025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.964660848222922e-05,
+      "loss": 2.8546,
+      "step": 23030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.963838939956752e-05,
+      "loss": 2.7282,
+      "step": 23035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.963016908200534e-05,
+      "loss": 2.9207,
+      "step": 23040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.962194752988519e-05,
+      "loss": 2.6924,
+      "step": 23045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.961372474354964e-05,
+      "loss": 2.7253,
+      "step": 23050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.960550072334127e-05,
+      "loss": 2.6869,
+      "step": 23055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.959727546960278e-05,
+      "loss": 2.8438,
+      "step": 23060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.958904898267687e-05,
+      "loss": 2.8542,
+      "step": 23065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.95808212629063e-05,
+      "loss": 2.9606,
+      "step": 23070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.957259231063384e-05,
+      "loss": 2.7441,
+      "step": 23075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.956436212620241e-05,
+      "loss": 2.8918,
+      "step": 23080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.955613070995489e-05,
+      "loss": 2.7084,
+      "step": 23085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.954789806223427e-05,
+      "loss": 2.6719,
+      "step": 23090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.953966418338357e-05,
+      "loss": 2.9763,
+      "step": 23095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.953142907374582e-05,
+      "loss": 2.7593,
+      "step": 23100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.952319273366418e-05,
+      "loss": 2.7974,
+      "step": 23105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.951495516348182e-05,
+      "loss": 2.7831,
+      "step": 23110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.950671636354194e-05,
+      "loss": 2.6819,
+      "step": 23115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.949847633418784e-05,
+      "loss": 2.874,
+      "step": 23120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.949023507576282e-05,
+      "loss": 2.7492,
+      "step": 23125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.948199258861028e-05,
+      "loss": 2.7737,
+      "step": 23130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.947374887307364e-05,
+      "loss": 2.787,
+      "step": 23135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.946550392949637e-05,
+      "loss": 2.8733,
+      "step": 23140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.9457257758222e-05,
+      "loss": 2.8452,
+      "step": 23145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.944901035959413e-05,
+      "loss": 2.7566,
+      "step": 23150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.944076173395637e-05,
+      "loss": 2.7186,
+      "step": 23155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.943251188165242e-05,
+      "loss": 2.7327,
+      "step": 23160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.9424260803026e-05,
+      "loss": 2.8089,
+      "step": 23165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.941600849842092e-05,
+      "loss": 2.8851,
+      "step": 23170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.9407754968181e-05,
+      "loss": 2.7099,
+      "step": 23175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.939950021265013e-05,
+      "loss": 2.6991,
+      "step": 23180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.939124423217226e-05,
+      "loss": 2.6969,
+      "step": 23185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.938298702709135e-05,
+      "loss": 2.9028,
+      "step": 23190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.937472859775147e-05,
+      "loss": 2.8131,
+      "step": 23195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.936646894449669e-05,
+      "loss": 2.8898,
+      "step": 23200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.935820806767119e-05,
+      "loss": 2.7377,
+      "step": 23205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.934994596761913e-05,
+      "loss": 2.6802,
+      "step": 23210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.934168264468476e-05,
+      "loss": 2.8539,
+      "step": 23215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.933341809921238e-05,
+      "loss": 2.9551,
+      "step": 23220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.932515233154635e-05,
+      "loss": 2.8306,
+      "step": 23225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.931688534203105e-05,
+      "loss": 2.7596,
+      "step": 23230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.930861713101095e-05,
+      "loss": 2.8937,
+      "step": 23235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.93003476988305e-05,
+      "loss": 2.6403,
+      "step": 23240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.929207704583432e-05,
+      "loss": 2.8179,
+      "step": 23245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.928380517236695e-05,
+      "loss": 2.6478,
+      "step": 23250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.927553207877309e-05,
+      "loss": 2.733,
+      "step": 23255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.926725776539743e-05,
+      "loss": 2.8804,
+      "step": 23260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.92589822325847e-05,
+      "loss": 2.7489,
+      "step": 23265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.925070548067972e-05,
+      "loss": 2.6766,
+      "step": 23270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.924242751002735e-05,
+      "loss": 2.8691,
+      "step": 23275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.923414832097251e-05,
+      "loss": 2.7225,
+      "step": 23280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.922586791386013e-05,
+      "loss": 2.7783,
+      "step": 23285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.921758628903522e-05,
+      "loss": 2.6827,
+      "step": 23290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.920930344684286e-05,
+      "loss": 2.7988,
+      "step": 23295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.920101938762814e-05,
+      "loss": 2.8507,
+      "step": 23300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.919273411173623e-05,
+      "loss": 2.7425,
+      "step": 23305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.918444761951233e-05,
+      "loss": 2.7484,
+      "step": 23310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.917615991130172e-05,
+      "loss": 2.899,
+      "step": 23315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.91678709874497e-05,
+      "loss": 2.8382,
+      "step": 23320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.915958084830162e-05,
+      "loss": 2.7459,
+      "step": 23325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.915128949420292e-05,
+      "loss": 2.8155,
+      "step": 23330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.914299692549904e-05,
+      "loss": 2.9185,
+      "step": 23335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.913470314253552e-05,
+      "loss": 2.8292,
+      "step": 23340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.912640814565791e-05,
+      "loss": 2.807,
+      "step": 23345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.911811193521182e-05,
+      "loss": 2.7997,
+      "step": 23350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.910981451154292e-05,
+      "loss": 2.9175,
+      "step": 23355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.910151587499695e-05,
+      "loss": 2.8169,
+      "step": 23360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.909321602591963e-05,
+      "loss": 2.7536,
+      "step": 23365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.908491496465682e-05,
+      "loss": 2.8522,
+      "step": 23370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.907661269155438e-05,
+      "loss": 2.7915,
+      "step": 23375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.90683092069582e-05,
+      "loss": 2.7614,
+      "step": 23380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.90600045112143e-05,
+      "loss": 2.8611,
+      "step": 23385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.905169860466866e-05,
+      "loss": 2.9247,
+      "step": 23390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.904339148766735e-05,
+      "loss": 2.7707,
+      "step": 23395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.903508316055652e-05,
+      "loss": 2.7092,
+      "step": 23400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.902677362368234e-05,
+      "loss": 2.7458,
+      "step": 23405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.901846287739099e-05,
+      "loss": 2.6921,
+      "step": 23410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.901015092202877e-05,
+      "loss": 2.8175,
+      "step": 23415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.900183775794201e-05,
+      "loss": 2.8041,
+      "step": 23420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.899352338547707e-05,
+      "loss": 2.6976,
+      "step": 23425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.898520780498038e-05,
+      "loss": 2.8121,
+      "step": 23430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.897689101679841e-05,
+      "loss": 2.7642,
+      "step": 23435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.896857302127767e-05,
+      "loss": 2.6999,
+      "step": 23440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.896025381876477e-05,
+      "loss": 2.7841,
+      "step": 23445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.89519334096063e-05,
+      "loss": 2.9242,
+      "step": 23450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.894361179414896e-05,
+      "loss": 2.7634,
+      "step": 23455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.893528897273943e-05,
+      "loss": 2.8248,
+      "step": 23460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.892696494572453e-05,
+      "loss": 2.7734,
+      "step": 23465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.89186397134511e-05,
+      "loss": 2.8146,
+      "step": 23470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.891031327626596e-05,
+      "loss": 2.651,
+      "step": 23475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.890198563451607e-05,
+      "loss": 2.7987,
+      "step": 23480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.889365678854838e-05,
+      "loss": 2.7287,
+      "step": 23485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.888532673870997e-05,
+      "loss": 2.7714,
+      "step": 23490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.887699548534786e-05,
+      "loss": 2.8474,
+      "step": 23495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.886866302880921e-05,
+      "loss": 2.848,
+      "step": 23500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.886032936944117e-05,
+      "loss": 2.76,
+      "step": 23505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.885199450759099e-05,
+      "loss": 2.8102,
+      "step": 23510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.884365844360594e-05,
+      "loss": 2.7691,
+      "step": 23515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.883532117783334e-05,
+      "loss": 2.8221,
+      "step": 23520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.88269827106206e-05,
+      "loss": 2.865,
+      "step": 23525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.881864304231508e-05,
+      "loss": 2.7,
+      "step": 23530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.88103021732643e-05,
+      "loss": 2.6476,
+      "step": 23535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.88019601038158e-05,
+      "loss": 2.7774,
+      "step": 23540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.879361683431715e-05,
+      "loss": 2.8642,
+      "step": 23545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.878527236511595e-05,
+      "loss": 2.7916,
+      "step": 23550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.87769266965599e-05,
+      "loss": 2.819,
+      "step": 23555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.876857982899671e-05,
+      "loss": 2.863,
+      "step": 23560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.876023176277419e-05,
+      "loss": 2.7272,
+      "step": 23565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.875188249824013e-05,
+      "loss": 2.8717,
+      "step": 23570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.874353203574242e-05,
+      "loss": 2.7532,
+      "step": 23575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.8735180375629e-05,
+      "loss": 2.7154,
+      "step": 23580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.872682751824782e-05,
+      "loss": 2.8587,
+      "step": 23585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.871847346394694e-05,
+      "loss": 2.7055,
+      "step": 23590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.871011821307442e-05,
+      "loss": 2.7125,
+      "step": 23595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.870176176597837e-05,
+      "loss": 2.8499,
+      "step": 23600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.869340412300699e-05,
+      "loss": 2.8324,
+      "step": 23605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.868504528450848e-05,
+      "loss": 2.8304,
+      "step": 23610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.867668525083117e-05,
+      "loss": 2.7162,
+      "step": 23615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.866832402232331e-05,
+      "loss": 2.7861,
+      "step": 23620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.865996159933334e-05,
+      "loss": 2.8032,
+      "step": 23625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.865159798220965e-05,
+      "loss": 2.8769,
+      "step": 23630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.864323317130073e-05,
+      "loss": 2.775,
+      "step": 23635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.86348671669551e-05,
+      "loss": 2.5288,
+      "step": 23640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.862649996952133e-05,
+      "loss": 2.7653,
+      "step": 23645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.861813157934804e-05,
+      "loss": 2.7936,
+      "step": 23650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.860976199678391e-05,
+      "loss": 2.8085,
+      "step": 23655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.860139122217769e-05,
+      "loss": 2.8846,
+      "step": 23660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.85930192558781e-05,
+      "loss": 2.8208,
+      "step": 23665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.858464609823399e-05,
+      "loss": 2.801,
+      "step": 23670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.857627174959425e-05,
+      "loss": 2.855,
+      "step": 23675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.856789621030778e-05,
+      "loss": 2.8141,
+      "step": 23680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.855951948072355e-05,
+      "loss": 2.6488,
+      "step": 23685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.85511415611906e-05,
+      "loss": 2.7511,
+      "step": 23690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.854276245205799e-05,
+      "loss": 2.7501,
+      "step": 23695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.853438215367481e-05,
+      "loss": 2.8122,
+      "step": 23700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.852600066639028e-05,
+      "loss": 2.8371,
+      "step": 23705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.85176179905536e-05,
+      "loss": 2.6846,
+      "step": 23710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.850923412651404e-05,
+      "loss": 2.8161,
+      "step": 23715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.850084907462093e-05,
+      "loss": 2.9318,
+      "step": 23720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.849246283522358e-05,
+      "loss": 2.845,
+      "step": 23725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.848407540867147e-05,
+      "loss": 2.8298,
+      "step": 23730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.847568679531406e-05,
+      "loss": 2.9752,
+      "step": 23735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.846729699550084e-05,
+      "loss": 2.6812,
+      "step": 23740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.845890600958142e-05,
+      "loss": 2.8606,
+      "step": 23745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.845051383790535e-05,
+      "loss": 2.7795,
+      "step": 23750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.844212048082235e-05,
+      "loss": 2.8667,
+      "step": 23755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.84337259386821e-05,
+      "loss": 2.8754,
+      "step": 23760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.84253302118344e-05,
+      "loss": 2.9064,
+      "step": 23765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.841693330062903e-05,
+      "loss": 2.8283,
+      "step": 23770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.840853520541585e-05,
+      "loss": 2.8097,
+      "step": 23775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.84001359265448e-05,
+      "loss": 2.791,
+      "step": 23780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.839173546436583e-05,
+      "loss": 2.6826,
+      "step": 23785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.83833338192289e-05,
+      "loss": 2.7338,
+      "step": 23790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.837493099148416e-05,
+      "loss": 2.8199,
+      "step": 23795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.836652698148166e-05,
+      "loss": 2.6777,
+      "step": 23800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.835812178957158e-05,
+      "loss": 2.724,
+      "step": 23805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.83497154161041e-05,
+      "loss": 2.7725,
+      "step": 23810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.83413078614295e-05,
+      "loss": 2.834,
+      "step": 23815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.833289912589809e-05,
+      "loss": 2.8034,
+      "step": 23820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.83244892098602e-05,
+      "loss": 2.8742,
+      "step": 23825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.831607811366625e-05,
+      "loss": 2.7711,
+      "step": 23830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.830766583766671e-05,
+      "loss": 2.735,
+      "step": 23835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.829925238221206e-05,
+      "loss": 2.8416,
+      "step": 23840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.829083774765285e-05,
+      "loss": 2.767,
+      "step": 23845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.828242193433967e-05,
+      "loss": 2.7938,
+      "step": 23850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.82740049426232e-05,
+      "loss": 2.7794,
+      "step": 23855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.826558677285412e-05,
+      "loss": 2.702,
+      "step": 23860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.825716742538319e-05,
+      "loss": 2.8173,
+      "step": 23865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.82487469005612e-05,
+      "loss": 2.7814,
+      "step": 23870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.824032519873899e-05,
+      "loss": 2.9072,
+      "step": 23875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.823190232026746e-05,
+      "loss": 2.8366,
+      "step": 23880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.822347826549756e-05,
+      "loss": 2.6783,
+      "step": 23885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.821505303478027e-05,
+      "loss": 2.5515,
+      "step": 23890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.820662662846663e-05,
+      "loss": 2.7161,
+      "step": 23895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.819819904690777e-05,
+      "loss": 2.7674,
+      "step": 23900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.818977029045477e-05,
+      "loss": 2.6557,
+      "step": 23905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.818134035945888e-05,
+      "loss": 2.8469,
+      "step": 23910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.817290925427128e-05,
+      "loss": 2.7446,
+      "step": 23915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.81644769752433e-05,
+      "loss": 2.8543,
+      "step": 23920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.815604352272625e-05,
+      "loss": 2.7579,
+      "step": 23925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.814760889707154e-05,
+      "loss": 2.7775,
+      "step": 23930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.813917309863057e-05,
+      "loss": 2.6997,
+      "step": 23935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.813073612775485e-05,
+      "loss": 2.8601,
+      "step": 23940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.812229798479589e-05,
+      "loss": 2.7219,
+      "step": 23945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.811385867010529e-05,
+      "loss": 2.8366,
+      "step": 23950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.810541818403467e-05,
+      "loss": 2.9197,
+      "step": 23955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.809697652693571e-05,
+      "loss": 2.9147,
+      "step": 23960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.808853369916014e-05,
+      "loss": 2.8437,
+      "step": 23965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.808008970105972e-05,
+      "loss": 2.7988,
+      "step": 23970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.80716445329863e-05,
+      "loss": 2.7214,
+      "step": 23975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.806319819529171e-05,
+      "loss": 2.7327,
+      "step": 23980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.805475068832791e-05,
+      "loss": 2.8539,
+      "step": 23985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.804630201244688e-05,
+      "loss": 2.6925,
+      "step": 23990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.803785216800059e-05,
+      "loss": 2.7767,
+      "step": 23995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.802940115534114e-05,
+      "loss": 2.94,
+      "step": 24000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.802094897482066e-05,
+      "loss": 2.819,
+      "step": 24005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.801249562679127e-05,
+      "loss": 2.7535,
+      "step": 24010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.800404111160522e-05,
+      "loss": 2.835,
+      "step": 24015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.799558542961477e-05,
+      "loss": 2.6844,
+      "step": 24020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.798712858117221e-05,
+      "loss": 2.7856,
+      "step": 24025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.797867056662992e-05,
+      "loss": 2.667,
+      "step": 24030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.79702113863403e-05,
+      "loss": 2.7081,
+      "step": 24035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.796175104065582e-05,
+      "loss": 2.7378,
+      "step": 24040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.795328952992895e-05,
+      "loss": 2.8526,
+      "step": 24045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.794482685451228e-05,
+      "loss": 2.7131,
+      "step": 24050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.79363630147584e-05,
+      "loss": 2.7594,
+      "step": 24055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.792789801101994e-05,
+      "loss": 2.9779,
+      "step": 24060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.791943184364964e-05,
+      "loss": 2.7984,
+      "step": 24065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.791096451300021e-05,
+      "loss": 2.7362,
+      "step": 24070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.790249601942445e-05,
+      "loss": 2.7492,
+      "step": 24075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.789402636327525e-05,
+      "loss": 2.8872,
+      "step": 24080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.788555554490545e-05,
+      "loss": 2.7784,
+      "step": 24085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.7877083564668e-05,
+      "loss": 2.7403,
+      "step": 24090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.786861042291593e-05,
+      "loss": 2.8475,
+      "step": 24095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.786013612000222e-05,
+      "loss": 2.7871,
+      "step": 24100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.785166065628002e-05,
+      "loss": 2.766,
+      "step": 24105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.78431840321024e-05,
+      "loss": 2.6778,
+      "step": 24110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.783470624782258e-05,
+      "loss": 2.9106,
+      "step": 24115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.782622730379381e-05,
+      "loss": 2.8928,
+      "step": 24120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.781774720036932e-05,
+      "loss": 2.7576,
+      "step": 24125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.780926593790248e-05,
+      "loss": 2.7332,
+      "step": 24130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.780078351674665e-05,
+      "loss": 2.8918,
+      "step": 24135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.779229993725523e-05,
+      "loss": 2.9114,
+      "step": 24140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.778381519978175e-05,
+      "loss": 2.6788,
+      "step": 24145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.77753293046797e-05,
+      "loss": 2.7674,
+      "step": 24150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.776684225230262e-05,
+      "loss": 2.8404,
+      "step": 24155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.775835404300418e-05,
+      "loss": 2.8974,
+      "step": 24160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.774986467713803e-05,
+      "loss": 2.8108,
+      "step": 24165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.774137415505786e-05,
+      "loss": 2.7645,
+      "step": 24170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.773288247711747e-05,
+      "loss": 2.7727,
+      "step": 24175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.772438964367066e-05,
+      "loss": 2.7484,
+      "step": 24180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.771589565507125e-05,
+      "loss": 2.6837,
+      "step": 24185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.77074005116732e-05,
+      "loss": 2.7942,
+      "step": 24190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.769890421383044e-05,
+      "loss": 2.8458,
+      "step": 24195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.769040676189698e-05,
+      "loss": 2.7271,
+      "step": 24200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.768190815622685e-05,
+      "loss": 2.8759,
+      "step": 24205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.767340839717418e-05,
+      "loss": 2.7624,
+      "step": 24210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.76649074850931e-05,
+      "loss": 2.9926,
+      "step": 24215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.765640542033781e-05,
+      "loss": 2.7524,
+      "step": 24220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.764790220326253e-05,
+      "loss": 2.7334,
+      "step": 24225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.76393978342216e-05,
+      "loss": 2.8164,
+      "step": 24230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.763089231356932e-05,
+      "loss": 2.7269,
+      "step": 24235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.76223856416601e-05,
+      "loss": 2.8174,
+      "step": 24240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.761387781884836e-05,
+      "loss": 2.7261,
+      "step": 24245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.760536884548859e-05,
+      "loss": 2.8309,
+      "step": 24250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.759685872193532e-05,
+      "loss": 2.7887,
+      "step": 24255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.758834744854312e-05,
+      "loss": 2.7326,
+      "step": 24260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.757983502566666e-05,
+      "loss": 2.7924,
+      "step": 24265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.757132145366054e-05,
+      "loss": 2.8326,
+      "step": 24270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.756280673287957e-05,
+      "loss": 2.7667,
+      "step": 24275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.755429086367844e-05,
+      "loss": 2.9315,
+      "step": 24280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.754577384641199e-05,
+      "loss": 2.6966,
+      "step": 24285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.753725568143514e-05,
+      "loss": 2.6448,
+      "step": 24290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.752873636910275e-05,
+      "loss": 2.7272,
+      "step": 24295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.752021590976979e-05,
+      "loss": 2.7318,
+      "step": 24300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.751169430379128e-05,
+      "loss": 2.8771,
+      "step": 24305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.750317155152228e-05,
+      "loss": 2.6728,
+      "step": 24310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.749464765331787e-05,
+      "loss": 2.7774,
+      "step": 24315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.748612260953323e-05,
+      "loss": 2.8407,
+      "step": 24320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.747759642052357e-05,
+      "loss": 2.7943,
+      "step": 24325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.74690690866441e-05,
+      "loss": 2.7163,
+      "step": 24330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.746054060825015e-05,
+      "loss": 2.8158,
+      "step": 24335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.745201098569706e-05,
+      "loss": 2.6555,
+      "step": 24340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.744348021934017e-05,
+      "loss": 2.8247,
+      "step": 24345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.743494830953501e-05,
+      "loss": 2.7927,
+      "step": 24350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.7426415256637e-05,
+      "loss": 2.7883,
+      "step": 24355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.74178810610017e-05,
+      "loss": 2.6104,
+      "step": 24360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.740934572298468e-05,
+      "loss": 2.79,
+      "step": 24365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.74008092429416e-05,
+      "loss": 2.8362,
+      "step": 24370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.739227162122808e-05,
+      "loss": 2.8267,
+      "step": 24375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.738373285819988e-05,
+      "loss": 2.8193,
+      "step": 24380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.73751929542128e-05,
+      "loss": 2.8051,
+      "step": 24385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.73666519096226e-05,
+      "loss": 2.8451,
+      "step": 24390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.73581097247852e-05,
+      "loss": 2.7992,
+      "step": 24395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.734956640005649e-05,
+      "loss": 2.848,
+      "step": 24400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.734102193579242e-05,
+      "loss": 2.7585,
+      "step": 24405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.733247633234902e-05,
+      "loss": 2.7005,
+      "step": 24410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.732392959008235e-05,
+      "loss": 2.842,
+      "step": 24415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.731538170934851e-05,
+      "loss": 2.8379,
+      "step": 24420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.730683269050365e-05,
+      "loss": 2.7006,
+      "step": 24425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.729828253390396e-05,
+      "loss": 2.9238,
+      "step": 24430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.728973123990572e-05,
+      "loss": 2.8192,
+      "step": 24435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.728117880886517e-05,
+      "loss": 2.5711,
+      "step": 24440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.72726252411387e-05,
+      "loss": 2.7922,
+      "step": 24445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.726407053708268e-05,
+      "loss": 2.6268,
+      "step": 24450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.725551469705356e-05,
+      "loss": 2.7167,
+      "step": 24455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.724695772140782e-05,
+      "loss": 2.6868,
+      "step": 24460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.723839961050196e-05,
+      "loss": 2.7804,
+      "step": 24465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.72298403646926e-05,
+      "loss": 2.6775,
+      "step": 24470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.722127998433635e-05,
+      "loss": 2.8942,
+      "step": 24475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.721271846978988e-05,
+      "loss": 2.7497,
+      "step": 24480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.72041558214099e-05,
+      "loss": 2.8258,
+      "step": 24485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.71955920395532e-05,
+      "loss": 2.7562,
+      "step": 24490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.71870271245766e-05,
+      "loss": 2.8015,
+      "step": 24495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.717846107683693e-05,
+      "loss": 2.6506,
+      "step": 24500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.716989389669112e-05,
+      "loss": 2.8153,
+      "step": 24505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.716132558449612e-05,
+      "loss": 2.8321,
+      "step": 24510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.715275614060896e-05,
+      "loss": 2.8047,
+      "step": 24515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.714418556538664e-05,
+      "loss": 2.8062,
+      "step": 24520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.71356138591863e-05,
+      "loss": 2.7805,
+      "step": 24525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.712704102236505e-05,
+      "loss": 2.8112,
+      "step": 24530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.711846705528011e-05,
+      "loss": 2.6958,
+      "step": 24535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.710989195828873e-05,
+      "loss": 2.7536,
+      "step": 24540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.710131573174818e-05,
+      "loss": 2.8109,
+      "step": 24545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.709273837601578e-05,
+      "loss": 2.9217,
+      "step": 24550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.708415989144893e-05,
+      "loss": 2.9011,
+      "step": 24555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.707558027840504e-05,
+      "loss": 2.8865,
+      "step": 24560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.70669995372416e-05,
+      "loss": 2.706,
+      "step": 24565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.705841766831612e-05,
+      "loss": 2.8428,
+      "step": 24570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.704983467198619e-05,
+      "loss": 2.7227,
+      "step": 24575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.70412505486094e-05,
+      "loss": 2.7369,
+      "step": 24580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.703266529854342e-05,
+      "loss": 2.8548,
+      "step": 24585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.702407892214596e-05,
+      "loss": 2.6738,
+      "step": 24590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.701549141977478e-05,
+      "loss": 2.8733,
+      "step": 24595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.700690279178768e-05,
+      "loss": 2.7217,
+      "step": 24600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.699831303854251e-05,
+      "loss": 2.6879,
+      "step": 24605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.698972216039717e-05,
+      "loss": 2.7604,
+      "step": 24610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.69811301577096e-05,
+      "loss": 2.8479,
+      "step": 24615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.69725370308378e-05,
+      "loss": 2.6374,
+      "step": 24620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.696394278013979e-05,
+      "loss": 2.921,
+      "step": 24625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.695534740597368e-05,
+      "loss": 2.7676,
+      "step": 24630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.694675090869756e-05,
+      "loss": 2.7672,
+      "step": 24635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.693815328866967e-05,
+      "loss": 2.6864,
+      "step": 24640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.692955454624818e-05,
+      "loss": 2.6774,
+      "step": 24645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.692095468179137e-05,
+      "loss": 2.8682,
+      "step": 24650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.691235369565757e-05,
+      "loss": 2.653,
+      "step": 24655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.690375158820516e-05,
+      "loss": 2.8163,
+      "step": 24660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.689514835979252e-05,
+      "loss": 2.798,
+      "step": 24665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.688654401077813e-05,
+      "loss": 2.9303,
+      "step": 24670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.687793854152049e-05,
+      "loss": 2.7116,
+      "step": 24675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.686933195237815e-05,
+      "loss": 2.7741,
+      "step": 24680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.686072424370969e-05,
+      "loss": 2.7351,
+      "step": 24685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.685211541587379e-05,
+      "loss": 2.7687,
+      "step": 24690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.684350546922913e-05,
+      "loss": 2.8143,
+      "step": 24695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.683489440413444e-05,
+      "loss": 2.6771,
+      "step": 24700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.682628222094851e-05,
+      "loss": 2.8131,
+      "step": 24705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.681766892003019e-05,
+      "loss": 2.7375,
+      "step": 24710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.68090545017383e-05,
+      "loss": 2.7449,
+      "step": 24715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.680043896643182e-05,
+      "loss": 2.6705,
+      "step": 24720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.679182231446971e-05,
+      "loss": 2.9181,
+      "step": 24725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.678320454621098e-05,
+      "loss": 2.6595,
+      "step": 24730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.67745856620147e-05,
+      "loss": 2.8036,
+      "step": 24735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.676596566223998e-05,
+      "loss": 2.7819,
+      "step": 24740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.675734454724597e-05,
+      "loss": 2.7268,
+      "step": 24745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.674872231739186e-05,
+      "loss": 2.7861,
+      "step": 24750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.674009897303693e-05,
+      "loss": 2.7481,
+      "step": 24755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.673147451454047e-05,
+      "loss": 2.835,
+      "step": 24760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.672284894226183e-05,
+      "loss": 2.769,
+      "step": 24765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.671422225656038e-05,
+      "loss": 2.8634,
+      "step": 24770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.670559445779555e-05,
+      "loss": 2.7395,
+      "step": 24775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.669696554632685e-05,
+      "loss": 2.8879,
+      "step": 24780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.668833552251378e-05,
+      "loss": 2.7535,
+      "step": 24785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.667970438671595e-05,
+      "loss": 2.8472,
+      "step": 24790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.667107213929294e-05,
+      "loss": 2.8393,
+      "step": 24795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.666243878060444e-05,
+      "loss": 2.6944,
+      "step": 24800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.665380431101017e-05,
+      "loss": 2.7415,
+      "step": 24805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.664516873086987e-05,
+      "loss": 2.6937,
+      "step": 24810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.663653204054337e-05,
+      "loss": 2.7471,
+      "step": 24815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.66278942403905e-05,
+      "loss": 2.8789,
+      "step": 24820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.66192553307712e-05,
+      "loss": 2.8415,
+      "step": 24825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.661061531204535e-05,
+      "loss": 2.7381,
+      "step": 24830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.660197418457298e-05,
+      "loss": 2.7768,
+      "step": 24835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.659333194871413e-05,
+      "loss": 2.7281,
+      "step": 24840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.658468860482886e-05,
+      "loss": 2.8031,
+      "step": 24845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.657604415327735e-05,
+      "loss": 2.7721,
+      "step": 24850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.656739859441973e-05,
+      "loss": 2.8813,
+      "step": 24855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.655875192861622e-05,
+      "loss": 2.6798,
+      "step": 24860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.655010415622712e-05,
+      "loss": 2.9166,
+      "step": 24865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.654145527761271e-05,
+      "loss": 2.7286,
+      "step": 24870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.653280529313338e-05,
+      "loss": 2.6494,
+      "step": 24875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.652415420314953e-05,
+      "loss": 2.8299,
+      "step": 24880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.65155020080216e-05,
+      "loss": 2.7481,
+      "step": 24885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.650684870811011e-05,
+      "loss": 2.8624,
+      "step": 24890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.649819430377558e-05,
+      "loss": 2.9485,
+      "step": 24895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.648953879537861e-05,
+      "loss": 2.6643,
+      "step": 24900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.648088218327985e-05,
+      "loss": 2.9251,
+      "step": 24905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.647222446783995e-05,
+      "loss": 2.8019,
+      "step": 24910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.646356564941968e-05,
+      "loss": 2.6385,
+      "step": 24915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.645490572837977e-05,
+      "loss": 2.7358,
+      "step": 24920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.644624470508109e-05,
+      "loss": 2.8262,
+      "step": 24925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.643758257988447e-05,
+      "loss": 2.7306,
+      "step": 24930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.642891935315084e-05,
+      "loss": 2.8041,
+      "step": 24935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.642025502524112e-05,
+      "loss": 2.8773,
+      "step": 24940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.64115895965164e-05,
+      "loss": 2.8089,
+      "step": 24945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.640292306733764e-05,
+      "loss": 2.8408,
+      "step": 24950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.639425543806599e-05,
+      "loss": 2.7445,
+      "step": 24955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.638558670906255e-05,
+      "loss": 2.8967,
+      "step": 24960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.637691688068855e-05,
+      "loss": 2.8468,
+      "step": 24965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.63682459533052e-05,
+      "loss": 2.6245,
+      "step": 24970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.63595739272738e-05,
+      "loss": 2.7984,
+      "step": 24975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.635090080295564e-05,
+      "loss": 2.7955,
+      "step": 24980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.634222658071211e-05,
+      "loss": 2.6354,
+      "step": 24985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.633355126090463e-05,
+      "loss": 2.8244,
+      "step": 24990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.632487484389466e-05,
+      "loss": 2.8063,
+      "step": 24995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.631619733004371e-05,
+      "loss": 2.5742,
+      "step": 25000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.630751871971331e-05,
+      "loss": 2.7767,
+      "step": 25005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.629883901326511e-05,
+      "loss": 2.7101,
+      "step": 25010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.62901582110607e-05,
+      "loss": 2.8054,
+      "step": 25015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.628147631346181e-05,
+      "loss": 2.7445,
+      "step": 25020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.627279332083015e-05,
+      "loss": 2.8662,
+      "step": 25025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.626410923352751e-05,
+      "loss": 2.8079,
+      "step": 25030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.625542405191573e-05,
+      "loss": 2.6666,
+      "step": 25035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.624673777635666e-05,
+      "loss": 2.8056,
+      "step": 25040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.623805040721224e-05,
+      "loss": 2.6498,
+      "step": 25045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.622936194484443e-05,
+      "loss": 2.7337,
+      "step": 25050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.622067238961523e-05,
+      "loss": 2.74,
+      "step": 25055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.621198174188671e-05,
+      "loss": 2.7966,
+      "step": 25060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.620329000202096e-05,
+      "loss": 2.7819,
+      "step": 25065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.619459717038012e-05,
+      "loss": 2.8332,
+      "step": 25070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.61859032473264e-05,
+      "loss": 2.6884,
+      "step": 25075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.6177208233222e-05,
+      "loss": 2.8377,
+      "step": 25080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.616851212842926e-05,
+      "loss": 2.7004,
+      "step": 25085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.615981493331046e-05,
+      "loss": 2.9544,
+      "step": 25090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.6151116648228e-05,
+      "loss": 2.8477,
+      "step": 25095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.61424172735443e-05,
+      "loss": 2.8401,
+      "step": 25100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.61337168096218e-05,
+      "loss": 2.7378,
+      "step": 25105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.612501525682303e-05,
+      "loss": 2.8488,
+      "step": 25110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.611631261551055e-05,
+      "loss": 2.787,
+      "step": 25115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.610760888604694e-05,
+      "loss": 2.6884,
+      "step": 25120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.609890406879486e-05,
+      "loss": 2.713,
+      "step": 25125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.6090198164117e-05,
+      "loss": 2.7358,
+      "step": 25130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.608149117237608e-05,
+      "loss": 2.8507,
+      "step": 25135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.607278309393491e-05,
+      "loss": 2.7559,
+      "step": 25140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.60640739291563e-05,
+      "loss": 2.752,
+      "step": 25145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.605536367840314e-05,
+      "loss": 2.8561,
+      "step": 25150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.604665234203832e-05,
+      "loss": 2.7326,
+      "step": 25155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.603793992042481e-05,
+      "loss": 2.6657,
+      "step": 25160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.602922641392563e-05,
+      "loss": 2.8389,
+      "step": 25165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.602051182290382e-05,
+      "loss": 2.9139,
+      "step": 25170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.60117961477225e-05,
+      "loss": 2.8755,
+      "step": 25175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.600307938874478e-05,
+      "loss": 2.6397,
+      "step": 25180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.599436154633388e-05,
+      "loss": 2.8501,
+      "step": 25185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.598564262085301e-05,
+      "loss": 2.8334,
+      "step": 25190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.597692261266548e-05,
+      "loss": 2.9387,
+      "step": 25195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.59682015221346e-05,
+      "loss": 2.7592,
+      "step": 25200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.595947934962371e-05,
+      "loss": 2.7533,
+      "step": 25205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.595075609549626e-05,
+      "loss": 2.747,
+      "step": 25210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.59420317601157e-05,
+      "loss": 2.6917,
+      "step": 25215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.593330634384553e-05,
+      "loss": 2.7695,
+      "step": 25220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.592457984704931e-05,
+      "loss": 2.7604,
+      "step": 25225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.591585227009061e-05,
+      "loss": 2.6891,
+      "step": 25230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.59071236133331e-05,
+      "loss": 2.86,
+      "step": 25235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.589839387714044e-05,
+      "loss": 2.8047,
+      "step": 25240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.588966306187636e-05,
+      "loss": 2.8404,
+      "step": 25245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.588093116790468e-05,
+      "loss": 2.6769,
+      "step": 25250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.587219819558915e-05,
+      "loss": 2.6914,
+      "step": 25255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.586346414529368e-05,
+      "loss": 2.8006,
+      "step": 25260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.585472901738217e-05,
+      "loss": 2.7631,
+      "step": 25265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.584599281221858e-05,
+      "loss": 2.9117,
+      "step": 25270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.583725553016689e-05,
+      "loss": 2.7042,
+      "step": 25275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.582851717159114e-05,
+      "loss": 2.5929,
+      "step": 25280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.581977773685544e-05,
+      "loss": 2.8205,
+      "step": 25285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.581103722632393e-05,
+      "loss": 3.0011,
+      "step": 25290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.580229564036076e-05,
+      "loss": 2.6781,
+      "step": 25295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.579355297933017e-05,
+      "loss": 2.7211,
+      "step": 25300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.578480924359644e-05,
+      "loss": 2.8121,
+      "step": 25305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.577606443352387e-05,
+      "loss": 2.8114,
+      "step": 25310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.57673185494768e-05,
+      "loss": 2.6578,
+      "step": 25315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.575857159181965e-05,
+      "loss": 2.7133,
+      "step": 25320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.574982356091686e-05,
+      "loss": 2.7906,
+      "step": 25325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.574107445713294e-05,
+      "loss": 2.762,
+      "step": 25330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.573232428083241e-05,
+      "loss": 2.8539,
+      "step": 25335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.572357303237985e-05,
+      "loss": 2.7804,
+      "step": 25340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.57148207121399e-05,
+      "loss": 2.7563,
+      "step": 25345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.570606732047722e-05,
+      "loss": 2.7009,
+      "step": 25350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.569731285775652e-05,
+      "loss": 2.7869,
+      "step": 25355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.568855732434256e-05,
+      "loss": 2.7639,
+      "step": 25360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.567980072060015e-05,
+      "loss": 2.7097,
+      "step": 25365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.567104304689415e-05,
+      "loss": 2.78,
+      "step": 25370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.566228430358945e-05,
+      "loss": 2.7692,
+      "step": 25375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.565352449105096e-05,
+      "loss": 2.8373,
+      "step": 25380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.56447636096437e-05,
+      "loss": 2.7785,
+      "step": 25385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.563600165973266e-05,
+      "loss": 2.6935,
+      "step": 25390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.562723864168294e-05,
+      "loss": 2.8076,
+      "step": 25395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.561847455585965e-05,
+      "loss": 2.8418,
+      "step": 25400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.560970940262795e-05,
+      "loss": 2.8667,
+      "step": 25405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.560094318235304e-05,
+      "loss": 2.7948,
+      "step": 25410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.559217589540018e-05,
+      "loss": 2.7656,
+      "step": 25415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.558340754213464e-05,
+      "loss": 2.8503,
+      "step": 25420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.55746381229218e-05,
+      "loss": 2.7203,
+      "step": 25425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5565867638127e-05,
+      "loss": 2.8949,
+      "step": 25430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.55570960881157e-05,
+      "loss": 2.9392,
+      "step": 25435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.554832347325335e-05,
+      "loss": 2.8183,
+      "step": 25440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.553954979390548e-05,
+      "loss": 2.786,
+      "step": 25445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.553077505043764e-05,
+      "loss": 2.7684,
+      "step": 25450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.552199924321543e-05,
+      "loss": 2.8336,
+      "step": 25455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.551322237260452e-05,
+      "loss": 2.775,
+      "step": 25460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.55044444389706e-05,
+      "loss": 2.8752,
+      "step": 25465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.549566544267939e-05,
+      "loss": 2.7176,
+      "step": 25470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.548688538409666e-05,
+      "loss": 2.782,
+      "step": 25475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.547810426358828e-05,
+      "loss": 2.8199,
+      "step": 25480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.546932208152008e-05,
+      "loss": 2.7952,
+      "step": 25485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.546053883825802e-05,
+      "loss": 2.7515,
+      "step": 25490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5451754534168e-05,
+      "loss": 2.7115,
+      "step": 25495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.544296916961608e-05,
+      "loss": 2.7768,
+      "step": 25500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.543418274496826e-05,
+      "loss": 2.8823,
+      "step": 25505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.542539526059065e-05,
+      "loss": 2.8518,
+      "step": 25510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.54166067168494e-05,
+      "loss": 2.679,
+      "step": 25515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.540781711411067e-05,
+      "loss": 2.769,
+      "step": 25520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.53990264527407e-05,
+      "loss": 2.8219,
+      "step": 25525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.539023473310576e-05,
+      "loss": 2.7618,
+      "step": 25530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.538144195557213e-05,
+      "loss": 2.5466,
+      "step": 25535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.53726481205062e-05,
+      "loss": 2.806,
+      "step": 25540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.536385322827435e-05,
+      "loss": 2.8367,
+      "step": 25545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.535505727924303e-05,
+      "loss": 2.883,
+      "step": 25550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.534626027377873e-05,
+      "loss": 2.7822,
+      "step": 25555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5337462212248e-05,
+      "loss": 2.7703,
+      "step": 25560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.532866309501738e-05,
+      "loss": 2.6444,
+      "step": 25565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.531986292245352e-05,
+      "loss": 2.7056,
+      "step": 25570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.531106169492305e-05,
+      "loss": 2.8024,
+      "step": 25575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.530225941279274e-05,
+      "loss": 2.714,
+      "step": 25580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.529345607642928e-05,
+      "loss": 2.8151,
+      "step": 25585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.528465168619949e-05,
+      "loss": 2.7506,
+      "step": 25590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.527584624247022e-05,
+      "loss": 2.7779,
+      "step": 25595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.526703974560833e-05,
+      "loss": 2.6872,
+      "step": 25600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.525823219598078e-05,
+      "loss": 2.7431,
+      "step": 25605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.52494235939545e-05,
+      "loss": 2.7929,
+      "step": 25610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.524061393989654e-05,
+      "loss": 2.862,
+      "step": 25615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.523180323417396e-05,
+      "loss": 2.7549,
+      "step": 25620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.522299147715384e-05,
+      "loss": 2.7766,
+      "step": 25625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.521417866920332e-05,
+      "loss": 2.857,
+      "step": 25630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.520536481068963e-05,
+      "loss": 2.7376,
+      "step": 25635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.519654990197998e-05,
+      "loss": 2.8319,
+      "step": 25640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.518773394344162e-05,
+      "loss": 2.7364,
+      "step": 25645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.517891693544194e-05,
+      "loss": 2.8566,
+      "step": 25650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.517009887834827e-05,
+      "loss": 2.8182,
+      "step": 25655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5161279772528e-05,
+      "loss": 2.7927,
+      "step": 25660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.51524596183486e-05,
+      "loss": 2.7217,
+      "step": 25665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.514363841617757e-05,
+      "loss": 2.808,
+      "step": 25670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.513481616638246e-05,
+      "loss": 2.7625,
+      "step": 25675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.512599286933084e-05,
+      "loss": 2.8232,
+      "step": 25680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.511716852539033e-05,
+      "loss": 2.8773,
+      "step": 25685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.510834313492863e-05,
+      "loss": 2.7089,
+      "step": 25690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.509951669831341e-05,
+      "loss": 2.8095,
+      "step": 25695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.509068921591246e-05,
+      "loss": 2.8505,
+      "step": 25700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.50818606880936e-05,
+      "loss": 2.7479,
+      "step": 25705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.507303111522463e-05,
+      "loss": 2.5953,
+      "step": 25710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.506420049767347e-05,
+      "loss": 2.7947,
+      "step": 25715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.505536883580807e-05,
+      "loss": 2.8949,
+      "step": 25720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.504653612999634e-05,
+      "loss": 2.7815,
+      "step": 25725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.503770238060635e-05,
+      "loss": 2.86,
+      "step": 25730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.502886758800616e-05,
+      "loss": 2.7829,
+      "step": 25735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.502003175256387e-05,
+      "loss": 2.7695,
+      "step": 25740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.501119487464763e-05,
+      "loss": 2.9037,
+      "step": 25745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.500235695462563e-05,
+      "loss": 2.8543,
+      "step": 25750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.499351799286612e-05,
+      "loss": 2.6736,
+      "step": 25755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.498467798973736e-05,
+      "loss": 2.7177,
+      "step": 25760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.497583694560769e-05,
+      "loss": 2.7648,
+      "step": 25765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.496699486084548e-05,
+      "loss": 2.8076,
+      "step": 25770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.495815173581914e-05,
+      "loss": 2.7768,
+      "step": 25775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.494930757089711e-05,
+      "loss": 2.7493,
+      "step": 25780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.494046236644789e-05,
+      "loss": 2.8326,
+      "step": 25785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.493161612284001e-05,
+      "loss": 2.7303,
+      "step": 25790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.492276884044207e-05,
+      "loss": 2.6663,
+      "step": 25795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.491392051962272e-05,
+      "loss": 2.8501,
+      "step": 25800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.490507116075059e-05,
+      "loss": 2.7639,
+      "step": 25805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.489622076419442e-05,
+      "loss": 2.7979,
+      "step": 25810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.488736933032296e-05,
+      "loss": 2.8568,
+      "step": 25815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4878516859505e-05,
+      "loss": 2.671,
+      "step": 25820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.486966335210937e-05,
+      "loss": 2.7052,
+      "step": 25825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.486080880850499e-05,
+      "loss": 2.7765,
+      "step": 25830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.48519532290608e-05,
+      "loss": 2.7348,
+      "step": 25835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.484309661414572e-05,
+      "loss": 2.7271,
+      "step": 25840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.48342389641288e-05,
+      "loss": 2.6704,
+      "step": 25845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.482538027937909e-05,
+      "loss": 2.7468,
+      "step": 25850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.481652056026571e-05,
+      "loss": 2.9232,
+      "step": 25855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.480765980715777e-05,
+      "loss": 2.8507,
+      "step": 25860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.479879802042449e-05,
+      "loss": 2.6873,
+      "step": 25865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.47899352004351e-05,
+      "loss": 2.6597,
+      "step": 25870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.478107134755886e-05,
+      "loss": 2.7348,
+      "step": 25875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.477220646216508e-05,
+      "loss": 2.7356,
+      "step": 25880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.476334054462316e-05,
+      "loss": 2.8937,
+      "step": 25885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.475447359530244e-05,
+      "loss": 2.6338,
+      "step": 25890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.474560561457244e-05,
+      "loss": 2.7881,
+      "step": 25895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.47367366028026e-05,
+      "loss": 2.8653,
+      "step": 25900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.472786656036245e-05,
+      "loss": 2.8996,
+      "step": 25905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.471899548762159e-05,
+      "loss": 2.8986,
+      "step": 25910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.471012338494963e-05,
+      "loss": 2.8334,
+      "step": 25915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.470125025271624e-05,
+      "loss": 2.8265,
+      "step": 25920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.46923760912911e-05,
+      "loss": 2.9414,
+      "step": 25925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.468350090104398e-05,
+      "loss": 2.6919,
+      "step": 25930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.467462468234467e-05,
+      "loss": 2.8323,
+      "step": 25935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.466574743556298e-05,
+      "loss": 2.7772,
+      "step": 25940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.46568691610688e-05,
+      "loss": 2.7262,
+      "step": 25945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.464798985923205e-05,
+      "loss": 2.7239,
+      "step": 25950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.463910953042268e-05,
+      "loss": 2.695,
+      "step": 25955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.463022817501071e-05,
+      "loss": 2.8163,
+      "step": 25960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.462134579336619e-05,
+      "loss": 2.753,
+      "step": 25965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.46124623858592e-05,
+      "loss": 2.849,
+      "step": 25970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.460357795285985e-05,
+      "loss": 2.8595,
+      "step": 25975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.459469249473836e-05,
+      "loss": 2.775,
+      "step": 25980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.45858060118649e-05,
+      "loss": 2.7151,
+      "step": 25985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.457691850460977e-05,
+      "loss": 2.6569,
+      "step": 25990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.456802997334327e-05,
+      "loss": 2.733,
+      "step": 25995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.455914041843572e-05,
+      "loss": 2.7597,
+      "step": 26000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.455024984025751e-05,
+      "loss": 2.7786,
+      "step": 26005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.45413582391791e-05,
+      "loss": 2.8256,
+      "step": 26010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.453246561557094e-05,
+      "loss": 2.7805,
+      "step": 26015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.452357196980357e-05,
+      "loss": 2.8051,
+      "step": 26020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.451467730224752e-05,
+      "loss": 2.8431,
+      "step": 26025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.450578161327339e-05,
+      "loss": 2.7969,
+      "step": 26030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.449688490325184e-05,
+      "loss": 2.9385,
+      "step": 26035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.448798717255356e-05,
+      "loss": 2.7553,
+      "step": 26040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.447908842154929e-05,
+      "loss": 2.8144,
+      "step": 26045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.447018865060977e-05,
+      "loss": 2.9535,
+      "step": 26050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.446128786010582e-05,
+      "loss": 2.8381,
+      "step": 26055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.445238605040831e-05,
+      "loss": 2.8195,
+      "step": 26060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.444348322188813e-05,
+      "loss": 2.7466,
+      "step": 26065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.443457937491623e-05,
+      "loss": 2.7338,
+      "step": 26070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.442567450986359e-05,
+      "loss": 2.5754,
+      "step": 26075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.441676862710124e-05,
+      "loss": 2.9244,
+      "step": 26080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.440786172700024e-05,
+      "loss": 2.7418,
+      "step": 26085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.439895380993171e-05,
+      "loss": 2.768,
+      "step": 26090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.439004487626679e-05,
+      "loss": 2.7566,
+      "step": 26095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.438113492637669e-05,
+      "loss": 2.7297,
+      "step": 26100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.437222396063263e-05,
+      "loss": 2.7564,
+      "step": 26105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.436331197940591e-05,
+      "loss": 2.6617,
+      "step": 26110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.435439898306786e-05,
+      "loss": 2.8286,
+      "step": 26115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.434548497198983e-05,
+      "loss": 2.8152,
+      "step": 26120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.433656994654321e-05,
+      "loss": 2.7408,
+      "step": 26125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.432765390709948e-05,
+      "loss": 2.9408,
+      "step": 26130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.431873685403012e-05,
+      "loss": 2.6316,
+      "step": 26135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.430981878770666e-05,
+      "loss": 2.7756,
+      "step": 26140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.430089970850068e-05,
+      "loss": 2.8407,
+      "step": 26145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.429197961678381e-05,
+      "loss": 2.8379,
+      "step": 26150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.428305851292769e-05,
+      "loss": 2.7241,
+      "step": 26155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.427413639730403e-05,
+      "loss": 2.7358,
+      "step": 26160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.426521327028459e-05,
+      "loss": 2.6397,
+      "step": 26165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.425628913224114e-05,
+      "loss": 2.6391,
+      "step": 26170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.42473639835455e-05,
+      "loss": 2.7555,
+      "step": 26175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.423843782456957e-05,
+      "loss": 2.7788,
+      "step": 26180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.422951065568526e-05,
+      "loss": 2.7144,
+      "step": 26185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.422058247726449e-05,
+      "loss": 2.7521,
+      "step": 26190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.42116532896793e-05,
+      "loss": 2.7189,
+      "step": 26195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.42027230933017e-05,
+      "loss": 2.6356,
+      "step": 26200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.419379188850379e-05,
+      "loss": 2.632,
+      "step": 26205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.418485967565768e-05,
+      "loss": 2.78,
+      "step": 26210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.417592645513556e-05,
+      "loss": 2.7847,
+      "step": 26215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.41669922273096e-05,
+      "loss": 2.7292,
+      "step": 26220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.41580569925521e-05,
+      "loss": 2.7334,
+      "step": 26225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.41491207512353e-05,
+      "loss": 2.8288,
+      "step": 26230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.414018350373157e-05,
+      "loss": 2.6913,
+      "step": 26235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.413124525041325e-05,
+      "loss": 2.8625,
+      "step": 26240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.41223059916528e-05,
+      "loss": 2.7607,
+      "step": 26245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.411336572782263e-05,
+      "loss": 2.7022,
+      "step": 26250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.41044244592953e-05,
+      "loss": 2.7603,
+      "step": 26255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.409548218644332e-05,
+      "loss": 2.6365,
+      "step": 26260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.408653890963928e-05,
+      "loss": 2.693,
+      "step": 26265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.407759462925579e-05,
+      "loss": 2.712,
+      "step": 26270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.406864934566553e-05,
+      "loss": 2.779,
+      "step": 26275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.405970305924123e-05,
+      "loss": 2.8025,
+      "step": 26280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.405075577035561e-05,
+      "loss": 2.7191,
+      "step": 26285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.40418074793815e-05,
+      "loss": 2.7412,
+      "step": 26290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.40328581866917e-05,
+      "loss": 2.7881,
+      "step": 26295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.402390789265909e-05,
+      "loss": 2.7375,
+      "step": 26300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.401495659765662e-05,
+      "loss": 2.7851,
+      "step": 26305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.400600430205723e-05,
+      "loss": 2.6453,
+      "step": 26310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.399705100623393e-05,
+      "loss": 2.9234,
+      "step": 26315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.398809671055975e-05,
+      "loss": 2.7886,
+      "step": 26320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.397914141540779e-05,
+      "loss": 2.7249,
+      "step": 26325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.397018512115118e-05,
+      "loss": 2.8564,
+      "step": 26330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.396122782816307e-05,
+      "loss": 2.7898,
+      "step": 26335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.395226953681667e-05,
+      "loss": 2.8326,
+      "step": 26340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.394331024748526e-05,
+      "loss": 2.6732,
+      "step": 26345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.393434996054213e-05,
+      "loss": 2.7284,
+      "step": 26350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.392538867636058e-05,
+      "loss": 2.7681,
+      "step": 26355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.391642639531404e-05,
+      "loss": 2.7887,
+      "step": 26360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.390746311777588e-05,
+      "loss": 2.8181,
+      "step": 26365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.389849884411958e-05,
+      "loss": 2.8819,
+      "step": 26370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.388953357471864e-05,
+      "loss": 2.8716,
+      "step": 26375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.388056730994661e-05,
+      "loss": 2.7438,
+      "step": 26380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.387160005017708e-05,
+      "loss": 2.8013,
+      "step": 26385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.386263179578365e-05,
+      "loss": 2.8466,
+      "step": 26390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.385366254714001e-05,
+      "loss": 2.7958,
+      "step": 26395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.384469230461986e-05,
+      "loss": 2.7788,
+      "step": 26400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.383572106859694e-05,
+      "loss": 2.8446,
+      "step": 26405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.382674883944506e-05,
+      "loss": 2.8683,
+      "step": 26410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.381777561753805e-05,
+      "loss": 2.8726,
+      "step": 26415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.38088014032498e-05,
+      "loss": 2.7699,
+      "step": 26420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.379982619695417e-05,
+      "loss": 2.8358,
+      "step": 26425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.379084999902517e-05,
+      "loss": 2.7397,
+      "step": 26430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.378187280983677e-05,
+      "loss": 2.7716,
+      "step": 26435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.377289462976305e-05,
+      "loss": 2.7691,
+      "step": 26440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.376391545917805e-05,
+      "loss": 2.6498,
+      "step": 26445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.375493529845591e-05,
+      "loss": 2.7608,
+      "step": 26450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.374595414797079e-05,
+      "loss": 2.7188,
+      "step": 26455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.373697200809687e-05,
+      "loss": 2.722,
+      "step": 26460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.372798887920843e-05,
+      "loss": 2.6892,
+      "step": 26465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.371900476167976e-05,
+      "loss": 2.7635,
+      "step": 26470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.371001965588517e-05,
+      "loss": 2.8382,
+      "step": 26475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.370103356219903e-05,
+      "loss": 2.6623,
+      "step": 26480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.369204648099575e-05,
+      "loss": 2.7472,
+      "step": 26485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.368305841264979e-05,
+      "loss": 2.812,
+      "step": 26490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.367406935753565e-05,
+      "loss": 2.8461,
+      "step": 26495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.366507931602785e-05,
+      "loss": 2.7232,
+      "step": 26500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.365608828850097e-05,
+      "loss": 2.8057,
+      "step": 26505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.364709627532964e-05,
+      "loss": 2.7816,
+      "step": 26510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.363810327688848e-05,
+      "loss": 2.803,
+      "step": 26515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.362910929355222e-05,
+      "loss": 2.8503,
+      "step": 26520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.362011432569559e-05,
+      "loss": 2.6253,
+      "step": 26525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.361111837369337e-05,
+      "loss": 2.8607,
+      "step": 26530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.360212143792039e-05,
+      "loss": 2.8271,
+      "step": 26535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.35931235187515e-05,
+      "loss": 2.6854,
+      "step": 26540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.358412461656162e-05,
+      "loss": 2.706,
+      "step": 26545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.357512473172568e-05,
+      "loss": 2.8328,
+      "step": 26550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.356612386461867e-05,
+      "loss": 2.7736,
+      "step": 26555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.35571220156156e-05,
+      "loss": 2.782,
+      "step": 26560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.354811918509157e-05,
+      "loss": 2.5839,
+      "step": 26565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.353911537342166e-05,
+      "loss": 2.8196,
+      "step": 26570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.353011058098104e-05,
+      "loss": 2.8291,
+      "step": 26575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.352110480814489e-05,
+      "loss": 2.7801,
+      "step": 26580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.351209805528843e-05,
+      "loss": 2.8811,
+      "step": 26585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.350309032278693e-05,
+      "loss": 2.7996,
+      "step": 26590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.349408161101573e-05,
+      "loss": 2.6825,
+      "step": 26595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.348507192035017e-05,
+      "loss": 2.7055,
+      "step": 26600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.347606125116564e-05,
+      "loss": 2.8112,
+      "step": 26605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.346704960383756e-05,
+      "loss": 2.6774,
+      "step": 26610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.345803697874144e-05,
+      "loss": 2.7688,
+      "step": 26615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.344902337625276e-05,
+      "loss": 2.7988,
+      "step": 26620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.34400087967471e-05,
+      "loss": 2.8912,
+      "step": 26625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.343099324060007e-05,
+      "loss": 2.8544,
+      "step": 26630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.342197670818728e-05,
+      "loss": 2.6965,
+      "step": 26635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.341295919988442e-05,
+      "loss": 2.758,
+      "step": 26640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.34039407160672e-05,
+      "loss": 2.7405,
+      "step": 26645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.339492125711139e-05,
+      "loss": 2.7312,
+      "step": 26650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.338590082339278e-05,
+      "loss": 2.7317,
+      "step": 26655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.337687941528725e-05,
+      "loss": 2.8095,
+      "step": 26660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.336785703317065e-05,
+      "loss": 2.7945,
+      "step": 26665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.335883367741889e-05,
+      "loss": 2.7562,
+      "step": 26670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.334980934840796e-05,
+      "loss": 2.7767,
+      "step": 26675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.334078404651383e-05,
+      "loss": 2.6709,
+      "step": 26680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.333175777211259e-05,
+      "loss": 2.7912,
+      "step": 26685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.33227305255803e-05,
+      "loss": 2.895,
+      "step": 26690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.331370230729309e-05,
+      "loss": 2.8146,
+      "step": 26695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.330467311762712e-05,
+      "loss": 2.7404,
+      "step": 26700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.329564295695861e-05,
+      "loss": 2.6868,
+      "step": 26705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.328661182566378e-05,
+      "loss": 2.7687,
+      "step": 26710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.327757972411894e-05,
+      "loss": 2.6496,
+      "step": 26715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.326854665270042e-05,
+      "loss": 2.5771,
+      "step": 26720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.325951261178457e-05,
+      "loss": 2.7889,
+      "step": 26725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.325047760174783e-05,
+      "loss": 2.663,
+      "step": 26730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.32414416229666e-05,
+      "loss": 2.8366,
+      "step": 26735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.32324046758174e-05,
+      "loss": 2.6239,
+      "step": 26740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.322336676067679e-05,
+      "loss": 2.6986,
+      "step": 26745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.321432787792126e-05,
+      "loss": 2.8291,
+      "step": 26750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.320528802792749e-05,
+      "loss": 2.675,
+      "step": 26755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.319624721107211e-05,
+      "loss": 2.5611,
+      "step": 26760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.318720542773181e-05,
+      "loss": 2.8293,
+      "step": 26765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.31781626782833e-05,
+      "loss": 2.9038,
+      "step": 26770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.316911896310338e-05,
+      "loss": 2.7102,
+      "step": 26775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.316007428256886e-05,
+      "loss": 2.7714,
+      "step": 26780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.315102863705657e-05,
+      "loss": 2.756,
+      "step": 26785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.314198202694343e-05,
+      "loss": 2.9062,
+      "step": 26790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.313293445260635e-05,
+      "loss": 2.8269,
+      "step": 26795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.312388591442232e-05,
+      "loss": 2.6758,
+      "step": 26800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.311483641276833e-05,
+      "loss": 2.8174,
+      "step": 26805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.310578594802144e-05,
+      "loss": 2.8965,
+      "step": 26810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.309673452055876e-05,
+      "loss": 2.7123,
+      "step": 26815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.308768213075741e-05,
+      "loss": 2.7188,
+      "step": 26820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.307862877899456e-05,
+      "loss": 2.6421,
+      "step": 26825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.306957446564744e-05,
+      "loss": 2.8256,
+      "step": 26830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.306051919109327e-05,
+      "loss": 2.78,
+      "step": 26835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.305146295570939e-05,
+      "loss": 2.6718,
+      "step": 26840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.304240575987307e-05,
+      "loss": 2.7133,
+      "step": 26845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.303334760396174e-05,
+      "loss": 2.7542,
+      "step": 26850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.30242884883528e-05,
+      "loss": 2.7163,
+      "step": 26855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.301522841342369e-05,
+      "loss": 2.7303,
+      "step": 26860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.300616737955192e-05,
+      "loss": 2.7152,
+      "step": 26865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.299710538711499e-05,
+      "loss": 2.8739,
+      "step": 26870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.29880424364905e-05,
+      "loss": 2.9041,
+      "step": 26875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.297897852805609e-05,
+      "loss": 2.8618,
+      "step": 26880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.296991366218935e-05,
+      "loss": 2.8433,
+      "step": 26885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.296084783926802e-05,
+      "loss": 2.6988,
+      "step": 26890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.295178105966981e-05,
+      "loss": 2.7498,
+      "step": 26895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.294271332377251e-05,
+      "loss": 2.8191,
+      "step": 26900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.29336446319539e-05,
+      "loss": 2.764,
+      "step": 26905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.292457498459188e-05,
+      "loss": 2.7736,
+      "step": 26910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.29155043820643e-05,
+      "loss": 2.6839,
+      "step": 26915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.290643282474911e-05,
+      "loss": 2.8465,
+      "step": 26920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.28973603130243e-05,
+      "loss": 2.7205,
+      "step": 26925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.288828684726783e-05,
+      "loss": 2.7032,
+      "step": 26930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.28792124278578e-05,
+      "loss": 2.8107,
+      "step": 26935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.287013705517229e-05,
+      "loss": 2.7421,
+      "step": 26940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.286106072958941e-05,
+      "loss": 2.6907,
+      "step": 26945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.285198345148736e-05,
+      "loss": 2.6676,
+      "step": 26950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.284290522124432e-05,
+      "loss": 2.8036,
+      "step": 26955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.283382603923856e-05,
+      "loss": 2.8134,
+      "step": 26960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.282474590584838e-05,
+      "loss": 2.6408,
+      "step": 26965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.281566482145205e-05,
+      "loss": 2.7838,
+      "step": 26970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.280658278642802e-05,
+      "loss": 2.8874,
+      "step": 26975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.279749980115466e-05,
+      "loss": 2.7084,
+      "step": 26980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.278841586601039e-05,
+      "loss": 2.8335,
+      "step": 26985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.277933098137374e-05,
+      "loss": 2.765,
+      "step": 26990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.277024514762323e-05,
+      "loss": 2.6136,
+      "step": 26995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.27611583651374e-05,
+      "loss": 2.6928,
+      "step": 27000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.275207063429488e-05,
+      "loss": 2.8458,
+      "step": 27005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.27429819554743e-05,
+      "loss": 2.8508,
+      "step": 27010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.27338923290544e-05,
+      "loss": 2.7878,
+      "step": 27015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.27248017554138e-05,
+      "loss": 2.707,
+      "step": 27020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.271571023493134e-05,
+      "loss": 2.9235,
+      "step": 27025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.270661776798582e-05,
+      "loss": 2.8011,
+      "step": 27030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.269752435495606e-05,
+      "loss": 2.8545,
+      "step": 27035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.268842999622095e-05,
+      "loss": 2.6596,
+      "step": 27040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.267933469215942e-05,
+      "loss": 2.8219,
+      "step": 27045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.267023844315042e-05,
+      "loss": 2.75,
+      "step": 27050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.266114124957295e-05,
+      "loss": 2.7957,
+      "step": 27055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.265204311180607e-05,
+      "loss": 2.6995,
+      "step": 27060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.264294403022883e-05,
+      "loss": 2.7861,
+      "step": 27065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.263384400522035e-05,
+      "loss": 2.6653,
+      "step": 27070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.262474303715981e-05,
+      "loss": 2.6667,
+      "step": 27075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.26156411264264e-05,
+      "loss": 2.7903,
+      "step": 27080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.260653827339937e-05,
+      "loss": 2.7075,
+      "step": 27085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.259743447845795e-05,
+      "loss": 2.7915,
+      "step": 27090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.258832974198149e-05,
+      "loss": 2.7465,
+      "step": 27095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.257922406434933e-05,
+      "loss": 2.801,
+      "step": 27100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.257011744594089e-05,
+      "loss": 2.7107,
+      "step": 27105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.256100988713557e-05,
+      "loss": 2.7785,
+      "step": 27110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.255190138831286e-05,
+      "loss": 2.8343,
+      "step": 27115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.254279194985228e-05,
+      "loss": 2.7097,
+      "step": 27120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.253368157213333e-05,
+      "loss": 2.596,
+      "step": 27125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.252457025553566e-05,
+      "loss": 2.85,
+      "step": 27130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.251545800043888e-05,
+      "loss": 2.7365,
+      "step": 27135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.250634480722262e-05,
+      "loss": 2.6664,
+      "step": 27140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.249723067626664e-05,
+      "loss": 2.6525,
+      "step": 27145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.248811560795066e-05,
+      "loss": 2.7332,
+      "step": 27150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.247899960265447e-05,
+      "loss": 2.6947,
+      "step": 27155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.246988266075787e-05,
+      "loss": 2.7752,
+      "step": 27160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.246076478264076e-05,
+      "loss": 2.7956,
+      "step": 27165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.245164596868302e-05,
+      "loss": 2.7179,
+      "step": 27170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.24425262192646e-05,
+      "loss": 2.6611,
+      "step": 27175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.243340553476547e-05,
+      "loss": 2.7389,
+      "step": 27180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.242428391556566e-05,
+      "loss": 2.716,
+      "step": 27185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.24151613620452e-05,
+      "loss": 2.8422,
+      "step": 27190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.240603787458422e-05,
+      "loss": 2.9021,
+      "step": 27195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.239691345356283e-05,
+      "loss": 2.8423,
+      "step": 27200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.238778809936124e-05,
+      "loss": 2.6374,
+      "step": 27205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.237866181235963e-05,
+      "loss": 2.7339,
+      "step": 27210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.236953459293827e-05,
+      "loss": 2.7383,
+      "step": 27215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.236040644147743e-05,
+      "loss": 2.6845,
+      "step": 27220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.235127735835744e-05,
+      "loss": 2.724,
+      "step": 27225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.234214734395871e-05,
+      "loss": 2.8992,
+      "step": 27230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.233301639866159e-05,
+      "loss": 2.7396,
+      "step": 27235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.232388452284657e-05,
+      "loss": 2.7964,
+      "step": 27240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.231475171689412e-05,
+      "loss": 2.7305,
+      "step": 27245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.230561798118476e-05,
+      "loss": 2.8003,
+      "step": 27250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.229648331609906e-05,
+      "loss": 2.767,
+      "step": 27255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.228734772201761e-05,
+      "loss": 2.6391,
+      "step": 27260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.227821119932104e-05,
+      "loss": 2.7798,
+      "step": 27265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.226907374839007e-05,
+      "loss": 2.7949,
+      "step": 27270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.225993536960538e-05,
+      "loss": 2.8669,
+      "step": 27275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.225079606334774e-05,
+      "loss": 2.6877,
+      "step": 27280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.224165582999794e-05,
+      "loss": 2.7288,
+      "step": 27285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.223251466993682e-05,
+      "loss": 2.6852,
+      "step": 27290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.222337258354521e-05,
+      "loss": 2.7608,
+      "step": 27295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.22142295712041e-05,
+      "loss": 2.6436,
+      "step": 27300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.220508563329438e-05,
+      "loss": 2.8643,
+      "step": 27305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.219594077019705e-05,
+      "loss": 2.8707,
+      "step": 27310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.218679498229313e-05,
+      "loss": 2.7504,
+      "step": 27315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.217764826996372e-05,
+      "loss": 2.7075,
+      "step": 27320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.216850063358986e-05,
+      "loss": 2.7996,
+      "step": 27325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.215935207355275e-05,
+      "loss": 2.8031,
+      "step": 27330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.215020259023356e-05,
+      "loss": 2.6743,
+      "step": 27335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.214105218401348e-05,
+      "loss": 2.8723,
+      "step": 27340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.213190085527378e-05,
+      "loss": 2.7507,
+      "step": 27345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.212274860439576e-05,
+      "loss": 2.766,
+      "step": 27350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.211359543176075e-05,
+      "loss": 2.8151,
+      "step": 27355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.210444133775014e-05,
+      "loss": 2.5903,
+      "step": 27360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.209528632274532e-05,
+      "loss": 2.6135,
+      "step": 27365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.208613038712773e-05,
+      "loss": 2.6795,
+      "step": 27370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.207697353127889e-05,
+      "loss": 2.6755,
+      "step": 27375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.206781575558029e-05,
+      "loss": 2.7091,
+      "step": 27380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.205865706041349e-05,
+      "loss": 2.6787,
+      "step": 27385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.204949744616014e-05,
+      "loss": 2.846,
+      "step": 27390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.204033691320185e-05,
+      "loss": 2.6787,
+      "step": 27395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.203117546192032e-05,
+      "loss": 2.7772,
+      "step": 27400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.202201309269722e-05,
+      "loss": 2.6179,
+      "step": 27405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.201284980591435e-05,
+      "loss": 2.7239,
+      "step": 27410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.200368560195349e-05,
+      "loss": 2.7563,
+      "step": 27415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.199452048119646e-05,
+      "loss": 2.8094,
+      "step": 27420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.198535444402514e-05,
+      "loss": 2.6962,
+      "step": 27425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.197618749082146e-05,
+      "loss": 2.7896,
+      "step": 27430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.196701962196732e-05,
+      "loss": 2.7992,
+      "step": 27435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.195785083784475e-05,
+      "loss": 2.6587,
+      "step": 27440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.194868113883573e-05,
+      "loss": 2.788,
+      "step": 27445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.193951052532235e-05,
+      "loss": 2.7572,
+      "step": 27450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.19303389976867e-05,
+      "loss": 2.8251,
+      "step": 27455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.192116655631094e-05,
+      "loss": 2.7823,
+      "step": 27460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.19119932015772e-05,
+      "loss": 2.7806,
+      "step": 27465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.190281893386773e-05,
+      "loss": 2.7732,
+      "step": 27470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.189364375356475e-05,
+      "loss": 2.8809,
+      "step": 27475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.188446766105057e-05,
+      "loss": 2.9392,
+      "step": 27480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.187529065670752e-05,
+      "loss": 2.7285,
+      "step": 27485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.186611274091797e-05,
+      "loss": 2.6676,
+      "step": 27490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.185693391406431e-05,
+      "loss": 2.6884,
+      "step": 27495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.184775417652899e-05,
+      "loss": 2.7087,
+      "step": 27500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.183857352869445e-05,
+      "loss": 2.6137,
+      "step": 27505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.182939197094324e-05,
+      "loss": 2.8829,
+      "step": 27510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.182020950365794e-05,
+      "loss": 2.6605,
+      "step": 27515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.181102612722111e-05,
+      "loss": 2.8111,
+      "step": 27520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.18018418420154e-05,
+      "loss": 2.8064,
+      "step": 27525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.179265664842345e-05,
+      "loss": 2.7956,
+      "step": 27530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.178347054682798e-05,
+      "loss": 2.6284,
+      "step": 27535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.177428353761173e-05,
+      "loss": 2.6212,
+      "step": 27540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.17650956211575e-05,
+      "loss": 2.6716,
+      "step": 27545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.175590679784811e-05,
+      "loss": 2.778,
+      "step": 27550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.17467170680664e-05,
+      "loss": 2.719,
+      "step": 27555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.173752643219526e-05,
+      "loss": 2.7554,
+      "step": 27560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.172833489061764e-05,
+      "loss": 2.7345,
+      "step": 27565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.17191424437165e-05,
+      "loss": 2.7719,
+      "step": 27570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.170994909187485e-05,
+      "loss": 2.7058,
+      "step": 27575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.170075483547576e-05,
+      "loss": 2.7899,
+      "step": 27580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.169155967490229e-05,
+      "loss": 2.6021,
+      "step": 27585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.168236361053755e-05,
+      "loss": 2.7888,
+      "step": 27590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.167316664276474e-05,
+      "loss": 2.7032,
+      "step": 27595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.166396877196699e-05,
+      "loss": 2.6927,
+      "step": 27600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.16547699985276e-05,
+      "loss": 2.7762,
+      "step": 27605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.164557032282982e-05,
+      "loss": 2.8292,
+      "step": 27610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.163636974525696e-05,
+      "loss": 2.6327,
+      "step": 27615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.162716826619237e-05,
+      "loss": 2.8225,
+      "step": 27620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.161796588601943e-05,
+      "loss": 2.7614,
+      "step": 27625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.160876260512154e-05,
+      "loss": 2.8729,
+      "step": 27630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.15995584238822e-05,
+      "loss": 2.8065,
+      "step": 27635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.159035334268489e-05,
+      "loss": 2.7436,
+      "step": 27640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.158114736191316e-05,
+      "loss": 2.7462,
+      "step": 27645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.157194048195055e-05,
+      "loss": 2.7313,
+      "step": 27650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.15627327031807e-05,
+      "loss": 2.8695,
+      "step": 27655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.155352402598723e-05,
+      "loss": 2.6842,
+      "step": 27660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.154431445075385e-05,
+      "loss": 2.8473,
+      "step": 27665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.153510397786425e-05,
+      "loss": 2.6349,
+      "step": 27670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.152589260770223e-05,
+      "loss": 2.7943,
+      "step": 27675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.151668034065158e-05,
+      "loss": 2.7555,
+      "step": 27680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.150746717709613e-05,
+      "loss": 2.7958,
+      "step": 27685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.149825311741972e-05,
+      "loss": 2.7778,
+      "step": 27690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.148903816200629e-05,
+      "loss": 2.722,
+      "step": 27695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.147982231123977e-05,
+      "loss": 2.713,
+      "step": 27700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.147060556550417e-05,
+      "loss": 2.8399,
+      "step": 27705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.14613879251835e-05,
+      "loss": 2.7977,
+      "step": 27710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.145216939066182e-05,
+      "loss": 2.8415,
+      "step": 27715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.144294996232319e-05,
+      "loss": 2.7263,
+      "step": 27720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.143372964055179e-05,
+      "loss": 2.7599,
+      "step": 27725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.142450842573176e-05,
+      "loss": 2.8006,
+      "step": 27730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.141528631824732e-05,
+      "loss": 2.8606,
+      "step": 27735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.140606331848273e-05,
+      "loss": 2.7761,
+      "step": 27740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.139683942682225e-05,
+      "loss": 2.6828,
+      "step": 27745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.13876146436502e-05,
+      "loss": 2.7878,
+      "step": 27750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.137838896935093e-05,
+      "loss": 2.9024,
+      "step": 27755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.136916240430885e-05,
+      "loss": 2.74,
+      "step": 27760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.135993494890838e-05,
+      "loss": 2.6638,
+      "step": 27765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.135070660353398e-05,
+      "loss": 2.8586,
+      "step": 27770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.13414773685702e-05,
+      "loss": 2.7143,
+      "step": 27775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.133224724440152e-05,
+      "loss": 2.7619,
+      "step": 27780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.132301623141253e-05,
+      "loss": 2.6814,
+      "step": 27785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.131378432998788e-05,
+      "loss": 2.8258,
+      "step": 27790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.130455154051219e-05,
+      "loss": 2.7972,
+      "step": 27795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.129531786337017e-05,
+      "loss": 2.6842,
+      "step": 27800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.128608329894654e-05,
+      "loss": 2.6833,
+      "step": 27805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.127684784762606e-05,
+      "loss": 2.7652,
+      "step": 27810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.126761150979352e-05,
+      "loss": 2.7577,
+      "step": 27815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.125837428583377e-05,
+      "loss": 2.7576,
+      "step": 27820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.124913617613169e-05,
+      "loss": 2.7899,
+      "step": 27825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.123989718107219e-05,
+      "loss": 2.7848,
+      "step": 27830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.123065730104021e-05,
+      "loss": 2.6545,
+      "step": 27835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.122141653642073e-05,
+      "loss": 2.632,
+      "step": 27840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.121217488759877e-05,
+      "loss": 2.7948,
+      "step": 27845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.120293235495942e-05,
+      "loss": 2.6523,
+      "step": 27850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.119368893888773e-05,
+      "loss": 2.8207,
+      "step": 27855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.118444463976887e-05,
+      "loss": 2.8431,
+      "step": 27860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.117519945798799e-05,
+      "loss": 2.6978,
+      "step": 27865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.11659533939303e-05,
+      "loss": 2.6956,
+      "step": 27870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.115670644798105e-05,
+      "loss": 2.6604,
+      "step": 27875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.11474586205255e-05,
+      "loss": 2.704,
+      "step": 27880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.113820991194898e-05,
+      "loss": 2.7327,
+      "step": 27885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.112896032263683e-05,
+      "loss": 2.6999,
+      "step": 27890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.111970985297446e-05,
+      "loss": 2.7826,
+      "step": 27895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.11104585033473e-05,
+      "loss": 2.8482,
+      "step": 27900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.110120627414079e-05,
+      "loss": 2.8915,
+      "step": 27905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.109195316574042e-05,
+      "loss": 2.7413,
+      "step": 27910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.108269917853177e-05,
+      "loss": 2.7451,
+      "step": 27915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.107344431290036e-05,
+      "loss": 2.7611,
+      "step": 27920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.106418856923185e-05,
+      "loss": 2.7225,
+      "step": 27925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.105493194791186e-05,
+      "loss": 2.6875,
+      "step": 27930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.104567444932605e-05,
+      "loss": 2.7478,
+      "step": 27935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.10364160738602e-05,
+      "loss": 2.812,
+      "step": 27940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.10271568219e-05,
+      "loss": 2.6706,
+      "step": 27945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.101789669383128e-05,
+      "loss": 2.7321,
+      "step": 27950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.100863569003985e-05,
+      "loss": 2.7444,
+      "step": 27955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.09993738109116e-05,
+      "loss": 2.8252,
+      "step": 27960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.099011105683241e-05,
+      "loss": 2.7771,
+      "step": 27965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.098084742818822e-05,
+      "loss": 2.7705,
+      "step": 27970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.097158292536502e-05,
+      "loss": 2.673,
+      "step": 27975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.09623175487488e-05,
+      "loss": 2.8048,
+      "step": 27980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.095305129872563e-05,
+      "loss": 2.9093,
+      "step": 27985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.094378417568156e-05,
+      "loss": 2.6919,
+      "step": 27990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.093451618000273e-05,
+      "loss": 2.7496,
+      "step": 27995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.092524731207532e-05,
+      "loss": 2.7844,
+      "step": 28000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.091597757228547e-05,
+      "loss": 2.7926,
+      "step": 28005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.090670696101946e-05,
+      "loss": 2.7734,
+      "step": 28010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.089743547866351e-05,
+      "loss": 2.8272,
+      "step": 28015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.088816312560398e-05,
+      "loss": 2.7216,
+      "step": 28020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.087888990222715e-05,
+      "loss": 2.7745,
+      "step": 28025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.086961580891942e-05,
+      "loss": 2.7058,
+      "step": 28030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.08603408460672e-05,
+      "loss": 2.7464,
+      "step": 28035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.085106501405693e-05,
+      "loss": 2.7142,
+      "step": 28040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.08417883132751e-05,
+      "loss": 2.7345,
+      "step": 28045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.083251074410821e-05,
+      "loss": 2.8023,
+      "step": 28050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.082323230694286e-05,
+      "loss": 2.7773,
+      "step": 28055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.081395300216559e-05,
+      "loss": 2.8398,
+      "step": 28060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.080467283016307e-05,
+      "loss": 2.5463,
+      "step": 28065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.079539179132194e-05,
+      "loss": 2.7183,
+      "step": 28070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.078610988602887e-05,
+      "loss": 2.7261,
+      "step": 28075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.077682711467066e-05,
+      "loss": 2.7342,
+      "step": 28080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.076754347763405e-05,
+      "loss": 2.7894,
+      "step": 28085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.075825897530586e-05,
+      "loss": 2.8239,
+      "step": 28090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.07489736080729e-05,
+      "loss": 2.7956,
+      "step": 28095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.073968737632208e-05,
+      "loss": 2.8299,
+      "step": 28100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.073040028044034e-05,
+      "loss": 2.7469,
+      "step": 28105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.072111232081456e-05,
+      "loss": 2.718,
+      "step": 28110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.071182349783178e-05,
+      "loss": 2.7317,
+      "step": 28115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.070253381187903e-05,
+      "loss": 2.7046,
+      "step": 28120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.069324326334334e-05,
+      "loss": 2.842,
+      "step": 28125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.068395185261183e-05,
+      "loss": 2.8062,
+      "step": 28130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.067465958007163e-05,
+      "loss": 2.7672,
+      "step": 28135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.066536644610988e-05,
+      "loss": 2.6045,
+      "step": 28140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.06560724511138e-05,
+      "loss": 2.6978,
+      "step": 28145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.064677759547065e-05,
+      "loss": 2.8013,
+      "step": 28150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.06374818795677e-05,
+      "loss": 2.6644,
+      "step": 28155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.062818530379222e-05,
+      "loss": 2.7626,
+      "step": 28160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.061888786853162e-05,
+      "loss": 2.8123,
+      "step": 28165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.060958957417323e-05,
+      "loss": 2.6103,
+      "step": 28170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.060029042110449e-05,
+      "loss": 2.7122,
+      "step": 28175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.059099040971286e-05,
+      "loss": 2.7388,
+      "step": 28180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.058168954038582e-05,
+      "loss": 2.7607,
+      "step": 28185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.05723878135109e-05,
+      "loss": 2.77,
+      "step": 28190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.056308522947568e-05,
+      "loss": 2.717,
+      "step": 28195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.055378178866773e-05,
+      "loss": 2.6265,
+      "step": 28200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.054447749147468e-05,
+      "loss": 2.555,
+      "step": 28205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.053517233828422e-05,
+      "loss": 2.6959,
+      "step": 28210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.052586632948404e-05,
+      "loss": 2.7472,
+      "step": 28215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.05165594654619e-05,
+      "loss": 2.7857,
+      "step": 28220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.050725174660556e-05,
+      "loss": 2.7616,
+      "step": 28225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.049794317330283e-05,
+      "loss": 2.7993,
+      "step": 28230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.048863374594154e-05,
+      "loss": 2.6582,
+      "step": 28235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.047932346490962e-05,
+      "loss": 2.7494,
+      "step": 28240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.047001233059495e-05,
+      "loss": 2.6089,
+      "step": 28245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.04607003433855e-05,
+      "loss": 2.7156,
+      "step": 28250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.045138750366926e-05,
+      "loss": 2.6862,
+      "step": 28255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.044207381183424e-05,
+      "loss": 2.677,
+      "step": 28260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.04327592682685e-05,
+      "loss": 2.8936,
+      "step": 28265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.042344387336016e-05,
+      "loss": 2.6841,
+      "step": 28270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.041412762749732e-05,
+      "loss": 2.8083,
+      "step": 28275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.040481053106819e-05,
+      "loss": 2.7453,
+      "step": 28280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.039549258446092e-05,
+      "loss": 2.6947,
+      "step": 28285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.03861737880638e-05,
+      "loss": 2.8615,
+      "step": 28290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.037685414226507e-05,
+      "loss": 2.6341,
+      "step": 28295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.036753364745304e-05,
+      "loss": 2.7474,
+      "step": 28300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.035821230401605e-05,
+      "loss": 2.7058,
+      "step": 28305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.034889011234247e-05,
+      "loss": 2.6306,
+      "step": 28310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.033956707282077e-05,
+      "loss": 2.7497,
+      "step": 28315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.033024318583934e-05,
+      "loss": 2.7781,
+      "step": 28320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.03209184517867e-05,
+      "loss": 2.8106,
+      "step": 28325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.031159287105134e-05,
+      "loss": 2.8026,
+      "step": 28330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.030226644402185e-05,
+      "loss": 2.7947,
+      "step": 28335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.029293917108678e-05,
+      "loss": 2.6051,
+      "step": 28340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.02836110526348e-05,
+      "loss": 2.7081,
+      "step": 28345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.027428208905456e-05,
+      "loss": 2.8681,
+      "step": 28350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.026495228073473e-05,
+      "loss": 2.8483,
+      "step": 28355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.025562162806404e-05,
+      "loss": 2.6896,
+      "step": 28360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.024629013143131e-05,
+      "loss": 2.7372,
+      "step": 28365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.02369577912253e-05,
+      "loss": 2.831,
+      "step": 28370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.022762460783486e-05,
+      "loss": 2.6469,
+      "step": 28375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.021829058164886e-05,
+      "loss": 2.6885,
+      "step": 28380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.02089557130562e-05,
+      "loss": 2.7137,
+      "step": 28385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.019962000244582e-05,
+      "loss": 2.8421,
+      "step": 28390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.019028345020671e-05,
+      "loss": 2.8378,
+      "step": 28395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.018094605672788e-05,
+      "loss": 2.807,
+      "step": 28400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.017160782239837e-05,
+      "loss": 2.756,
+      "step": 28405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.01622687476073e-05,
+      "loss": 2.749,
+      "step": 28410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.015292883274374e-05,
+      "loss": 2.8674,
+      "step": 28415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.014358807819685e-05,
+      "loss": 2.6352,
+      "step": 28420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.013424648435584e-05,
+      "loss": 2.7116,
+      "step": 28425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.012490405160991e-05,
+      "loss": 2.8587,
+      "step": 28430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.011556078034835e-05,
+      "loss": 2.6909,
+      "step": 28435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.010621667096041e-05,
+      "loss": 2.8342,
+      "step": 28440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.009687172383545e-05,
+      "loss": 2.639,
+      "step": 28445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.008752593936283e-05,
+      "loss": 2.6844,
+      "step": 28450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.007817931793192e-05,
+      "loss": 2.7384,
+      "step": 28455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.006883185993218e-05,
+      "loss": 2.7709,
+      "step": 28460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.005948356575309e-05,
+      "loss": 2.9668,
+      "step": 28465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.00501344357841e-05,
+      "loss": 2.7412,
+      "step": 28470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.00407844704148e-05,
+      "loss": 2.6797,
+      "step": 28475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.003143367003473e-05,
+      "loss": 2.7059,
+      "step": 28480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.002208203503351e-05,
+      "loss": 2.7928,
+      "step": 28485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.001272956580078e-05,
+      "loss": 2.8005,
+      "step": 28490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.000337626272619e-05,
+      "loss": 2.7831,
+      "step": 28495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.99940221261995e-05,
+      "loss": 2.704,
+      "step": 28500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.998466715661043e-05,
+      "loss": 2.8076,
+      "step": 28505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.997531135434876e-05,
+      "loss": 2.7428,
+      "step": 28510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.99659547198043e-05,
+      "loss": 2.7834,
+      "step": 28515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.995659725336689e-05,
+      "loss": 2.7959,
+      "step": 28520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.994723895542644e-05,
+      "loss": 2.8257,
+      "step": 28525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.993787982637285e-05,
+      "loss": 2.7902,
+      "step": 28530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.99285198665961e-05,
+      "loss": 2.7015,
+      "step": 28535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.991915907648613e-05,
+      "loss": 2.7954,
+      "step": 28540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.990979745643303e-05,
+      "loss": 2.8679,
+      "step": 28545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.990043500682678e-05,
+      "loss": 2.8184,
+      "step": 28550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.989107172805752e-05,
+      "loss": 2.6683,
+      "step": 28555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.988170762051538e-05,
+      "loss": 2.719,
+      "step": 28560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.987234268459051e-05,
+      "loss": 2.8535,
+      "step": 28565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.986297692067309e-05,
+      "loss": 2.6005,
+      "step": 28570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.985361032915337e-05,
+      "loss": 2.7035,
+      "step": 28575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.98442429104216e-05,
+      "loss": 2.8551,
+      "step": 28580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.983487466486809e-05,
+      "loss": 2.8388,
+      "step": 28585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.982550559288317e-05,
+      "loss": 2.7672,
+      "step": 28590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.98161356948572e-05,
+      "loss": 2.7604,
+      "step": 28595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.980676497118061e-05,
+      "loss": 2.6479,
+      "step": 28600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.979739342224381e-05,
+      "loss": 2.8391,
+      "step": 28605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.978802104843727e-05,
+      "loss": 2.7393,
+      "step": 28610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.97786478501515e-05,
+      "loss": 2.7837,
+      "step": 28615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.976927382777705e-05,
+      "loss": 2.7716,
+      "step": 28620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.975989898170449e-05,
+      "loss": 2.6532,
+      "step": 28625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.975052331232441e-05,
+      "loss": 2.7357,
+      "step": 28630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.974114682002748e-05,
+      "loss": 2.7422,
+      "step": 28635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.973176950520436e-05,
+      "loss": 2.7907,
+      "step": 28640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.972239136824575e-05,
+      "loss": 2.8385,
+      "step": 28645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.971301240954242e-05,
+      "loss": 2.8222,
+      "step": 28650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.970363262948515e-05,
+      "loss": 2.7172,
+      "step": 28655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.969425202846475e-05,
+      "loss": 2.7554,
+      "step": 28660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.968487060687205e-05,
+      "loss": 2.7465,
+      "step": 28665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.967548836509795e-05,
+      "loss": 2.9342,
+      "step": 28670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.966610530353334e-05,
+      "loss": 2.7577,
+      "step": 28675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.965672142256921e-05,
+      "loss": 2.6581,
+      "step": 28680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.964733672259653e-05,
+      "loss": 2.8153,
+      "step": 28685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.963795120400632e-05,
+      "loss": 2.8534,
+      "step": 28690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.962856486718962e-05,
+      "loss": 2.8104,
+      "step": 28695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.961917771253751e-05,
+      "loss": 2.8286,
+      "step": 28700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.960978974044115e-05,
+      "loss": 2.7709,
+      "step": 28705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.960040095129167e-05,
+      "loss": 2.7379,
+      "step": 28710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.959101134548025e-05,
+      "loss": 2.7763,
+      "step": 28715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.958162092339813e-05,
+      "loss": 2.6611,
+      "step": 28720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.957222968543657e-05,
+      "loss": 2.5962,
+      "step": 28725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.956283763198684e-05,
+      "loss": 2.8435,
+      "step": 28730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.955344476344029e-05,
+      "loss": 2.6195,
+      "step": 28735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.954405108018825e-05,
+      "loss": 2.6951,
+      "step": 28740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.953465658262214e-05,
+      "loss": 2.8839,
+      "step": 28745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.952526127113339e-05,
+      "loss": 2.8212,
+      "step": 28750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.951586514611344e-05,
+      "loss": 2.8015,
+      "step": 28755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.95064682079538e-05,
+      "loss": 2.8612,
+      "step": 28760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.949707045704598e-05,
+      "loss": 2.7113,
+      "step": 28765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.948767189378155e-05,
+      "loss": 2.8489,
+      "step": 28770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.947827251855212e-05,
+      "loss": 2.6638,
+      "step": 28775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.946887233174931e-05,
+      "loss": 2.8576,
+      "step": 28780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.94594713337648e-05,
+      "loss": 2.745,
+      "step": 28785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.945006952499026e-05,
+      "loss": 2.7489,
+      "step": 28790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.944066690581745e-05,
+      "loss": 2.7918,
+      "step": 28795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.943126347663811e-05,
+      "loss": 2.6345,
+      "step": 28800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.942185923784405e-05,
+      "loss": 2.6942,
+      "step": 28805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.941245418982711e-05,
+      "loss": 2.6116,
+      "step": 28810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.940304833297917e-05,
+      "loss": 2.8728,
+      "step": 28815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.939364166769208e-05,
+      "loss": 2.8484,
+      "step": 28820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.938423419435782e-05,
+      "loss": 2.6934,
+      "step": 28825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.937482591336834e-05,
+      "loss": 2.7044,
+      "step": 28830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.936541682511564e-05,
+      "loss": 2.7725,
+      "step": 28835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.935600692999176e-05,
+      "loss": 2.7007,
+      "step": 28840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.934659622838877e-05,
+      "loss": 2.7444,
+      "step": 28845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.933718472069877e-05,
+      "loss": 2.807,
+      "step": 28850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.932777240731391e-05,
+      "loss": 2.8114,
+      "step": 28855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.931835928862632e-05,
+      "loss": 2.864,
+      "step": 28860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.930894536502824e-05,
+      "loss": 2.8171,
+      "step": 28865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.92995306369119e-05,
+      "loss": 2.8489,
+      "step": 28870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.929011510466957e-05,
+      "loss": 2.8351,
+      "step": 28875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.928069876869354e-05,
+      "loss": 2.7491,
+      "step": 28880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.927128162937615e-05,
+      "loss": 2.7635,
+      "step": 28885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.926186368710979e-05,
+      "loss": 2.7648,
+      "step": 28890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.925244494228684e-05,
+      "loss": 2.7226,
+      "step": 28895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.924302539529974e-05,
+      "loss": 2.7384,
+      "step": 28900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.923360504654098e-05,
+      "loss": 2.7483,
+      "step": 28905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.922418389640306e-05,
+      "loss": 2.7415,
+      "step": 28910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.921476194527851e-05,
+      "loss": 2.8113,
+      "step": 28915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.920533919355988e-05,
+      "loss": 2.846,
+      "step": 28920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.919591564163981e-05,
+      "loss": 2.6238,
+      "step": 28925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.918649128991093e-05,
+      "loss": 2.8677,
+      "step": 28930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.917706613876588e-05,
+      "loss": 2.7959,
+      "step": 28935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.916764018859742e-05,
+      "loss": 2.7591,
+      "step": 28940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.915821343979824e-05,
+      "loss": 2.8628,
+      "step": 28945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.914878589276113e-05,
+      "loss": 2.7386,
+      "step": 28950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.913935754787888e-05,
+      "loss": 2.5535,
+      "step": 28955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.912992840554436e-05,
+      "loss": 2.7152,
+      "step": 28960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.91204984661504e-05,
+      "loss": 2.6107,
+      "step": 28965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.911106773008992e-05,
+      "loss": 2.8026,
+      "step": 28970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.910163619775588e-05,
+      "loss": 2.6968,
+      "step": 28975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.909220386954123e-05,
+      "loss": 2.6135,
+      "step": 28980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.908277074583895e-05,
+      "loss": 2.7904,
+      "step": 28985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.907333682704212e-05,
+      "loss": 2.7783,
+      "step": 28990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.906390211354378e-05,
+      "loss": 2.5926,
+      "step": 28995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.905446660573705e-05,
+      "loss": 2.6837,
+      "step": 29000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.904503030401506e-05,
+      "loss": 2.7314,
+      "step": 29005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.903559320877098e-05,
+      "loss": 2.7784,
+      "step": 29010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9026155320398e-05,
+      "loss": 2.7263,
+      "step": 29015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.901671663928935e-05,
+      "loss": 2.803,
+      "step": 29020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.900727716583835e-05,
+      "loss": 2.9158,
+      "step": 29025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.899783690043822e-05,
+      "loss": 2.8037,
+      "step": 29030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.898839584348237e-05,
+      "loss": 2.801,
+      "step": 29035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.897895399536413e-05,
+      "loss": 2.7022,
+      "step": 29040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.89695113564769e-05,
+      "loss": 2.716,
+      "step": 29045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.896006792721411e-05,
+      "loss": 2.6774,
+      "step": 29050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.895062370796925e-05,
+      "loss": 2.6854,
+      "step": 29055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.894117869913577e-05,
+      "loss": 2.7259,
+      "step": 29060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.893173290110726e-05,
+      "loss": 2.7739,
+      "step": 29065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.892228631427726e-05,
+      "loss": 2.6492,
+      "step": 29070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.891283893903935e-05,
+      "loss": 2.8701,
+      "step": 29075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.890339077578719e-05,
+      "loss": 2.8051,
+      "step": 29080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.889394182491444e-05,
+      "loss": 2.6081,
+      "step": 29085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.888449208681475e-05,
+      "loss": 2.7774,
+      "step": 29090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.887504156188191e-05,
+      "loss": 2.6277,
+      "step": 29095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.886559025050965e-05,
+      "loss": 2.7253,
+      "step": 29100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.885613815309177e-05,
+      "loss": 2.6817,
+      "step": 29105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.88466852700221e-05,
+      "loss": 2.7232,
+      "step": 29110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.883723160169449e-05,
+      "loss": 2.7467,
+      "step": 29115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.882777714850284e-05,
+      "loss": 2.8531,
+      "step": 29120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.881832191084107e-05,
+      "loss": 2.7299,
+      "step": 29125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.880886588910316e-05,
+      "loss": 2.8111,
+      "step": 29130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.879940908368307e-05,
+      "loss": 2.6466,
+      "step": 29135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.878995149497483e-05,
+      "loss": 2.6207,
+      "step": 29140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.878049312337253e-05,
+      "loss": 2.8341,
+      "step": 29145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.877103396927021e-05,
+      "loss": 2.8066,
+      "step": 29150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.8761574033062e-05,
+      "loss": 2.8838,
+      "step": 29155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.875211331514208e-05,
+      "loss": 2.6152,
+      "step": 29160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.874265181590462e-05,
+      "loss": 2.8092,
+      "step": 29165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.873318953574384e-05,
+      "loss": 2.734,
+      "step": 29170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.872372647505398e-05,
+      "loss": 2.7381,
+      "step": 29175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.871426263422935e-05,
+      "loss": 2.6766,
+      "step": 29180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.870479801366425e-05,
+      "loss": 2.7288,
+      "step": 29185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.869533261375303e-05,
+      "loss": 2.6799,
+      "step": 29190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.868586643489007e-05,
+      "loss": 2.6013,
+      "step": 29195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.867639947746977e-05,
+      "loss": 2.7417,
+      "step": 29200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.86669317418866e-05,
+      "loss": 2.6706,
+      "step": 29205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.865746322853506e-05,
+      "loss": 2.7297,
+      "step": 29210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.86479939378096e-05,
+      "loss": 2.6737,
+      "step": 29215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.86385238701048e-05,
+      "loss": 2.7949,
+      "step": 29220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.862905302581523e-05,
+      "loss": 2.9001,
+      "step": 29225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.861958140533551e-05,
+      "loss": 2.7667,
+      "step": 29230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.861010900906026e-05,
+      "loss": 2.8009,
+      "step": 29235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.860063583738417e-05,
+      "loss": 2.6619,
+      "step": 29240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.859116189070193e-05,
+      "loss": 2.651,
+      "step": 29245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.858168716940828e-05,
+      "loss": 2.7314,
+      "step": 29250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.857221167389801e-05,
+      "loss": 2.735,
+      "step": 29255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.856273540456588e-05,
+      "loss": 2.8218,
+      "step": 29260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.855325836180677e-05,
+      "loss": 2.6467,
+      "step": 29265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.854378054601553e-05,
+      "loss": 2.8105,
+      "step": 29270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.853430195758705e-05,
+      "loss": 2.9006,
+      "step": 29275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.852482259691627e-05,
+      "loss": 2.844,
+      "step": 29280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.851534246439814e-05,
+      "loss": 2.71,
+      "step": 29285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.850586156042765e-05,
+      "loss": 2.698,
+      "step": 29290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.849637988539986e-05,
+      "loss": 2.7443,
+      "step": 29295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.848689743970982e-05,
+      "loss": 2.7486,
+      "step": 29300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.847741422375262e-05,
+      "loss": 2.7421,
+      "step": 29305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.846793023792333e-05,
+      "loss": 2.7395,
+      "step": 29310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.84584454826172e-05,
+      "loss": 2.6473,
+      "step": 29315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.844895995822934e-05,
+      "loss": 2.7365,
+      "step": 29320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.843947366515502e-05,
+      "loss": 2.7179,
+      "step": 29325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.842998660378946e-05,
+      "loss": 2.7525,
+      "step": 29330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.842049877452796e-05,
+      "loss": 2.5885,
+      "step": 29335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.841101017776584e-05,
+      "loss": 2.7479,
+      "step": 29340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.840152081389841e-05,
+      "loss": 2.8499,
+      "step": 29345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.83920306833211e-05,
+      "loss": 2.689,
+      "step": 29350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.83825397864293e-05,
+      "loss": 2.7205,
+      "step": 29355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.837304812361846e-05,
+      "loss": 2.7422,
+      "step": 29360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.836355569528403e-05,
+      "loss": 2.5928,
+      "step": 29365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.835406250182155e-05,
+      "loss": 2.7026,
+      "step": 29370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.834456854362655e-05,
+      "loss": 2.7735,
+      "step": 29375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.833507382109457e-05,
+      "loss": 2.7746,
+      "step": 29380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.832557833462128e-05,
+      "loss": 2.6449,
+      "step": 29385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.831608208460225e-05,
+      "loss": 2.5433,
+      "step": 29390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.830658507143319e-05,
+      "loss": 2.6728,
+      "step": 29395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.829708729550977e-05,
+      "loss": 2.8303,
+      "step": 29400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.828758875722772e-05,
+      "loss": 2.8524,
+      "step": 29405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.827808945698284e-05,
+      "loss": 2.7076,
+      "step": 29410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.826858939517087e-05,
+      "loss": 2.83,
+      "step": 29415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.825908857218768e-05,
+      "loss": 2.7385,
+      "step": 29420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.82495869884291e-05,
+      "loss": 2.7716,
+      "step": 29425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.824008464429103e-05,
+      "loss": 2.6666,
+      "step": 29430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.82305815401694e-05,
+      "loss": 2.7969,
+      "step": 29435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.822107767646014e-05,
+      "loss": 2.7175,
+      "step": 29440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.821157305355924e-05,
+      "loss": 2.7516,
+      "step": 29445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.820206767186274e-05,
+      "loss": 2.777,
+      "step": 29450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.819256153176666e-05,
+      "loss": 2.8097,
+      "step": 29455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.818305463366708e-05,
+      "loss": 2.697,
+      "step": 29460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.817354697796012e-05,
+      "loss": 2.8593,
+      "step": 29465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.816403856504193e-05,
+      "loss": 2.7672,
+      "step": 29470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.815452939530864e-05,
+      "loss": 2.8527,
+      "step": 29475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.814501946915652e-05,
+      "loss": 2.7533,
+      "step": 29480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.813550878698177e-05,
+      "loss": 2.7263,
+      "step": 29485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.812599734918065e-05,
+      "loss": 2.8269,
+      "step": 29490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.811648515614948e-05,
+      "loss": 2.6946,
+      "step": 29495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.810697220828457e-05,
+      "loss": 2.7524,
+      "step": 29500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.809745850598231e-05,
+      "loss": 2.7439,
+      "step": 29505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.808794404963907e-05,
+      "loss": 2.7421,
+      "step": 29510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.807842883965128e-05,
+      "loss": 2.78,
+      "step": 29515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.806891287641541e-05,
+      "loss": 2.8246,
+      "step": 29520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.805939616032795e-05,
+      "loss": 2.6538,
+      "step": 29525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.80498786917854e-05,
+      "loss": 2.6174,
+      "step": 29530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.804036047118431e-05,
+      "loss": 2.7135,
+      "step": 29535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.803084149892127e-05,
+      "loss": 2.7229,
+      "step": 29540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.80213217753929e-05,
+      "loss": 2.8562,
+      "step": 29545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.801180130099585e-05,
+      "loss": 2.6094,
+      "step": 29550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.800228007612676e-05,
+      "loss": 2.935,
+      "step": 29555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.79927581011824e-05,
+      "loss": 2.6713,
+      "step": 29560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.798323537655943e-05,
+      "loss": 2.7691,
+      "step": 29565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.797371190265467e-05,
+      "loss": 2.7583,
+      "step": 29570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.796418767986494e-05,
+      "loss": 2.6776,
+      "step": 29575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.795466270858702e-05,
+      "loss": 2.6844,
+      "step": 29580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.794513698921781e-05,
+      "loss": 2.6384,
+      "step": 29585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.793561052215419e-05,
+      "loss": 2.6154,
+      "step": 29590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.792608330779306e-05,
+      "loss": 2.7114,
+      "step": 29595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.791655534653143e-05,
+      "loss": 2.7395,
+      "step": 29600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.790702663876625e-05,
+      "loss": 2.7071,
+      "step": 29605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.789749718489458e-05,
+      "loss": 2.6748,
+      "step": 29610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.788796698531342e-05,
+      "loss": 2.6577,
+      "step": 29615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.787843604041988e-05,
+      "loss": 2.6571,
+      "step": 29620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.786890435061106e-05,
+      "loss": 2.8604,
+      "step": 29625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.785937191628411e-05,
+      "loss": 2.7456,
+      "step": 29630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.784983873783619e-05,
+      "loss": 2.6841,
+      "step": 29635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.784030481566454e-05,
+      "loss": 2.5594,
+      "step": 29640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.783077015016637e-05,
+      "loss": 2.8315,
+      "step": 29645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.782123474173895e-05,
+      "loss": 2.5987,
+      "step": 29650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.781169859077958e-05,
+      "loss": 2.815,
+      "step": 29655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.780216169768558e-05,
+      "loss": 2.9346,
+      "step": 29660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.779262406285432e-05,
+      "loss": 2.6391,
+      "step": 29665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.77830856866832e-05,
+      "loss": 2.7977,
+      "step": 29670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.777354656956964e-05,
+      "loss": 2.7844,
+      "step": 29675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.776400671191108e-05,
+      "loss": 2.5799,
+      "step": 29680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.7754466114105e-05,
+      "loss": 2.5911,
+      "step": 29685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.774492477654891e-05,
+      "loss": 2.7381,
+      "step": 29690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.773538269964038e-05,
+      "loss": 2.7233,
+      "step": 29695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.772583988377698e-05,
+      "loss": 2.6557,
+      "step": 29700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.771629632935632e-05,
+      "loss": 2.6232,
+      "step": 29705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.770675203677602e-05,
+      "loss": 2.6902,
+      "step": 29710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.769720700643376e-05,
+      "loss": 2.6887,
+      "step": 29715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.768766123872723e-05,
+      "loss": 2.6815,
+      "step": 29720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.767811473405418e-05,
+      "loss": 2.8723,
+      "step": 29725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.766856749281233e-05,
+      "loss": 2.7276,
+      "step": 29730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.765901951539952e-05,
+      "loss": 2.86,
+      "step": 29735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.764947080221356e-05,
+      "loss": 2.7589,
+      "step": 29740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.763992135365227e-05,
+      "loss": 2.7591,
+      "step": 29745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.763037117011357e-05,
+      "loss": 2.692,
+      "step": 29750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.762082025199536e-05,
+      "loss": 2.761,
+      "step": 29755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.761126859969557e-05,
+      "loss": 2.7113,
+      "step": 29760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.760171621361221e-05,
+      "loss": 2.7677,
+      "step": 29765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.759216309414325e-05,
+      "loss": 2.6748,
+      "step": 29770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.758260924168675e-05,
+      "loss": 2.788,
+      "step": 29775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.757305465664077e-05,
+      "loss": 2.7289,
+      "step": 29780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.756349933940341e-05,
+      "loss": 2.7214,
+      "step": 29785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.755394329037277e-05,
+      "loss": 2.7047,
+      "step": 29790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.754438650994704e-05,
+      "loss": 2.7921,
+      "step": 29795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.753482899852442e-05,
+      "loss": 2.72,
+      "step": 29800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.752527075650309e-05,
+      "loss": 2.7324,
+      "step": 29805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.751571178428132e-05,
+      "loss": 2.8514,
+      "step": 29810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.750615208225739e-05,
+      "loss": 2.7984,
+      "step": 29815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.749659165082961e-05,
+      "loss": 2.6721,
+      "step": 29820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.748703049039633e-05,
+      "loss": 2.6794,
+      "step": 29825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.747746860135591e-05,
+      "loss": 2.6915,
+      "step": 29830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.746790598410676e-05,
+      "loss": 2.7797,
+      "step": 29835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.745834263904729e-05,
+      "loss": 2.8041,
+      "step": 29840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.744877856657596e-05,
+      "loss": 2.6378,
+      "step": 29845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.743921376709131e-05,
+      "loss": 2.7435,
+      "step": 29850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.742964824099182e-05,
+      "loss": 2.787,
+      "step": 29855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.742008198867606e-05,
+      "loss": 2.7096,
+      "step": 29860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.74105150105426e-05,
+      "loss": 2.8045,
+      "step": 29865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.740094730699009e-05,
+      "loss": 2.8395,
+      "step": 29870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.739137887841712e-05,
+      "loss": 2.743,
+      "step": 29875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.73818097252224e-05,
+      "loss": 2.7329,
+      "step": 29880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.737223984780461e-05,
+      "loss": 2.762,
+      "step": 29885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.736266924656253e-05,
+      "loss": 2.8492,
+      "step": 29890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.735309792189488e-05,
+      "loss": 2.6369,
+      "step": 29895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.734352587420046e-05,
+      "loss": 2.9381,
+      "step": 29900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.733395310387812e-05,
+      "loss": 2.7921,
+      "step": 29905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.732437961132669e-05,
+      "loss": 2.7239,
+      "step": 29910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.731480539694504e-05,
+      "loss": 2.5713,
+      "step": 29915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.730523046113215e-05,
+      "loss": 2.7159,
+      "step": 29920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.72956548042869e-05,
+      "loss": 2.7596,
+      "step": 29925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.72860784268083e-05,
+      "loss": 2.6179,
+      "step": 29930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.727650132909533e-05,
+      "loss": 2.7353,
+      "step": 29935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.726692351154704e-05,
+      "loss": 2.7576,
+      "step": 29940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.72573449745625e-05,
+      "loss": 2.7831,
+      "step": 29945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.724776571854081e-05,
+      "loss": 2.8331,
+      "step": 29950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.723818574388107e-05,
+      "loss": 2.8196,
+      "step": 29955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.722860505098248e-05,
+      "loss": 2.7771,
+      "step": 29960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.721902364024416e-05,
+      "loss": 2.7443,
+      "step": 29965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.720944151206538e-05,
+      "loss": 2.6314,
+      "step": 29970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.719985866684535e-05,
+      "loss": 2.7781,
+      "step": 29975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.719027510498338e-05,
+      "loss": 2.7295,
+      "step": 29980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.718069082687874e-05,
+      "loss": 2.7873,
+      "step": 29985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.71711058329308e-05,
+      "loss": 2.563,
+      "step": 29990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.716152012353887e-05,
+      "loss": 2.8837,
+      "step": 29995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.71519336991024e-05,
+      "loss": 2.6708,
+      "step": 30000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.714234656002078e-05,
+      "loss": 2.7435,
+      "step": 30005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.71327587066935e-05,
+      "loss": 2.7439,
+      "step": 30010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.712317013951999e-05,
+      "loss": 2.7714,
+      "step": 30015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.71135808588998e-05,
+      "loss": 2.7242,
+      "step": 30020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.710399086523246e-05,
+      "loss": 2.6258,
+      "step": 30025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.709440015891755e-05,
+      "loss": 2.7621,
+      "step": 30030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.708480874035467e-05,
+      "loss": 2.6925,
+      "step": 30035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.707521660994346e-05,
+      "loss": 2.709,
+      "step": 30040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.706562376808357e-05,
+      "loss": 2.6874,
+      "step": 30045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.705603021517469e-05,
+      "loss": 2.8373,
+      "step": 30050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.704643595161654e-05,
+      "loss": 2.7644,
+      "step": 30055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.703684097780886e-05,
+      "loss": 2.7937,
+      "step": 30060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.702724529415148e-05,
+      "loss": 2.7998,
+      "step": 30065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.701764890104416e-05,
+      "loss": 2.6896,
+      "step": 30070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.700805179888675e-05,
+      "loss": 2.6671,
+      "step": 30075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.699845398807913e-05,
+      "loss": 2.7756,
+      "step": 30080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.69888554690212e-05,
+      "loss": 2.7249,
+      "step": 30085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.697925624211285e-05,
+      "loss": 2.7173,
+      "step": 30090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.696965630775409e-05,
+      "loss": 2.765,
+      "step": 30095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.696005566634487e-05,
+      "loss": 2.7368,
+      "step": 30100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.695045431828524e-05,
+      "loss": 2.7628,
+      "step": 30105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.69408522639752e-05,
+      "loss": 2.5783,
+      "step": 30110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.693124950381488e-05,
+      "loss": 2.766,
+      "step": 30115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.692164603820432e-05,
+      "loss": 2.797,
+      "step": 30120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.69120418675437e-05,
+      "loss": 2.8563,
+      "step": 30125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.690243699223319e-05,
+      "loss": 2.8356,
+      "step": 30130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.689283141267294e-05,
+      "loss": 2.9127,
+      "step": 30135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.688322512926322e-05,
+      "loss": 2.7705,
+      "step": 30140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.687361814240424e-05,
+      "loss": 2.687,
+      "step": 30145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.68640104524963e-05,
+      "loss": 2.8018,
+      "step": 30150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.685440205993971e-05,
+      "loss": 2.8239,
+      "step": 30155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.684479296513481e-05,
+      "loss": 2.6891,
+      "step": 30160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.683518316848199e-05,
+      "loss": 2.6809,
+      "step": 30165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.682557267038159e-05,
+      "loss": 2.7062,
+      "step": 30170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.681596147123408e-05,
+      "loss": 2.7443,
+      "step": 30175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.68063495714399e-05,
+      "loss": 2.7432,
+      "step": 30180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.679673697139954e-05,
+      "loss": 2.8279,
+      "step": 30185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.678712367151352e-05,
+      "loss": 2.8602,
+      "step": 30190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.677750967218239e-05,
+      "loss": 2.745,
+      "step": 30195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.676789497380671e-05,
+      "loss": 2.6688,
+      "step": 30200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.675827957678708e-05,
+      "loss": 2.6606,
+      "step": 30205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.674866348152413e-05,
+      "loss": 2.5776,
+      "step": 30210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.673904668841854e-05,
+      "loss": 2.7477,
+      "step": 30215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.672942919787097e-05,
+      "loss": 2.62,
+      "step": 30220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.671981101028217e-05,
+      "loss": 2.7441,
+      "step": 30225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.671019212605286e-05,
+      "loss": 2.845,
+      "step": 30230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.670057254558383e-05,
+      "loss": 2.7943,
+      "step": 30235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.669095226927587e-05,
+      "loss": 2.6212,
+      "step": 30240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.668133129752985e-05,
+      "loss": 2.6667,
+      "step": 30245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.667170963074659e-05,
+      "loss": 2.6667,
+      "step": 30250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.666208726932703e-05,
+      "loss": 2.8,
+      "step": 30255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.665246421367204e-05,
+      "loss": 2.8016,
+      "step": 30260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.664284046418259e-05,
+      "loss": 2.8574,
+      "step": 30265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.663321602125967e-05,
+      "loss": 2.8097,
+      "step": 30270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.662359088530427e-05,
+      "loss": 2.8037,
+      "step": 30275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.661396505671745e-05,
+      "loss": 2.6371,
+      "step": 30280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.660433853590026e-05,
+      "loss": 2.687,
+      "step": 30285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.65947113232538e-05,
+      "loss": 2.8047,
+      "step": 30290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.658508341917917e-05,
+      "loss": 2.8051,
+      "step": 30295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.657545482407756e-05,
+      "loss": 2.8606,
+      "step": 30300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.656582553835013e-05,
+      "loss": 2.6558,
+      "step": 30305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.655619556239811e-05,
+      "loss": 2.6388,
+      "step": 30310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.654656489662271e-05,
+      "loss": 2.6569,
+      "step": 30315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.653693354142521e-05,
+      "loss": 2.7232,
+      "step": 30320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.65273014972069e-05,
+      "loss": 2.6964,
+      "step": 30325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.65176687643691e-05,
+      "loss": 2.8785,
+      "step": 30330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.650803534331319e-05,
+      "loss": 2.6289,
+      "step": 30335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.649840123444053e-05,
+      "loss": 2.6907,
+      "step": 30340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.648876643815255e-05,
+      "loss": 2.7288,
+      "step": 30345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.647913095485065e-05,
+      "loss": 2.6376,
+      "step": 30350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.646949478493633e-05,
+      "loss": 2.8268,
+      "step": 30355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.645985792881109e-05,
+      "loss": 2.75,
+      "step": 30360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.645022038687642e-05,
+      "loss": 2.7399,
+      "step": 30365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.644058215953393e-05,
+      "loss": 2.6916,
+      "step": 30370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.643094324718517e-05,
+      "loss": 2.6834,
+      "step": 30375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.642130365023174e-05,
+      "loss": 2.8043,
+      "step": 30380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.641166336907529e-05,
+      "loss": 2.6964,
+      "step": 30385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.640202240411747e-05,
+      "loss": 2.7788,
+      "step": 30390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.639238075576001e-05,
+      "loss": 2.8035,
+      "step": 30395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.638273842440462e-05,
+      "loss": 2.6434,
+      "step": 30400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.637309541045305e-05,
+      "loss": 2.8482,
+      "step": 30405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.63634517143071e-05,
+      "loss": 2.6782,
+      "step": 30410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.635380733636855e-05,
+      "loss": 2.6745,
+      "step": 30415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.634416227703926e-05,
+      "loss": 2.7907,
+      "step": 30420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.633451653672108e-05,
+      "loss": 2.6842,
+      "step": 30425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.632487011581592e-05,
+      "loss": 2.6806,
+      "step": 30430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.63152230147257e-05,
+      "loss": 2.7169,
+      "step": 30435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.630557523385238e-05,
+      "loss": 2.6954,
+      "step": 30440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.629592677359793e-05,
+      "loss": 2.683,
+      "step": 30445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.628627763436436e-05,
+      "loss": 2.6487,
+      "step": 30450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.627662781655368e-05,
+      "loss": 2.643,
+      "step": 30455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.6266977320568e-05,
+      "loss": 2.7686,
+      "step": 30460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.62573261468094e-05,
+      "loss": 2.8001,
+      "step": 30465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.624767429568e-05,
+      "loss": 2.8341,
+      "step": 30470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.623802176758194e-05,
+      "loss": 2.8657,
+      "step": 30475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.622836856291743e-05,
+      "loss": 2.8615,
+      "step": 30480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.621871468208861e-05,
+      "loss": 2.7405,
+      "step": 30485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.620906012549776e-05,
+      "loss": 2.7167,
+      "step": 30490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.619940489354716e-05,
+      "loss": 2.7262,
+      "step": 30495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.61897489866391e-05,
+      "loss": 2.7404,
+      "step": 30500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.618009240517584e-05,
+      "loss": 2.7148,
+      "step": 30505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.617043514955979e-05,
+      "loss": 2.817,
+      "step": 30510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.616077722019326e-05,
+      "loss": 2.682,
+      "step": 30515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.615111861747872e-05,
+      "loss": 2.7685,
+      "step": 30520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.614145934181859e-05,
+      "loss": 2.9227,
+      "step": 30525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.613179939361529e-05,
+      "loss": 2.6836,
+      "step": 30530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.612213877327135e-05,
+      "loss": 2.6091,
+      "step": 30535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.611247748118926e-05,
+      "loss": 2.6058,
+      "step": 30540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.610281551777156e-05,
+      "loss": 2.556,
+      "step": 30545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.609315288342084e-05,
+      "loss": 2.755,
+      "step": 30550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.608348957853968e-05,
+      "loss": 2.7184,
+      "step": 30555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.607382560353072e-05,
+      "loss": 2.6724,
+      "step": 30560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.606416095879663e-05,
+      "loss": 2.8502,
+      "step": 30565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.605449564474007e-05,
+      "loss": 2.8055,
+      "step": 30570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.604482966176375e-05,
+      "loss": 2.7243,
+      "step": 30575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.60351630102704e-05,
+      "loss": 2.7151,
+      "step": 30580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.602549569066282e-05,
+      "loss": 2.7034,
+      "step": 30585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.601582770334378e-05,
+      "loss": 2.8525,
+      "step": 30590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.600615904871612e-05,
+      "loss": 2.6805,
+      "step": 30595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.599648972718266e-05,
+      "loss": 2.7599,
+      "step": 30600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.598681973914632e-05,
+      "loss": 2.742,
+      "step": 30605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.597714908500996e-05,
+      "loss": 2.845,
+      "step": 30610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.596747776517653e-05,
+      "loss": 2.8094,
+      "step": 30615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.595780578004901e-05,
+      "loss": 2.6492,
+      "step": 30620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.594813313003038e-05,
+      "loss": 2.9098,
+      "step": 30625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.593845981552365e-05,
+      "loss": 2.5996,
+      "step": 30630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.592878583693185e-05,
+      "loss": 2.6778,
+      "step": 30635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.591911119465807e-05,
+      "loss": 2.6575,
+      "step": 30640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.59094358891054e-05,
+      "loss": 2.7015,
+      "step": 30645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.589975992067699e-05,
+      "loss": 2.6259,
+      "step": 30650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.589008328977599e-05,
+      "loss": 2.6276,
+      "step": 30655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.588040599680556e-05,
+      "loss": 2.7365,
+      "step": 30660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.587072804216891e-05,
+      "loss": 2.7585,
+      "step": 30665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.58610494262693e-05,
+      "loss": 2.7894,
+      "step": 30670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.585137014950998e-05,
+      "loss": 2.7505,
+      "step": 30675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.584169021229424e-05,
+      "loss": 2.6693,
+      "step": 30680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.583200961502541e-05,
+      "loss": 2.8905,
+      "step": 30685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.582232835810684e-05,
+      "loss": 2.8793,
+      "step": 30690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.58126464419419e-05,
+      "loss": 2.5945,
+      "step": 30695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.580296386693399e-05,
+      "loss": 2.8621,
+      "step": 30700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.579328063348653e-05,
+      "loss": 2.724,
+      "step": 30705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.5783596742003e-05,
+      "loss": 2.6369,
+      "step": 30710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.577391219288687e-05,
+      "loss": 2.8155,
+      "step": 30715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.576422698654167e-05,
+      "loss": 2.7553,
+      "step": 30720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.575454112337092e-05,
+      "loss": 2.6746,
+      "step": 30725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.574485460377818e-05,
+      "loss": 2.6401,
+      "step": 30730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.573516742816707e-05,
+      "loss": 2.5624,
+      "step": 30735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.572547959694118e-05,
+      "loss": 2.7488,
+      "step": 30740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.57157911105042e-05,
+      "loss": 2.7215,
+      "step": 30745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.570610196925977e-05,
+      "loss": 2.7503,
+      "step": 30750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.569641217361162e-05,
+      "loss": 2.7255,
+      "step": 30755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.568672172396348e-05,
+      "loss": 2.7396,
+      "step": 30760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.567703062071907e-05,
+      "loss": 2.7473,
+      "step": 30765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.566733886428222e-05,
+      "loss": 2.7237,
+      "step": 30770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.56576464550567e-05,
+      "loss": 2.7976,
+      "step": 30775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.56479533934464e-05,
+      "loss": 2.8061,
+      "step": 30780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.563825967985517e-05,
+      "loss": 2.7921,
+      "step": 30785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.562856531468688e-05,
+      "loss": 2.879,
+      "step": 30790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.561887029834547e-05,
+      "loss": 2.8251,
+      "step": 30795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.560917463123488e-05,
+      "loss": 2.6294,
+      "step": 30800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.55994783137591e-05,
+      "loss": 2.7441,
+      "step": 30805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.558978134632213e-05,
+      "loss": 2.7029,
+      "step": 30810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.558008372932797e-05,
+      "loss": 2.6886,
+      "step": 30815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.557038546318073e-05,
+      "loss": 2.8076,
+      "step": 30820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.556068654828444e-05,
+      "loss": 2.6252,
+      "step": 30825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.555098698504325e-05,
+      "loss": 2.7232,
+      "step": 30830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.554128677386126e-05,
+      "loss": 2.7145,
+      "step": 30835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.553158591514268e-05,
+      "loss": 2.698,
+      "step": 30840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.552188440929168e-05,
+      "loss": 2.8436,
+      "step": 30845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.551218225671247e-05,
+      "loss": 2.678,
+      "step": 30850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.550247945780931e-05,
+      "loss": 2.8556,
+      "step": 30855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.549277601298647e-05,
+      "loss": 2.7481,
+      "step": 30860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.548307192264824e-05,
+      "loss": 2.6811,
+      "step": 30865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.547336718719894e-05,
+      "loss": 2.6479,
+      "step": 30870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.546366180704296e-05,
+      "loss": 2.804,
+      "step": 30875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.545395578258464e-05,
+      "loss": 2.696,
+      "step": 30880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.544424911422841e-05,
+      "loss": 2.7512,
+      "step": 30885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.543454180237869e-05,
+      "loss": 2.7036,
+      "step": 30890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.542483384743996e-05,
+      "loss": 2.7841,
+      "step": 30895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.541512524981667e-05,
+      "loss": 2.7895,
+      "step": 30900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.54054160099134e-05,
+      "loss": 2.6303,
+      "step": 30905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.539570612813462e-05,
+      "loss": 2.7299,
+      "step": 30910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.538599560488494e-05,
+      "loss": 2.9148,
+      "step": 30915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.537628444056894e-05,
+      "loss": 2.7203,
+      "step": 30920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.536657263559125e-05,
+      "loss": 2.7102,
+      "step": 30925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.53568601903565e-05,
+      "loss": 2.6674,
+      "step": 30930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.53471471052694e-05,
+      "loss": 2.7258,
+      "step": 30935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.533743338073461e-05,
+      "loss": 2.6825,
+      "step": 30940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.532771901715688e-05,
+      "loss": 2.6388,
+      "step": 30945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.531800401494095e-05,
+      "loss": 2.7029,
+      "step": 30950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.530828837449164e-05,
+      "loss": 2.6795,
+      "step": 30955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.529857209621369e-05,
+      "loss": 2.8048,
+      "step": 30960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.528885518051199e-05,
+      "loss": 2.7402,
+      "step": 30965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.52791376277914e-05,
+      "loss": 2.6732,
+      "step": 30970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.526941943845679e-05,
+      "loss": 2.6744,
+      "step": 30975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.525970061291306e-05,
+      "loss": 2.7363,
+      "step": 30980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.524998115156517e-05,
+      "loss": 2.8721,
+      "step": 30985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.52402610548181e-05,
+      "loss": 2.6172,
+      "step": 30990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.52305403230768e-05,
+      "loss": 2.7614,
+      "step": 30995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.522081895674633e-05,
+      "loss": 2.6898,
+      "step": 31000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.521109695623174e-05,
+      "loss": 2.7686,
+      "step": 31005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.520137432193804e-05,
+      "loss": 2.8032,
+      "step": 31010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.519165105427042e-05,
+      "loss": 2.7732,
+      "step": 31015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.518192715363394e-05,
+      "loss": 2.6897,
+      "step": 31020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.517220262043377e-05,
+      "loss": 2.5949,
+      "step": 31025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.51624774550751e-05,
+      "loss": 2.7504,
+      "step": 31030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.515275165796311e-05,
+      "loss": 2.5674,
+      "step": 31035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.514302522950307e-05,
+      "loss": 2.696,
+      "step": 31040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.513329817010018e-05,
+      "loss": 2.7854,
+      "step": 31045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.51235704801598e-05,
+      "loss": 2.6291,
+      "step": 31050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.511384216008716e-05,
+      "loss": 2.7991,
+      "step": 31055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.510411321028764e-05,
+      "loss": 2.5892,
+      "step": 31060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.50943836311666e-05,
+      "loss": 2.6041,
+      "step": 31065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.508465342312942e-05,
+      "loss": 2.6313,
+      "step": 31070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.507492258658152e-05,
+      "loss": 2.7322,
+      "step": 31075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.506519112192834e-05,
+      "loss": 2.8177,
+      "step": 31080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.505545902957534e-05,
+      "loss": 2.7959,
+      "step": 31085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.504572630992802e-05,
+      "loss": 2.6765,
+      "step": 31090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.50359929633919e-05,
+      "loss": 2.8082,
+      "step": 31095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.502625899037252e-05,
+      "loss": 2.6598,
+      "step": 31100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.501652439127547e-05,
+      "loss": 2.6797,
+      "step": 31105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.500678916650632e-05,
+      "loss": 2.6863,
+      "step": 31110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.499705331647072e-05,
+      "loss": 2.731,
+      "step": 31115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.498731684157429e-05,
+      "loss": 2.8168,
+      "step": 31120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.497757974222272e-05,
+      "loss": 2.6654,
+      "step": 31125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.496784201882172e-05,
+      "loss": 2.672,
+      "step": 31130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.495810367177701e-05,
+      "loss": 2.75,
+      "step": 31135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.494836470149435e-05,
+      "loss": 2.6544,
+      "step": 31140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.493862510837951e-05,
+      "loss": 2.732,
+      "step": 31145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.49288848928383e-05,
+      "loss": 2.8059,
+      "step": 31150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.491914405527657e-05,
+      "loss": 2.6484,
+      "step": 31155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.490940259610015e-05,
+      "loss": 2.7794,
+      "step": 31160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.489966051571492e-05,
+      "loss": 2.7849,
+      "step": 31165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.488991781452682e-05,
+      "loss": 2.8011,
+      "step": 31170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.488017449294177e-05,
+      "loss": 2.7795,
+      "step": 31175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.487043055136573e-05,
+      "loss": 2.8117,
+      "step": 31180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.486068599020466e-05,
+      "loss": 2.8006,
+      "step": 31185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.485094080986463e-05,
+      "loss": 2.8982,
+      "step": 31190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.484119501075162e-05,
+      "loss": 2.8349,
+      "step": 31195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.483144859327174e-05,
+      "loss": 2.59,
+      "step": 31200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.482170155783106e-05,
+      "loss": 2.8368,
+      "step": 31205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.48119539048357e-05,
+      "loss": 2.7044,
+      "step": 31210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.480220563469179e-05,
+      "loss": 2.8152,
+      "step": 31215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.479245674780552e-05,
+      "loss": 2.6806,
+      "step": 31220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.478270724458305e-05,
+      "loss": 2.8203,
+      "step": 31225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.477295712543064e-05,
+      "loss": 2.6403,
+      "step": 31230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.47632063907545e-05,
+      "loss": 2.7154,
+      "step": 31235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.475345504096092e-05,
+      "loss": 2.7798,
+      "step": 31240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.474370307645617e-05,
+      "loss": 2.8876,
+      "step": 31245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.47339504976466e-05,
+      "loss": 2.7277,
+      "step": 31250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.472419730493855e-05,
+      "loss": 2.756,
+      "step": 31255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.47144434987384e-05,
+      "loss": 2.7437,
+      "step": 31260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.470468907945252e-05,
+      "loss": 2.5245,
+      "step": 31265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.469493404748736e-05,
+      "loss": 2.6432,
+      "step": 31270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.468517840324935e-05,
+      "loss": 2.7712,
+      "step": 31275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.467542214714497e-05,
+      "loss": 2.7624,
+      "step": 31280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.466566527958071e-05,
+      "loss": 2.7392,
+      "step": 31285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.465590780096313e-05,
+      "loss": 2.7174,
+      "step": 31290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.464614971169877e-05,
+      "loss": 2.7339,
+      "step": 31295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.463639101219418e-05,
+      "loss": 2.6523,
+      "step": 31300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.462663170285597e-05,
+      "loss": 2.6271,
+      "step": 31305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.461687178409078e-05,
+      "loss": 2.6313,
+      "step": 31310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.460711125630526e-05,
+      "loss": 2.5876,
+      "step": 31315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.459735011990608e-05,
+      "loss": 2.7289,
+      "step": 31320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.458758837529996e-05,
+      "loss": 2.7383,
+      "step": 31325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.457782602289363e-05,
+      "loss": 2.5743,
+      "step": 31330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.456806306309382e-05,
+      "loss": 2.6609,
+      "step": 31335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.455829949630731e-05,
+      "loss": 2.8237,
+      "step": 31340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.454853532294092e-05,
+      "loss": 2.7895,
+      "step": 31345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.453877054340148e-05,
+      "loss": 2.6684,
+      "step": 31350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.452900515809584e-05,
+      "loss": 2.707,
+      "step": 31355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.451923916743087e-05,
+      "loss": 2.5769,
+      "step": 31360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.45094725718135e-05,
+      "loss": 2.7394,
+      "step": 31365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.449970537165064e-05,
+      "loss": 2.7309,
+      "step": 31370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.448993756734925e-05,
+      "loss": 2.682,
+      "step": 31375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.448016915931632e-05,
+      "loss": 2.7128,
+      "step": 31380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.447040014795885e-05,
+      "loss": 2.6891,
+      "step": 31385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.446063053368388e-05,
+      "loss": 2.7046,
+      "step": 31390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.445086031689844e-05,
+      "loss": 2.7651,
+      "step": 31395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.444108949800965e-05,
+      "loss": 2.7867,
+      "step": 31400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.443131807742457e-05,
+      "loss": 2.7737,
+      "step": 31405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.442154605555037e-05,
+      "loss": 2.6857,
+      "step": 31410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.441177343279421e-05,
+      "loss": 2.6661,
+      "step": 31415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.440200020956327e-05,
+      "loss": 2.6249,
+      "step": 31420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.439222638626472e-05,
+      "loss": 2.8528,
+      "step": 31425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.438245196330582e-05,
+      "loss": 2.7305,
+      "step": 31430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.437267694109382e-05,
+      "loss": 2.7617,
+      "step": 31435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.436290132003601e-05,
+      "loss": 2.7876,
+      "step": 31440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.43531251005397e-05,
+      "loss": 2.786,
+      "step": 31445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.43433482830122e-05,
+      "loss": 2.7541,
+      "step": 31450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.43335708678609e-05,
+      "loss": 2.5317,
+      "step": 31455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.432379285549315e-05,
+      "loss": 2.6218,
+      "step": 31460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.431401424631637e-05,
+      "loss": 2.7593,
+      "step": 31465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.430423504073798e-05,
+      "loss": 2.7259,
+      "step": 31470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.429445523916546e-05,
+      "loss": 2.7204,
+      "step": 31475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.428467484200627e-05,
+      "loss": 2.7963,
+      "step": 31480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.427489384966794e-05,
+      "loss": 2.4695,
+      "step": 31485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.426511226255797e-05,
+      "loss": 2.6956,
+      "step": 31490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.425533008108393e-05,
+      "loss": 2.8705,
+      "step": 31495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.42455473056534e-05,
+      "loss": 2.645,
+      "step": 31500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.423576393667398e-05,
+      "loss": 2.5338,
+      "step": 31505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.42259799745533e-05,
+      "loss": 2.6846,
+      "step": 31510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.421619541969904e-05,
+      "loss": 2.7433,
+      "step": 31515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.420641027251885e-05,
+      "loss": 2.8052,
+      "step": 31520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.419662453342043e-05,
+      "loss": 2.7121,
+      "step": 31525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.418683820281153e-05,
+      "loss": 2.7543,
+      "step": 31530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.417705128109988e-05,
+      "loss": 2.6302,
+      "step": 31535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.416726376869329e-05,
+      "loss": 2.6089,
+      "step": 31540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.415747566599954e-05,
+      "loss": 2.7268,
+      "step": 31545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.414768697342646e-05,
+      "loss": 2.6399,
+      "step": 31550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.41378976913819e-05,
+      "loss": 2.8022,
+      "step": 31555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.412810782027373e-05,
+      "loss": 2.7487,
+      "step": 31560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.411831736050985e-05,
+      "loss": 2.7391,
+      "step": 31565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.41085263124982e-05,
+      "loss": 2.776,
+      "step": 31570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.409873467664673e-05,
+      "loss": 2.6719,
+      "step": 31575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.40889424533634e-05,
+      "loss": 2.7193,
+      "step": 31580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.407914964305623e-05,
+      "loss": 2.6876,
+      "step": 31585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.406935624613322e-05,
+      "loss": 2.7482,
+      "step": 31590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.405956226300241e-05,
+      "loss": 2.7974,
+      "step": 31595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.40497676940719e-05,
+      "loss": 2.7202,
+      "step": 31600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.403997253974977e-05,
+      "loss": 2.6918,
+      "step": 31605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.403017680044415e-05,
+      "loss": 2.8218,
+      "step": 31610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.402038047656319e-05,
+      "loss": 2.8111,
+      "step": 31615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.401058356851504e-05,
+      "loss": 2.6825,
+      "step": 31620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.40007860767079e-05,
+      "loss": 2.8021,
+      "step": 31625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.399098800154997e-05,
+      "loss": 2.8344,
+      "step": 31630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.398118934344955e-05,
+      "loss": 2.8637,
+      "step": 31635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.397139010281487e-05,
+      "loss": 2.7806,
+      "step": 31640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.396159028005421e-05,
+      "loss": 2.7642,
+      "step": 31645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.39517898755759e-05,
+      "loss": 2.7511,
+      "step": 31650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.394198888978826e-05,
+      "loss": 2.7238,
+      "step": 31655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.393218732309967e-05,
+      "loss": 2.6478,
+      "step": 31660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.392238517591853e-05,
+      "loss": 2.5891,
+      "step": 31665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.391258244865325e-05,
+      "loss": 2.8893,
+      "step": 31670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.390277914171226e-05,
+      "loss": 2.7871,
+      "step": 31675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.389297525550399e-05,
+      "loss": 2.7454,
+      "step": 31680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.388317079043695e-05,
+      "loss": 2.8751,
+      "step": 31685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.387336574691967e-05,
+      "loss": 2.7365,
+      "step": 31690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.386356012536065e-05,
+      "loss": 2.7599,
+      "step": 31695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.385375392616847e-05,
+      "loss": 2.6977,
+      "step": 31700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.384394714975169e-05,
+      "loss": 2.7792,
+      "step": 31705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.383413979651894e-05,
+      "loss": 2.6663,
+      "step": 31710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.382433186687882e-05,
+      "loss": 2.6098,
+      "step": 31715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.381452336124e-05,
+      "loss": 2.6656,
+      "step": 31720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.380471428001113e-05,
+      "loss": 2.696,
+      "step": 31725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.379490462360097e-05,
+      "loss": 2.5643,
+      "step": 31730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.37850943924182e-05,
+      "loss": 2.6395,
+      "step": 31735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.377528358687159e-05,
+      "loss": 2.7493,
+      "step": 31740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.376547220736988e-05,
+      "loss": 2.5231,
+      "step": 31745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.37556602543219e-05,
+      "loss": 2.6977,
+      "step": 31750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.374584772813645e-05,
+      "loss": 2.8142,
+      "step": 31755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.37360346292224e-05,
+      "loss": 2.6638,
+      "step": 31760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.372622095798859e-05,
+      "loss": 2.702,
+      "step": 31765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.371640671484393e-05,
+      "loss": 2.7541,
+      "step": 31770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.370659190019732e-05,
+      "loss": 2.6821,
+      "step": 31775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.369677651445771e-05,
+      "loss": 2.8724,
+      "step": 31780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.368696055803409e-05,
+      "loss": 2.6822,
+      "step": 31785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.367714403133538e-05,
+      "loss": 2.7,
+      "step": 31790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.366732693477066e-05,
+      "loss": 2.505,
+      "step": 31795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.365750926874893e-05,
+      "loss": 2.6252,
+      "step": 31800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.364769103367926e-05,
+      "loss": 2.7813,
+      "step": 31805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.363787222997073e-05,
+      "loss": 2.6569,
+      "step": 31810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.362805285803244e-05,
+      "loss": 2.8021,
+      "step": 31815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.361823291827351e-05,
+      "loss": 2.6786,
+      "step": 31820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.360841241110313e-05,
+      "loss": 2.8243,
+      "step": 31825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.359859133693045e-05,
+      "loss": 2.8379,
+      "step": 31830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.358876969616468e-05,
+      "loss": 2.5278,
+      "step": 31835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.357894748921502e-05,
+      "loss": 2.8193,
+      "step": 31840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.356912471649076e-05,
+      "loss": 2.6694,
+      "step": 31845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.355930137840114e-05,
+      "loss": 2.6938,
+      "step": 31850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.354947747535546e-05,
+      "loss": 2.7747,
+      "step": 31855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.353965300776305e-05,
+      "loss": 2.6911,
+      "step": 31860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.352982797603324e-05,
+      "loss": 2.8175,
+      "step": 31865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.35200023805754e-05,
+      "loss": 2.7313,
+      "step": 31870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.351017622179892e-05,
+      "loss": 2.7208,
+      "step": 31875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.350034950011322e-05,
+      "loss": 2.7025,
+      "step": 31880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.34905222159277e-05,
+      "loss": 2.6892,
+      "step": 31885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.348069436965187e-05,
+      "loss": 2.7036,
+      "step": 31890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.34708659616952e-05,
+      "loss": 2.8318,
+      "step": 31895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.346103699246718e-05,
+      "loss": 2.7057,
+      "step": 31900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.345120746237733e-05,
+      "loss": 2.6772,
+      "step": 31905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.344137737183523e-05,
+      "loss": 2.8621,
+      "step": 31910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.343154672125043e-05,
+      "loss": 2.6996,
+      "step": 31915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.342171551103256e-05,
+      "loss": 2.7885,
+      "step": 31920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.341188374159124e-05,
+      "loss": 2.6813,
+      "step": 31925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.340205141333609e-05,
+      "loss": 2.7502,
+      "step": 31930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.33922185266768e-05,
+      "loss": 2.7732,
+      "step": 31935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.338238508202304e-05,
+      "loss": 2.7318,
+      "step": 31940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.337255107978455e-05,
+      "loss": 2.787,
+      "step": 31945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.336271652037107e-05,
+      "loss": 2.7569,
+      "step": 31950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.335288140419235e-05,
+      "loss": 2.6871,
+      "step": 31955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.334304573165819e-05,
+      "loss": 2.7886,
+      "step": 31960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.333320950317838e-05,
+      "loss": 2.8254,
+      "step": 31965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.332337271916276e-05,
+      "loss": 2.6876,
+      "step": 31970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.331353538002118e-05,
+      "loss": 2.756,
+      "step": 31975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.330369748616355e-05,
+      "loss": 2.5953,
+      "step": 31980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.329385903799974e-05,
+      "loss": 2.7923,
+      "step": 31985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.328402003593968e-05,
+      "loss": 2.641,
+      "step": 31990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.327418048039329e-05,
+      "loss": 2.756,
+      "step": 31995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.326434037177059e-05,
+      "loss": 2.776,
+      "step": 32000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.325449971048156e-05,
+      "loss": 2.6332,
+      "step": 32005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.32446584969362e-05,
+      "loss": 2.7728,
+      "step": 32010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.323481673154456e-05,
+      "loss": 2.8028,
+      "step": 32015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.322497441471672e-05,
+      "loss": 2.7082,
+      "step": 32020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.321513154686273e-05,
+      "loss": 2.7323,
+      "step": 32025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.320528812839273e-05,
+      "loss": 2.8235,
+      "step": 32030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.319544415971683e-05,
+      "loss": 2.8373,
+      "step": 32035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.318559964124519e-05,
+      "loss": 2.6476,
+      "step": 32040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.317575457338799e-05,
+      "loss": 2.6524,
+      "step": 32045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.316590895655545e-05,
+      "loss": 2.7997,
+      "step": 32050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.315606279115775e-05,
+      "loss": 2.6099,
+      "step": 32055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.314621607760518e-05,
+      "loss": 2.811,
+      "step": 32060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.313636881630798e-05,
+      "loss": 2.8249,
+      "step": 32065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.312652100767645e-05,
+      "loss": 2.7175,
+      "step": 32070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.311667265212092e-05,
+      "loss": 2.8077,
+      "step": 32075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.310682375005169e-05,
+      "loss": 2.6551,
+      "step": 32080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.309697430187916e-05,
+      "loss": 2.7904,
+      "step": 32085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.308712430801369e-05,
+      "loss": 2.8284,
+      "step": 32090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.30772737688657e-05,
+      "loss": 2.7642,
+      "step": 32095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.30674226848456e-05,
+      "loss": 2.6763,
+      "step": 32100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.305757105636385e-05,
+      "loss": 2.6743,
+      "step": 32105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.304771888383094e-05,
+      "loss": 2.6877,
+      "step": 32110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.303786616765735e-05,
+      "loss": 2.7118,
+      "step": 32115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.302801290825359e-05,
+      "loss": 2.8028,
+      "step": 32120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.301815910603022e-05,
+      "loss": 2.6022,
+      "step": 32125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.300830476139779e-05,
+      "loss": 2.7999,
+      "step": 32130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.29984498747669e-05,
+      "loss": 2.6282,
+      "step": 32135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.298859444654814e-05,
+      "loss": 2.6502,
+      "step": 32140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.297873847715217e-05,
+      "loss": 2.6514,
+      "step": 32145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.296888196698962e-05,
+      "loss": 2.7385,
+      "step": 32150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.29590249164712e-05,
+      "loss": 2.8323,
+      "step": 32155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.294916732600757e-05,
+      "loss": 2.7888,
+      "step": 32160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.293930919600946e-05,
+      "loss": 2.6352,
+      "step": 32165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.292945052688763e-05,
+      "loss": 2.6809,
+      "step": 32170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.291959131905283e-05,
+      "loss": 2.8387,
+      "step": 32175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.290973157291587e-05,
+      "loss": 2.7041,
+      "step": 32180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.289987128888756e-05,
+      "loss": 2.7318,
+      "step": 32185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.289001046737873e-05,
+      "loss": 2.6104,
+      "step": 32190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.28801491088002e-05,
+      "loss": 2.7827,
+      "step": 32195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.287028721356292e-05,
+      "loss": 2.6066,
+      "step": 32200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.286042478207775e-05,
+      "loss": 2.6986,
+      "step": 32205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.285056181475558e-05,
+      "loss": 2.7537,
+      "step": 32210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.284069831200742e-05,
+      "loss": 2.7446,
+      "step": 32215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.283083427424422e-05,
+      "loss": 2.789,
+      "step": 32220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.282096970187697e-05,
+      "loss": 2.7578,
+      "step": 32225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.281110459531665e-05,
+      "loss": 2.7534,
+      "step": 32230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.280123895497434e-05,
+      "loss": 2.6788,
+      "step": 32235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.279137278126106e-05,
+      "loss": 2.6491,
+      "step": 32240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.278150607458792e-05,
+      "loss": 2.7339,
+      "step": 32245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.277163883536602e-05,
+      "loss": 2.7328,
+      "step": 32250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.276177106400647e-05,
+      "loss": 2.7275,
+      "step": 32255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.27519027609204e-05,
+      "loss": 2.7518,
+      "step": 32260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.274203392651903e-05,
+      "loss": 2.7267,
+      "step": 32265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.273216456121349e-05,
+      "loss": 2.7379,
+      "step": 32270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.272229466541505e-05,
+      "loss": 2.5682,
+      "step": 32275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.271242423953492e-05,
+      "loss": 2.7566,
+      "step": 32280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.270255328398434e-05,
+      "loss": 2.7041,
+      "step": 32285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.26926817991746e-05,
+      "loss": 2.7248,
+      "step": 32290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.268280978551701e-05,
+      "loss": 2.7668,
+      "step": 32295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.267293724342289e-05,
+      "loss": 2.6181,
+      "step": 32300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.266306417330359e-05,
+      "loss": 2.7595,
+      "step": 32305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.265319057557046e-05,
+      "loss": 2.7869,
+      "step": 32310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.264331645063492e-05,
+      "loss": 2.6585,
+      "step": 32315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.263344179890834e-05,
+      "loss": 2.7004,
+      "step": 32320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.262356662080217e-05,
+      "loss": 2.6512,
+      "step": 32325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.261369091672787e-05,
+      "loss": 2.7266,
+      "step": 32330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.260381468709693e-05,
+      "loss": 2.7871,
+      "step": 32335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.259393793232083e-05,
+      "loss": 2.8658,
+      "step": 32340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.25840606528111e-05,
+      "loss": 2.6706,
+      "step": 32345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.257418284897927e-05,
+      "loss": 2.8016,
+      "step": 32350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.25643045212369e-05,
+      "loss": 2.6945,
+      "step": 32355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.255442566999559e-05,
+      "loss": 2.6889,
+      "step": 32360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.254454629566696e-05,
+      "loss": 2.7967,
+      "step": 32365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.253466639866262e-05,
+      "loss": 2.7974,
+      "step": 32370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.252478597939424e-05,
+      "loss": 2.7507,
+      "step": 32375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.251490503827347e-05,
+      "loss": 2.6774,
+      "step": 32380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.250502357571199e-05,
+      "loss": 2.6099,
+      "step": 32385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.249514159212156e-05,
+      "loss": 2.7051,
+      "step": 32390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.248525908791392e-05,
+      "loss": 2.7471,
+      "step": 32395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.24753760635008e-05,
+      "loss": 2.6784,
+      "step": 32400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.246549251929399e-05,
+      "loss": 2.7076,
+      "step": 32405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.24556084557053e-05,
+      "loss": 2.7143,
+      "step": 32410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.244572387314655e-05,
+      "loss": 2.7569,
+      "step": 32415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.243583877202958e-05,
+      "loss": 2.6959,
+      "step": 32420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.242595315276627e-05,
+      "loss": 2.5862,
+      "step": 32425
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.241606701576852e-05,
+      "loss": 2.7814,
+      "step": 32430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.240618036144821e-05,
+      "loss": 2.6706,
+      "step": 32435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.239629319021732e-05,
+      "loss": 2.5301,
+      "step": 32440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.238640550248776e-05,
+      "loss": 2.6761,
+      "step": 32445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.23765172986715e-05,
+      "loss": 2.7607,
+      "step": 32450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.236662857918057e-05,
+      "loss": 2.6563,
+      "step": 32455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.2356739344427e-05,
+      "loss": 2.7466,
+      "step": 32460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.23468495948228e-05,
+      "loss": 2.6732,
+      "step": 32465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.233695933078004e-05,
+      "loss": 2.7198,
+      "step": 32470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.23270685527108e-05,
+      "loss": 2.8107,
+      "step": 32475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.231717726102719e-05,
+      "loss": 2.7398,
+      "step": 32480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.230728545614133e-05,
+      "loss": 2.8237,
+      "step": 32485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.229739313846542e-05,
+      "loss": 2.7491,
+      "step": 32490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.228750030841154e-05,
+      "loss": 2.7647,
+      "step": 32495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.227760696639193e-05,
+      "loss": 2.8281,
+      "step": 32500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.22677131128188e-05,
+      "loss": 2.7839,
+      "step": 32505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.225781874810439e-05,
+      "loss": 2.8168,
+      "step": 32510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.224792387266091e-05,
+      "loss": 2.7604,
+      "step": 32515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.22380284869007e-05,
+      "loss": 2.8154,
+      "step": 32520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.2228132591236e-05,
+      "loss": 2.5955,
+      "step": 32525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.221823618607919e-05,
+      "loss": 2.9691,
+      "step": 32530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.220833927184255e-05,
+      "loss": 2.6916,
+      "step": 32535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.219844184893848e-05,
+      "loss": 2.774,
+      "step": 32540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.218854391777931e-05,
+      "loss": 2.7116,
+      "step": 32545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.21786454787775e-05,
+      "loss": 2.7253,
+      "step": 32550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.216874653234545e-05,
+      "loss": 2.7611,
+      "step": 32555
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.215884707889562e-05,
+      "loss": 2.6121,
+      "step": 32560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.214894711884046e-05,
+      "loss": 2.6935,
+      "step": 32565
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.213904665259247e-05,
+      "loss": 2.7775,
+      "step": 32570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.212914568056413e-05,
+      "loss": 2.6942,
+      "step": 32575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.211924420316797e-05,
+      "loss": 2.7861,
+      "step": 32580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.210934222081661e-05,
+      "loss": 2.6164,
+      "step": 32585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.209943973392255e-05,
+      "loss": 2.7031,
+      "step": 32590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.208953674289841e-05,
+      "loss": 2.6608,
+      "step": 32595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.207963324815679e-05,
+      "loss": 2.7135,
+      "step": 32600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.206972925011034e-05,
+      "loss": 2.7542,
+      "step": 32605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.205982474917172e-05,
+      "loss": 2.6829,
+      "step": 32610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.204991974575359e-05,
+      "loss": 2.7211,
+      "step": 32615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.204001424026864e-05,
+      "loss": 2.6957,
+      "step": 32620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.203010823312961e-05,
+      "loss": 2.7479,
+      "step": 32625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.202020172474926e-05,
+      "loss": 2.8659,
+      "step": 32630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.201029471554029e-05,
+      "loss": 2.8422,
+      "step": 32635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.200038720591551e-05,
+      "loss": 2.6875,
+      "step": 32640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.199047919628773e-05,
+      "loss": 2.7763,
+      "step": 32645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.198057068706978e-05,
+      "loss": 2.7325,
+      "step": 32650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.197066167867449e-05,
+      "loss": 2.5159,
+      "step": 32655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.196075217151473e-05,
+      "loss": 2.6884,
+      "step": 32660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.195084216600337e-05,
+      "loss": 2.7716,
+      "step": 32665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.194093166255333e-05,
+      "loss": 2.7372,
+      "step": 32670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.193102066157754e-05,
+      "loss": 2.7837,
+      "step": 32675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.192110916348895e-05,
+      "loss": 2.8005,
+      "step": 32680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.191119716870051e-05,
+      "loss": 2.7148,
+      "step": 32685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.190128467762523e-05,
+      "loss": 2.7629,
+      "step": 32690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.18913716906761e-05,
+      "loss": 2.8668,
+      "step": 32695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.188145820826616e-05,
+      "loss": 2.663,
+      "step": 32700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.187154423080845e-05,
+      "loss": 2.5702,
+      "step": 32705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.186162975871606e-05,
+      "loss": 2.7174,
+      "step": 32710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.185171479240208e-05,
+      "loss": 2.7105,
+      "step": 32715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.184179933227963e-05,
+      "loss": 2.6292,
+      "step": 32720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.183188337876181e-05,
+      "loss": 2.7392,
+      "step": 32725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.18219669322618e-05,
+      "loss": 2.6891,
+      "step": 32730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.181204999319277e-05,
+      "loss": 2.7657,
+      "step": 32735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.18021325619679e-05,
+      "loss": 2.7246,
+      "step": 32740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.179221463900043e-05,
+      "loss": 2.7623,
+      "step": 32745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.178229622470359e-05,
+      "loss": 2.8266,
+      "step": 32750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.177237731949064e-05,
+      "loss": 2.7091,
+      "step": 32755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.176245792377482e-05,
+      "loss": 2.7031,
+      "step": 32760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.175253803796947e-05,
+      "loss": 2.6851,
+      "step": 32765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.17426176624879e-05,
+      "loss": 2.7127,
+      "step": 32770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.173269679774343e-05,
+      "loss": 2.8365,
+      "step": 32775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.172277544414943e-05,
+      "loss": 2.5483,
+      "step": 32780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.17128536021193e-05,
+      "loss": 2.8059,
+      "step": 32785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.170293127206639e-05,
+      "loss": 2.7841,
+      "step": 32790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.169300845440415e-05,
+      "loss": 2.6934,
+      "step": 32795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.168308514954602e-05,
+      "loss": 2.6799,
+      "step": 32800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.167316135790548e-05,
+      "loss": 2.8302,
+      "step": 32805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.166323707989597e-05,
+      "loss": 2.7429,
+      "step": 32810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.165331231593101e-05,
+      "loss": 2.6818,
+      "step": 32815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.164338706642412e-05,
+      "loss": 2.8101,
+      "step": 32820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.163346133178885e-05,
+      "loss": 2.8414,
+      "step": 32825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.162353511243874e-05,
+      "loss": 2.8443,
+      "step": 32830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.161360840878737e-05,
+      "loss": 2.6904,
+      "step": 32835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.160368122124838e-05,
+      "loss": 2.785,
+      "step": 32840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.159375355023537e-05,
+      "loss": 2.7155,
+      "step": 32845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.158382539616195e-05,
+      "loss": 2.5949,
+      "step": 32850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.157389675944184e-05,
+      "loss": 2.7432,
+      "step": 32855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.15639676404887e-05,
+      "loss": 2.7507,
+      "step": 32860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.15540380397162e-05,
+      "loss": 2.5419,
+      "step": 32865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.154410795753811e-05,
+      "loss": 2.7542,
+      "step": 32870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.153417739436814e-05,
+      "loss": 2.8145,
+      "step": 32875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.152424635062007e-05,
+      "loss": 2.7179,
+      "step": 32880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.151431482670768e-05,
+      "loss": 2.7058,
+      "step": 32885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.150438282304476e-05,
+      "loss": 2.5459,
+      "step": 32890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.149445034004515e-05,
+      "loss": 2.7325,
+      "step": 32895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.148451737812268e-05,
+      "loss": 2.793,
+      "step": 32900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.147458393769123e-05,
+      "loss": 2.6518,
+      "step": 32905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.146465001916465e-05,
+      "loss": 2.6508,
+      "step": 32910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.145471562295687e-05,
+      "loss": 2.6497,
+      "step": 32915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.14447807494818e-05,
+      "loss": 2.7006,
+      "step": 32920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.143484539915341e-05,
+      "loss": 2.6382,
+      "step": 32925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.142490957238563e-05,
+      "loss": 2.7333,
+      "step": 32930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.141497326959243e-05,
+      "loss": 2.6758,
+      "step": 32935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.140503649118786e-05,
+      "loss": 2.6484,
+      "step": 32940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.139509923758589e-05,
+      "loss": 2.8161,
+      "step": 32945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.13851615092006e-05,
+      "loss": 2.7324,
+      "step": 32950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.137522330644604e-05,
+      "loss": 2.7396,
+      "step": 32955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.136528462973628e-05,
+      "loss": 2.8019,
+      "step": 32960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.135534547948543e-05,
+      "loss": 2.7663,
+      "step": 32965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.13454058561076e-05,
+      "loss": 2.7638,
+      "step": 32970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.133546576001695e-05,
+      "loss": 2.5873,
+      "step": 32975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.132552519162761e-05,
+      "loss": 2.8214,
+      "step": 32980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.13155841513538e-05,
+      "loss": 2.8279,
+      "step": 32985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.130564263960968e-05,
+      "loss": 2.7117,
+      "step": 32990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.129570065680951e-05,
+      "loss": 2.7143,
+      "step": 32995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.128575820336749e-05,
+      "loss": 2.6646,
+      "step": 33000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.127581527969788e-05,
+      "loss": 2.7585,
+      "step": 33005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.126587188621499e-05,
+      "loss": 2.7545,
+      "step": 33010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.125592802333309e-05,
+      "loss": 2.7076,
+      "step": 33015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.124598369146649e-05,
+      "loss": 2.7885,
+      "step": 33020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.123603889102955e-05,
+      "loss": 2.7081,
+      "step": 33025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.122609362243663e-05,
+      "loss": 2.5749,
+      "step": 33030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.121614788610206e-05,
+      "loss": 2.7886,
+      "step": 33035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.12062016824403e-05,
+      "loss": 2.8124,
+      "step": 33040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.119625501186572e-05,
+      "loss": 2.6771,
+      "step": 33045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.118630787479276e-05,
+      "loss": 2.8551,
+      "step": 33050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.117636027163587e-05,
+      "loss": 2.7097,
+      "step": 33055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.116641220280954e-05,
+      "loss": 2.7746,
+      "step": 33060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.115646366872824e-05,
+      "loss": 2.6933,
+      "step": 33065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.114651466980652e-05,
+      "loss": 2.7914,
+      "step": 33070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.113656520645886e-05,
+      "loss": 2.6437,
+      "step": 33075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.112661527909984e-05,
+      "loss": 2.7964,
+      "step": 33080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.111666488814402e-05,
+      "loss": 2.5938,
+      "step": 33085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1106714034006e-05,
+      "loss": 2.7284,
+      "step": 33090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.109676271710037e-05,
+      "loss": 2.6781,
+      "step": 33095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.108681093784177e-05,
+      "loss": 2.8412,
+      "step": 33100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.107685869664485e-05,
+      "loss": 2.7315,
+      "step": 33105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.106690599392427e-05,
+      "loss": 2.6836,
+      "step": 33110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.10569528300947e-05,
+      "loss": 2.7132,
+      "step": 33115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.104699920557088e-05,
+      "loss": 2.6391,
+      "step": 33120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.10370451207675e-05,
+      "loss": 2.6903,
+      "step": 33125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.102709057609931e-05,
+      "loss": 2.6938,
+      "step": 33130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1017135571981086e-05,
+      "loss": 2.6123,
+      "step": 33135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.10071801088276e-05,
+      "loss": 2.5325,
+      "step": 33140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.099722418705366e-05,
+      "loss": 2.7296,
+      "step": 33145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0987267807074054e-05,
+      "loss": 2.7365,
+      "step": 33150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.097731096930367e-05,
+      "loss": 2.8885,
+      "step": 33155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.096735367415731e-05,
+      "loss": 2.6559,
+      "step": 33160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.095739592204991e-05,
+      "loss": 2.8397,
+      "step": 33165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0947437713396315e-05,
+      "loss": 2.6907,
+      "step": 33170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0937479048611465e-05,
+      "loss": 2.834,
+      "step": 33175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.092751992811029e-05,
+      "loss": 2.8537,
+      "step": 33180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.091756035230775e-05,
+      "loss": 2.7449,
+      "step": 33185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0907600321618775e-05,
+      "loss": 2.7611,
+      "step": 33190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.089763983645841e-05,
+      "loss": 2.7506,
+      "step": 33195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.088767889724164e-05,
+      "loss": 2.7618,
+      "step": 33200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.08777175043835e-05,
+      "loss": 2.6947,
+      "step": 33205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.086775565829902e-05,
+      "loss": 2.6745,
+      "step": 33210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.085779335940328e-05,
+      "loss": 2.905,
+      "step": 33215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.084783060811136e-05,
+      "loss": 2.6806,
+      "step": 33220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0837867404838366e-05,
+      "loss": 2.6454,
+      "step": 33225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.082790374999944e-05,
+      "loss": 2.5818,
+      "step": 33230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.081793964400969e-05,
+      "loss": 2.7857,
+      "step": 33235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.080797508728429e-05,
+      "loss": 2.6551,
+      "step": 33240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.079801008023841e-05,
+      "loss": 2.628,
+      "step": 33245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0788044623287265e-05,
+      "loss": 2.8291,
+      "step": 33250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0778078716846063e-05,
+      "loss": 2.7828,
+      "step": 33255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0768112361330043e-05,
+      "loss": 2.6866,
+      "step": 33260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.075814555715447e-05,
+      "loss": 2.6621,
+      "step": 33265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.074817830473458e-05,
+      "loss": 2.7632,
+      "step": 33270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.073821060448569e-05,
+      "loss": 2.7791,
+      "step": 33275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0728242456823104e-05,
+      "loss": 2.7847,
+      "step": 33280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.071827386216216e-05,
+      "loss": 2.7486,
+      "step": 33285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0708304820918204e-05,
+      "loss": 2.7499,
+      "step": 33290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0698335333506596e-05,
+      "loss": 2.5478,
+      "step": 33295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.068836540034271e-05,
+      "loss": 2.765,
+      "step": 33300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.067839502184196e-05,
+      "loss": 2.6754,
+      "step": 33305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0668424198419774e-05,
+      "loss": 2.8037,
+      "step": 33310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.065845293049157e-05,
+      "loss": 2.7204,
+      "step": 33315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.064848121847284e-05,
+      "loss": 2.64,
+      "step": 33320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0638509062779036e-05,
+      "loss": 2.8232,
+      "step": 33325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.062853646382567e-05,
+      "loss": 2.7238,
+      "step": 33330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.061856342202823e-05,
+      "loss": 2.7787,
+      "step": 33335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.060858993780226e-05,
+      "loss": 2.6339,
+      "step": 33340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0598616011563324e-05,
+      "loss": 2.7869,
+      "step": 33345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0588641643727e-05,
+      "loss": 2.588,
+      "step": 33350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.057866683470884e-05,
+      "loss": 2.7817,
+      "step": 33355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0568691584924474e-05,
+      "loss": 2.7703,
+      "step": 33360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.055871589478953e-05,
+      "loss": 2.8355,
+      "step": 33365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.054873976471963e-05,
+      "loss": 2.6212,
+      "step": 33370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.053876319513044e-05,
+      "loss": 2.6564,
+      "step": 33375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.052878618643767e-05,
+      "loss": 2.7508,
+      "step": 33380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.051880873905698e-05,
+      "loss": 2.6988,
+      "step": 33385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.050883085340411e-05,
+      "loss": 2.5796,
+      "step": 33390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.049885252989478e-05,
+      "loss": 2.7322,
+      "step": 33395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0488873768944745e-05,
+      "loss": 2.7657,
+      "step": 33400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.047889457096977e-05,
+      "loss": 2.6367,
+      "step": 33405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0468914936385665e-05,
+      "loss": 2.6724,
+      "step": 33410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.045893486560823e-05,
+      "loss": 2.7866,
+      "step": 33415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.044895435905328e-05,
+      "loss": 2.7308,
+      "step": 33420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.043897341713667e-05,
+      "loss": 2.7878,
+      "step": 33425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.042899204027423e-05,
+      "loss": 2.6931,
+      "step": 33430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.041901022888189e-05,
+      "loss": 2.5716,
+      "step": 33435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.040902798337551e-05,
+      "loss": 2.7354,
+      "step": 33440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.039904530417102e-05,
+      "loss": 2.7933,
+      "step": 33445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0389062191684366e-05,
+      "loss": 2.6704,
+      "step": 33450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.037907864633148e-05,
+      "loss": 2.7686,
+      "step": 33455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.036909466852834e-05,
+      "loss": 2.6681,
+      "step": 33460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.035911025869092e-05,
+      "loss": 2.7384,
+      "step": 33465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.034912541723524e-05,
+      "loss": 2.7441,
+      "step": 33470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0339140144577346e-05,
+      "loss": 2.6202,
+      "step": 33475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0329154441133254e-05,
+      "loss": 2.6664,
+      "step": 33480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.031916830731902e-05,
+      "loss": 2.8469,
+      "step": 33485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0309181743550725e-05,
+      "loss": 2.6904,
+      "step": 33490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.029919475024447e-05,
+      "loss": 2.6229,
+      "step": 33495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0289207327816365e-05,
+      "loss": 2.763,
+      "step": 33500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.027921947668256e-05,
+      "loss": 2.8568,
+      "step": 33505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.026923119725918e-05,
+      "loss": 2.7598,
+      "step": 33510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.025924248996241e-05,
+      "loss": 2.7329,
+      "step": 33515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0249253355208414e-05,
+      "loss": 2.7354,
+      "step": 33520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0239263793413414e-05,
+      "loss": 2.6741,
+      "step": 33525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0229273804993613e-05,
+      "loss": 2.7812,
+      "step": 33530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0219283390365286e-05,
+      "loss": 2.7401,
+      "step": 33535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.020929254994465e-05,
+      "loss": 2.7078,
+      "step": 33540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0199301284148004e-05,
+      "loss": 2.7512,
+      "step": 33545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.018930959339163e-05,
+      "loss": 2.666,
+      "step": 33550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.017931747809184e-05,
+      "loss": 2.5773,
+      "step": 33555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.016932493866494e-05,
+      "loss": 2.7502,
+      "step": 33560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0159331975527286e-05,
+      "loss": 2.5737,
+      "step": 33565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0149338589095275e-05,
+      "loss": 2.6043,
+      "step": 33570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0139344779785256e-05,
+      "loss": 2.7032,
+      "step": 33575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.012935054801363e-05,
+      "loss": 2.6756,
+      "step": 33580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0119355894196815e-05,
+      "loss": 2.6339,
+      "step": 33585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.010936081875122e-05,
+      "loss": 2.6188,
+      "step": 33590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.009936532209334e-05,
+      "loss": 2.7171,
+      "step": 33595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.008936940463961e-05,
+      "loss": 2.7315,
+      "step": 33600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.007937306680653e-05,
+      "loss": 2.8033,
+      "step": 33605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0069376309010615e-05,
+      "loss": 2.5506,
+      "step": 33610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.005937913166836e-05,
+      "loss": 2.7297,
+      "step": 33615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.00493815351963e-05,
+      "loss": 2.6879,
+      "step": 33620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0039383520011e-05,
+      "loss": 2.6337,
+      "step": 33625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.002938508652907e-05,
+      "loss": 2.8736,
+      "step": 33630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0019386235167055e-05,
+      "loss": 2.6201,
+      "step": 33635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.000938696634156e-05,
+      "loss": 2.7334,
+      "step": 33640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.999938728046924e-05,
+      "loss": 2.6604,
+      "step": 33645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9989387177966714e-05,
+      "loss": 2.7612,
+      "step": 33650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.997938665925067e-05,
+      "loss": 2.6665,
+      "step": 33655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9969385724737735e-05,
+      "loss": 2.7415,
+      "step": 33660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.995938437484466e-05,
+      "loss": 2.7953,
+      "step": 33665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9949382609988124e-05,
+      "loss": 2.887,
+      "step": 33670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.993938043058488e-05,
+      "loss": 2.6281,
+      "step": 33675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.992937783705164e-05,
+      "loss": 2.667,
+      "step": 33680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.991937482980519e-05,
+      "loss": 2.7269,
+      "step": 33685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9909371409262305e-05,
+      "loss": 2.5798,
+      "step": 33690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.989936757583979e-05,
+      "loss": 2.5966,
+      "step": 33695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.988936332995446e-05,
+      "loss": 2.8078,
+      "step": 33700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9879358672023147e-05,
+      "loss": 2.7674,
+      "step": 33705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9869353602462685e-05,
+      "loss": 2.7368,
+      "step": 33710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9859348121689974e-05,
+      "loss": 2.6427,
+      "step": 33715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9849342230121855e-05,
+      "loss": 2.6577,
+      "step": 33720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9839335928175265e-05,
+      "loss": 2.7841,
+      "step": 33725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9829329216267114e-05,
+      "loss": 2.6786,
+      "step": 33730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.981932209481433e-05,
+      "loss": 2.7829,
+      "step": 33735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.980931456423386e-05,
+      "loss": 2.6436,
+      "step": 33740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.979930662494269e-05,
+      "loss": 2.678,
+      "step": 33745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9789298277357796e-05,
+      "loss": 2.8462,
+      "step": 33750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9779289521896174e-05,
+      "loss": 2.7641,
+      "step": 33755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9769280358974876e-05,
+      "loss": 2.6871,
+      "step": 33760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.975927078901091e-05,
+      "loss": 2.6672,
+      "step": 33765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9749260812421336e-05,
+      "loss": 2.5663,
+      "step": 33770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.973925042962322e-05,
+      "loss": 2.6724,
+      "step": 33775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.972923964103367e-05,
+      "loss": 2.6316,
+      "step": 33780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.971922844706976e-05,
+      "loss": 2.7973,
+      "step": 33785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.970921684814865e-05,
+      "loss": 2.6354,
+      "step": 33790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.969920484468746e-05,
+      "loss": 2.5836,
+      "step": 33795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9689192437103345e-05,
+      "loss": 2.6316,
+      "step": 33800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.967917962581347e-05,
+      "loss": 2.7777,
+      "step": 33805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.966916641123505e-05,
+      "loss": 2.681,
+      "step": 33810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.965915279378525e-05,
+      "loss": 2.7106,
+      "step": 33815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.964913877388134e-05,
+      "loss": 2.7625,
+      "step": 33820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9639124351940545e-05,
+      "loss": 2.844,
+      "step": 33825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.96291095283801e-05,
+      "loss": 2.7228,
+      "step": 33830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.961909430361728e-05,
+      "loss": 2.8071,
+      "step": 33835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.960907867806942e-05,
+      "loss": 2.654,
+      "step": 33840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.959906265215376e-05,
+      "loss": 2.7212,
+      "step": 33845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.958904622628768e-05,
+      "loss": 2.6829,
+      "step": 33850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9579029400888496e-05,
+      "loss": 2.8293,
+      "step": 33855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9569012176373576e-05,
+      "loss": 2.6851,
+      "step": 33860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.955899455316026e-05,
+      "loss": 2.6946,
+      "step": 33865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9548976531665975e-05,
+      "loss": 2.7018,
+      "step": 33870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.953895811230812e-05,
+      "loss": 2.7447,
+      "step": 33875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9528939295504105e-05,
+      "loss": 2.6014,
+      "step": 33880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.951892008167138e-05,
+      "loss": 2.7392,
+      "step": 33885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9508900471227416e-05,
+      "loss": 2.7883,
+      "step": 33890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9498880464589636e-05,
+      "loss": 2.7069,
+      "step": 33895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.948886006217559e-05,
+      "loss": 2.6957,
+      "step": 33900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9478839264402755e-05,
+      "loss": 2.716,
+      "step": 33905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.946881807168864e-05,
+      "loss": 2.6909,
+      "step": 33910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9458796484450794e-05,
+      "loss": 2.8136,
+      "step": 33915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9448774503106794e-05,
+      "loss": 2.6608,
+      "step": 33920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.943875212807418e-05,
+      "loss": 2.7602,
+      "step": 33925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9428729359770565e-05,
+      "loss": 2.6548,
+      "step": 33930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9418706198613536e-05,
+      "loss": 2.8386,
+      "step": 33935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.940868264502072e-05,
+      "loss": 2.7659,
+      "step": 33940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.939865869940976e-05,
+      "loss": 2.7542,
+      "step": 33945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9388634362198304e-05,
+      "loss": 2.6689,
+      "step": 33950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.937860963380403e-05,
+      "loss": 2.7703,
+      "step": 33955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.936858451464459e-05,
+      "loss": 2.6722,
+      "step": 33960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.935855900513773e-05,
+      "loss": 2.7719,
+      "step": 33965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9348533105701146e-05,
+      "loss": 2.6881,
+      "step": 33970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.933850681675257e-05,
+      "loss": 2.7671,
+      "step": 33975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9328480138709775e-05,
+      "loss": 2.645,
+      "step": 33980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.931845307199051e-05,
+      "loss": 2.7852,
+      "step": 33985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.930842561701254e-05,
+      "loss": 2.7631,
+      "step": 33990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.929839777419372e-05,
+      "loss": 2.6696,
+      "step": 33995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9288369543951814e-05,
+      "loss": 2.8434,
+      "step": 34000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.927834092670467e-05,
+      "loss": 2.7373,
+      "step": 34005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9268311922870145e-05,
+      "loss": 2.7279,
+      "step": 34010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9258282532866095e-05,
+      "loss": 2.6956,
+      "step": 34015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.92482527571104e-05,
+      "loss": 2.8364,
+      "step": 34020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9238222596020966e-05,
+      "loss": 2.8407,
+      "step": 34025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.92281920500157e-05,
+      "loss": 2.7566,
+      "step": 34030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.921816111951251e-05,
+      "loss": 2.6755,
+      "step": 34035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.920812980492937e-05,
+      "loss": 2.7299,
+      "step": 34040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.919809810668423e-05,
+      "loss": 2.7643,
+      "step": 34045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.918806602519505e-05,
+      "loss": 2.7813,
+      "step": 34050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9178033560879866e-05,
+      "loss": 2.6639,
+      "step": 34055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9168000714156646e-05,
+      "loss": 2.8338,
+      "step": 34060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.915796748544343e-05,
+      "loss": 2.8178,
+      "step": 34065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.914793387515824e-05,
+      "loss": 2.7548,
+      "step": 34070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.913789988371916e-05,
+      "loss": 2.7161,
+      "step": 34075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.912786551154424e-05,
+      "loss": 2.5868,
+      "step": 34080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.911783075905158e-05,
+      "loss": 2.707,
+      "step": 34085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.910779562665929e-05,
+      "loss": 2.8315,
+      "step": 34090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9097760114785486e-05,
+      "loss": 2.6006,
+      "step": 34095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9087724223848276e-05,
+      "loss": 2.7414,
+      "step": 34100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.907768795426585e-05,
+      "loss": 2.6875,
+      "step": 34105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.906765130645634e-05,
+      "loss": 2.6458,
+      "step": 34110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.905761428083796e-05,
+      "loss": 2.7925,
+      "step": 34115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.904757687782889e-05,
+      "loss": 2.8167,
+      "step": 34120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.903753909784735e-05,
+      "loss": 2.7423,
+      "step": 34125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.902750094131157e-05,
+      "loss": 2.6419,
+      "step": 34130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.901746240863979e-05,
+      "loss": 2.5608,
+      "step": 34135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.900742350025027e-05,
+      "loss": 2.6478,
+      "step": 34140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.89973842165613e-05,
+      "loss": 2.7523,
+      "step": 34145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8987344557991164e-05,
+      "loss": 2.7174,
+      "step": 34150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8977304524958176e-05,
+      "loss": 2.8026,
+      "step": 34155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.896726411788066e-05,
+      "loss": 2.6939,
+      "step": 34160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.895722333717692e-05,
+      "loss": 2.749,
+      "step": 34165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.894718218326537e-05,
+      "loss": 2.77,
+      "step": 34170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.893714065656434e-05,
+      "loss": 2.6259,
+      "step": 34175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.892709875749223e-05,
+      "loss": 2.6673,
+      "step": 34180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.891705648646744e-05,
+      "loss": 2.6138,
+      "step": 34185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8907013843908384e-05,
+      "loss": 2.7063,
+      "step": 34190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.889697083023349e-05,
+      "loss": 2.8353,
+      "step": 34195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.888692744586122e-05,
+      "loss": 2.7063,
+      "step": 34200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8876883691210014e-05,
+      "loss": 2.6741,
+      "step": 34205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8866839566698385e-05,
+      "loss": 2.737,
+      "step": 34210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.885679507274481e-05,
+      "loss": 2.7805,
+      "step": 34215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.884675020976779e-05,
+      "loss": 2.7921,
+      "step": 34220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.883670497818585e-05,
+      "loss": 2.7309,
+      "step": 34225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.882665937841755e-05,
+      "loss": 2.9162,
+      "step": 34230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8816613410881416e-05,
+      "loss": 2.7334,
+      "step": 34235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8806567075996043e-05,
+      "loss": 2.7071,
+      "step": 34240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8796520374180016e-05,
+      "loss": 2.7029,
+      "step": 34245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.878647330585194e-05,
+      "loss": 2.6577,
+      "step": 34250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.877642587143039e-05,
+      "loss": 2.6851,
+      "step": 34255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.876637807133406e-05,
+      "loss": 2.6957,
+      "step": 34260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.875632990598155e-05,
+      "loss": 2.7808,
+      "step": 34265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.874628137579156e-05,
+      "loss": 2.7007,
+      "step": 34270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.873623248118274e-05,
+      "loss": 2.6142,
+      "step": 34275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.872618322257379e-05,
+      "loss": 2.762,
+      "step": 34280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8716133600383426e-05,
+      "loss": 2.6619,
+      "step": 34285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.870608361503035e-05,
+      "loss": 2.8103,
+      "step": 34290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.869603326693332e-05,
+      "loss": 2.6046,
+      "step": 34295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8685982556511086e-05,
+      "loss": 2.6588,
+      "step": 34300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.867593148418242e-05,
+      "loss": 2.6864,
+      "step": 34305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8665880050366106e-05,
+      "loss": 2.7935,
+      "step": 34310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8655828255480924e-05,
+      "loss": 2.6543,
+      "step": 34315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.86457760999457e-05,
+      "loss": 2.7718,
+      "step": 34320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.863572358417927e-05,
+      "loss": 2.7867,
+      "step": 34325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.862567070860048e-05,
+      "loss": 2.7507,
+      "step": 34330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.861561747362817e-05,
+      "loss": 2.5815,
+      "step": 34335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8605563879681235e-05,
+      "loss": 2.7389,
+      "step": 34340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.859550992717855e-05,
+      "loss": 2.8115,
+      "step": 34345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8585455616539006e-05,
+      "loss": 2.8477,
+      "step": 34350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8575400948181544e-05,
+      "loss": 2.6124,
+      "step": 34355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.856534592252511e-05,
+      "loss": 2.7988,
+      "step": 34360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.855529053998862e-05,
+      "loss": 2.6412,
+      "step": 34365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.854523480099106e-05,
+      "loss": 2.8822,
+      "step": 34370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8535178705951385e-05,
+      "loss": 2.6728,
+      "step": 34375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.852512225528862e-05,
+      "loss": 2.741,
+      "step": 34380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.851506544942174e-05,
+      "loss": 2.7575,
+      "step": 34385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8505008288769766e-05,
+      "loss": 2.697,
+      "step": 34390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8494950773751775e-05,
+      "loss": 2.6688,
+      "step": 34395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.84848929047868e-05,
+      "loss": 2.8781,
+      "step": 34400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.84748346822939e-05,
+      "loss": 2.7091,
+      "step": 34405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8464776106692166e-05,
+      "loss": 2.6955,
+      "step": 34410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8454717178400664e-05,
+      "loss": 2.774,
+      "step": 34415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.844465789783854e-05,
+      "loss": 2.5788,
+      "step": 34420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.843459826542492e-05,
+      "loss": 2.7488,
+      "step": 34425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.842453828157893e-05,
+      "loss": 2.5475,
+      "step": 34430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8414477946719725e-05,
+      "loss": 2.8037,
+      "step": 34435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8404417261266485e-05,
+      "loss": 2.7406,
+      "step": 34440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.839435622563838e-05,
+      "loss": 2.7512,
+      "step": 34445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.83842948402546e-05,
+      "loss": 2.6854,
+      "step": 34450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.83742331055344e-05,
+      "loss": 2.7312,
+      "step": 34455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.836417102189699e-05,
+      "loss": 2.6598,
+      "step": 34460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.835410858976159e-05,
+      "loss": 2.7719,
+      "step": 34465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8344045809547476e-05,
+      "loss": 2.7551,
+      "step": 34470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.833398268167391e-05,
+      "loss": 2.8444,
+      "step": 34475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.832391920656018e-05,
+      "loss": 2.6785,
+      "step": 34480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8313855384625594e-05,
+      "loss": 2.7788,
+      "step": 34485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.830379121628946e-05,
+      "loss": 2.7053,
+      "step": 34490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.829372670197112e-05,
+      "loss": 2.7398,
+      "step": 34495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.828366184208991e-05,
+      "loss": 2.8075,
+      "step": 34500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.827359663706519e-05,
+      "loss": 2.7417,
+      "step": 34505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8263531087316304e-05,
+      "loss": 2.7223,
+      "step": 34510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.825346519326267e-05,
+      "loss": 2.8428,
+      "step": 34515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.824339895532369e-05,
+      "loss": 2.7788,
+      "step": 34520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.823333237391878e-05,
+      "loss": 2.6893,
+      "step": 34525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8223265449467354e-05,
+      "loss": 2.7946,
+      "step": 34530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8213198182388875e-05,
+      "loss": 2.6921,
+      "step": 34535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.820313057310277e-05,
+      "loss": 2.7802,
+      "step": 34540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.819306262202854e-05,
+      "loss": 2.8227,
+      "step": 34545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818299432958567e-05,
+      "loss": 2.681,
+      "step": 34550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817292569619366e-05,
+      "loss": 2.7891,
+      "step": 34555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816285672227202e-05,
+      "loss": 2.5976,
+      "step": 34560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815278740824027e-05,
+      "loss": 2.5626,
+      "step": 34565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8142717754517964e-05,
+      "loss": 2.6421,
+      "step": 34570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8132647761524675e-05,
+      "loss": 2.8545,
+      "step": 34575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.812257742967995e-05,
+      "loss": 2.7185,
+      "step": 34580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8112506759403384e-05,
+      "loss": 2.6645,
+      "step": 34585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.810243575111458e-05,
+      "loss": 2.684,
+      "step": 34590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.809236440523316e-05,
+      "loss": 2.6774,
+      "step": 34595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.808229272217874e-05,
+      "loss": 2.7415,
+      "step": 34600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.807222070237095e-05,
+      "loss": 2.7663,
+      "step": 34605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.806214834622947e-05,
+      "loss": 2.8021,
+      "step": 34610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.805207565417398e-05,
+      "loss": 2.5957,
+      "step": 34615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8042002626624135e-05,
+      "loss": 2.5778,
+      "step": 34620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.803192926399964e-05,
+      "loss": 2.6134,
+      "step": 34625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.802185556672023e-05,
+      "loss": 2.8222,
+      "step": 34630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.801178153520559e-05,
+      "loss": 2.7224,
+      "step": 34635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8001707169875497e-05,
+      "loss": 2.6487,
+      "step": 34640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.799163247114969e-05,
+      "loss": 2.6267,
+      "step": 34645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7981557439447953e-05,
+      "loss": 2.7702,
+      "step": 34650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.797148207519004e-05,
+      "loss": 2.6024,
+      "step": 34655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.796140637879578e-05,
+      "loss": 2.8044,
+      "step": 34660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.795133035068494e-05,
+      "loss": 2.7218,
+      "step": 34665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.794125399127739e-05,
+      "loss": 2.6023,
+      "step": 34670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.793117730099292e-05,
+      "loss": 2.6741,
+      "step": 34675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7921100280251425e-05,
+      "loss": 2.6127,
+      "step": 34680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7911022929472755e-05,
+      "loss": 2.6927,
+      "step": 34685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.790094524907679e-05,
+      "loss": 2.6928,
+      "step": 34690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7890867239483395e-05,
+      "loss": 2.5293,
+      "step": 34695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.78807889011125e-05,
+      "loss": 2.6613,
+      "step": 34700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.787071023438403e-05,
+      "loss": 2.7717,
+      "step": 34705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7860631239717925e-05,
+      "loss": 2.8209,
+      "step": 34710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.78505519175341e-05,
+      "loss": 2.5396,
+      "step": 34715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7840472268252556e-05,
+      "loss": 2.7731,
+      "step": 34720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7830392292293214e-05,
+      "loss": 2.6979,
+      "step": 34725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.782031199007611e-05,
+      "loss": 2.664,
+      "step": 34730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.781023136202123e-05,
+      "loss": 2.7133,
+      "step": 34735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.780015040854858e-05,
+      "loss": 2.7484,
+      "step": 34740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.779006913007821e-05,
+      "loss": 2.6549,
+      "step": 34745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7779987527030144e-05,
+      "loss": 2.7359,
+      "step": 34750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7769905599824434e-05,
+      "loss": 2.6227,
+      "step": 34755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7759823348881174e-05,
+      "loss": 2.6795,
+      "step": 34760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.774974077462041e-05,
+      "loss": 2.7452,
+      "step": 34765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.773965787746228e-05,
+      "loss": 2.6096,
+      "step": 34770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.772957465782687e-05,
+      "loss": 2.6725,
+      "step": 34775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7719491116134306e-05,
+      "loss": 2.6383,
+      "step": 34780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.77094072528047e-05,
+      "loss": 2.742,
+      "step": 34785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.769932306825826e-05,
+      "loss": 2.7193,
+      "step": 34790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7689238562915105e-05,
+      "loss": 2.7464,
+      "step": 34795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.767915373719541e-05,
+      "loss": 2.7261,
+      "step": 34800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.766906859151939e-05,
+      "loss": 2.809,
+      "step": 34805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7658983126307234e-05,
+      "loss": 2.6391,
+      "step": 34810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.764889734197916e-05,
+      "loss": 2.6619,
+      "step": 34815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.763881123895539e-05,
+      "loss": 2.8739,
+      "step": 34820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.762872481765619e-05,
+      "loss": 2.6691,
+      "step": 34825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.761863807850179e-05,
+      "loss": 2.5672,
+      "step": 34830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.760855102191247e-05,
+      "loss": 2.6226,
+      "step": 34835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7598463648308535e-05,
+      "loss": 2.5855,
+      "step": 34840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.758837595811025e-05,
+      "loss": 2.8127,
+      "step": 34845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.757828795173793e-05,
+      "loss": 2.7213,
+      "step": 34850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.756819962961191e-05,
+      "loss": 2.7384,
+      "step": 34855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.75581109921525e-05,
+      "loss": 2.7951,
+      "step": 34860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7548022039780095e-05,
+      "loss": 2.6391,
+      "step": 34865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.753793277291503e-05,
+      "loss": 2.7692,
+      "step": 34870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.752784319197766e-05,
+      "loss": 2.8477,
+      "step": 34875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.75177532973884e-05,
+      "loss": 2.6787,
+      "step": 34880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.750766308956764e-05,
+      "loss": 2.7492,
+      "step": 34885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.749757256893582e-05,
+      "loss": 2.711,
+      "step": 34890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.748748173591332e-05,
+      "loss": 2.7088,
+      "step": 34895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.747739059092062e-05,
+      "loss": 2.7593,
+      "step": 34900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.746729913437816e-05,
+      "loss": 2.7907,
+      "step": 34905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.745720736670639e-05,
+      "loss": 2.6938,
+      "step": 34910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.744711528832582e-05,
+      "loss": 2.6163,
+      "step": 34915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7437022899656935e-05,
+      "loss": 2.6395,
+      "step": 34920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.742693020112021e-05,
+      "loss": 2.7389,
+      "step": 34925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7416837193136194e-05,
+      "loss": 2.6677,
+      "step": 34930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7406743876125423e-05,
+      "loss": 2.7906,
+      "step": 34935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.739665025050841e-05,
+      "loss": 2.7287,
+      "step": 34940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7386556316705744e-05,
+      "loss": 2.7345,
+      "step": 34945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.737646207513797e-05,
+      "loss": 2.7558,
+      "step": 34950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.736636752622567e-05,
+      "loss": 2.6798,
+      "step": 34955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7356272670389466e-05,
+      "loss": 2.8905,
+      "step": 34960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.734617750804995e-05,
+      "loss": 2.6414,
+      "step": 34965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7336082039627724e-05,
+      "loss": 2.7221,
+      "step": 34970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7325986265543455e-05,
+      "loss": 2.7484,
+      "step": 34975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.731589018621777e-05,
+      "loss": 2.6233,
+      "step": 34980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.730579380207134e-05,
+      "loss": 2.6754,
+      "step": 34985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.72956971135248e-05,
+      "loss": 2.7224,
+      "step": 34990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.728560012099889e-05,
+      "loss": 2.5391,
+      "step": 34995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7275502824914264e-05,
+      "loss": 2.6134,
+      "step": 35000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.726540522569166e-05,
+      "loss": 2.7261,
+      "step": 35005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.725530732375178e-05,
+      "loss": 2.7739,
+      "step": 35010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.724520911951536e-05,
+      "loss": 2.6786,
+      "step": 35015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7235110613403165e-05,
+      "loss": 2.634,
+      "step": 35020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.722501180583594e-05,
+      "loss": 2.8237,
+      "step": 35025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.721491269723446e-05,
+      "loss": 2.6878,
+      "step": 35030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.72048132880195e-05,
+      "loss": 2.7658,
+      "step": 35035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.719471357861188e-05,
+      "loss": 2.7555,
+      "step": 35040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.718461356943241e-05,
+      "loss": 2.6761,
+      "step": 35045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.717451326090188e-05,
+      "loss": 2.649,
+      "step": 35050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.716441265344116e-05,
+      "loss": 2.7798,
+      "step": 35055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.715431174747108e-05,
+      "loss": 2.6452,
+      "step": 35060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7144210543412505e-05,
+      "loss": 2.649,
+      "step": 35065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.71341090416863e-05,
+      "loss": 2.7657,
+      "step": 35070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.712400724271337e-05,
+      "loss": 2.6546,
+      "step": 35075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7113905146914594e-05,
+      "loss": 2.5519,
+      "step": 35080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.710380275471089e-05,
+      "loss": 2.6759,
+      "step": 35085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.709370006652317e-05,
+      "loss": 2.7589,
+      "step": 35090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.708359708277238e-05,
+      "loss": 2.6941,
+      "step": 35095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7073493803879454e-05,
+      "loss": 2.6173,
+      "step": 35100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.706339023026537e-05,
+      "loss": 2.7411,
+      "step": 35105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.705328636235108e-05,
+      "loss": 2.5888,
+      "step": 35110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7043182200557575e-05,
+      "loss": 2.7591,
+      "step": 35115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7033077745305854e-05,
+      "loss": 2.6882,
+      "step": 35120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.702297299701691e-05,
+      "loss": 2.6847,
+      "step": 35125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7012867956111784e-05,
+      "loss": 2.9016,
+      "step": 35130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.700276262301151e-05,
+      "loss": 2.744,
+      "step": 35135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.699265699813712e-05,
+      "loss": 2.7307,
+      "step": 35140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.698255108190964e-05,
+      "loss": 2.5455,
+      "step": 35145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.697244487475021e-05,
+      "loss": 2.7966,
+      "step": 35150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6962338377079836e-05,
+      "loss": 2.7407,
+      "step": 35155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.695223158931968e-05,
+      "loss": 2.7795,
+      "step": 35160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6942124511890806e-05,
+      "loss": 2.7794,
+      "step": 35165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6932017145214335e-05,
+      "loss": 2.7912,
+      "step": 35170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6921909489711386e-05,
+      "loss": 2.7988,
+      "step": 35175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.691180154580313e-05,
+      "loss": 2.6469,
+      "step": 35180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.69016933139107e-05,
+      "loss": 2.7617,
+      "step": 35185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.689158479445528e-05,
+      "loss": 2.6789,
+      "step": 35190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.688147598785802e-05,
+      "loss": 2.6479,
+      "step": 35195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.687136689454014e-05,
+      "loss": 2.6569,
+      "step": 35200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.686125751492281e-05,
+      "loss": 2.7505,
+      "step": 35205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.685114784942724e-05,
+      "loss": 2.7373,
+      "step": 35210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.684103789847468e-05,
+      "loss": 2.6314,
+      "step": 35215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.683092766248637e-05,
+      "loss": 2.5645,
+      "step": 35220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.682081714188355e-05,
+      "loss": 2.6789,
+      "step": 35225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.681070633708747e-05,
+      "loss": 2.5961,
+      "step": 35230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.680059524851942e-05,
+      "loss": 2.6645,
+      "step": 35235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.679048387660066e-05,
+      "loss": 2.6707,
+      "step": 35240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.678037222175252e-05,
+      "loss": 2.6513,
+      "step": 35245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.677026028439627e-05,
+      "loss": 2.766,
+      "step": 35250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6760148064953255e-05,
+      "loss": 2.7558,
+      "step": 35255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.675003556384481e-05,
+      "loss": 2.7692,
+      "step": 35260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6739922781492264e-05,
+      "loss": 2.6314,
+      "step": 35265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6729809718316965e-05,
+      "loss": 2.6596,
+      "step": 35270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.671969637474029e-05,
+      "loss": 2.873,
+      "step": 35275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6709582751183635e-05,
+      "loss": 2.6128,
+      "step": 35280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.669946884806836e-05,
+      "loss": 2.7843,
+      "step": 35285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6689354665815887e-05,
+      "loss": 2.691,
+      "step": 35290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6679240204847614e-05,
+      "loss": 2.6862,
+      "step": 35295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.666912546558498e-05,
+      "loss": 2.5921,
+      "step": 35300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.665901044844941e-05,
+      "loss": 2.7435,
+      "step": 35305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.664889515386235e-05,
+      "loss": 2.7303,
+      "step": 35310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6638779582245284e-05,
+      "loss": 2.7806,
+      "step": 35315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.662866373401965e-05,
+      "loss": 2.6273,
+      "step": 35320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6618547609606965e-05,
+      "loss": 2.7211,
+      "step": 35325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.660843120942869e-05,
+      "loss": 2.7694,
+      "step": 35330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6598314533906335e-05,
+      "loss": 2.5682,
+      "step": 35335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6588197583461434e-05,
+      "loss": 2.6163,
+      "step": 35340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6578080358515516e-05,
+      "loss": 2.766,
+      "step": 35345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.656796285949011e-05,
+      "loss": 2.712,
+      "step": 35350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.655784508680677e-05,
+      "loss": 2.6182,
+      "step": 35355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.654772704088707e-05,
+      "loss": 2.5653,
+      "step": 35360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.653760872215255e-05,
+      "loss": 2.6248,
+      "step": 35365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.652749013102483e-05,
+      "loss": 2.6902,
+      "step": 35370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.65173712679255e-05,
+      "loss": 2.6867,
+      "step": 35375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6507252133276164e-05,
+      "loss": 2.6642,
+      "step": 35380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.649713272749844e-05,
+      "loss": 2.7799,
+      "step": 35385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.648701305101397e-05,
+      "loss": 2.5538,
+      "step": 35390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.647689310424437e-05,
+      "loss": 2.6153,
+      "step": 35395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.646677288761132e-05,
+      "loss": 2.7505,
+      "step": 35400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6456652401536456e-05,
+      "loss": 2.6165,
+      "step": 35405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6446531646441505e-05,
+      "loss": 2.7874,
+      "step": 35410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6436410622748104e-05,
+      "loss": 2.7102,
+      "step": 35415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.642628933087798e-05,
+      "loss": 2.759,
+      "step": 35420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6416167771252826e-05,
+      "loss": 2.7473,
+      "step": 35425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.640604594429435e-05,
+      "loss": 2.6289,
+      "step": 35430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6395923850424314e-05,
+      "loss": 2.6549,
+      "step": 35435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.638580149006445e-05,
+      "loss": 2.9211,
+      "step": 35440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.637567886363652e-05,
+      "loss": 2.5233,
+      "step": 35445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6365555971562266e-05,
+      "loss": 2.6262,
+      "step": 35450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6355432814263486e-05,
+      "loss": 2.8177,
+      "step": 35455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.634530939216195e-05,
+      "loss": 2.7949,
+      "step": 35460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.633518570567946e-05,
+      "loss": 2.6324,
+      "step": 35465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.632506175523784e-05,
+      "loss": 2.6976,
+      "step": 35470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6314937541258905e-05,
+      "loss": 2.6189,
+      "step": 35475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6304813064164486e-05,
+      "loss": 2.7692,
+      "step": 35480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.629468832437641e-05,
+      "loss": 2.749,
+      "step": 35485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.628456332231653e-05,
+      "loss": 2.6769,
+      "step": 35490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6274438058406745e-05,
+      "loss": 2.668,
+      "step": 35495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.626431253306888e-05,
+      "loss": 2.6575,
+      "step": 35500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6254186746724855e-05,
+      "loss": 2.731,
+      "step": 35505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6244060699796574e-05,
+      "loss": 2.677,
+      "step": 35510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6233934392705925e-05,
+      "loss": 2.7803,
+      "step": 35515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.622380782587482e-05,
+      "loss": 2.7661,
+      "step": 35520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.621368099972519e-05,
+      "loss": 2.76,
+      "step": 35525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.620355391467899e-05,
+      "loss": 2.603,
+      "step": 35530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.619342657115817e-05,
+      "loss": 2.6484,
+      "step": 35535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6183298969584685e-05,
+      "loss": 2.7116,
+      "step": 35540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.617317111038051e-05,
+      "loss": 2.6981,
+      "step": 35545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6163042993967616e-05,
+      "loss": 2.6752,
+      "step": 35550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.615291462076802e-05,
+      "loss": 2.7183,
+      "step": 35555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.614278599120368e-05,
+      "loss": 2.6665,
+      "step": 35560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.613265710569668e-05,
+      "loss": 2.6808,
+      "step": 35565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6122527964669e-05,
+      "loss": 2.7356,
+      "step": 35570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.611239856854269e-05,
+      "loss": 2.8128,
+      "step": 35575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.61022689177398e-05,
+      "loss": 2.7083,
+      "step": 35580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6092139012682365e-05,
+      "loss": 2.7408,
+      "step": 35585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6082008853792465e-05,
+      "loss": 2.7371,
+      "step": 35590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.60718784414922e-05,
+      "loss": 2.7093,
+      "step": 35595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6061747776203646e-05,
+      "loss": 2.7644,
+      "step": 35600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.605161685834889e-05,
+      "loss": 2.8105,
+      "step": 35605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.604148568835006e-05,
+      "loss": 2.7569,
+      "step": 35610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6031354266629256e-05,
+      "loss": 2.7031,
+      "step": 35615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6021222593608636e-05,
+      "loss": 2.7056,
+      "step": 35620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.601109066971031e-05,
+      "loss": 2.6886,
+      "step": 35625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.600095849535647e-05,
+      "loss": 2.7083,
+      "step": 35630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.599082607096926e-05,
+      "loss": 2.7167,
+      "step": 35635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.598069339697084e-05,
+      "loss": 2.6108,
+      "step": 35640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5970560473783406e-05,
+      "loss": 2.5964,
+      "step": 35645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.596042730182915e-05,
+      "loss": 2.733,
+      "step": 35650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.595029388153028e-05,
+      "loss": 2.7211,
+      "step": 35655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.594016021330901e-05,
+      "loss": 2.7559,
+      "step": 35660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.593002629758757e-05,
+      "loss": 2.9186,
+      "step": 35665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5919892134788175e-05,
+      "loss": 2.5881,
+      "step": 35670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5909757725333104e-05,
+      "loss": 2.7158,
+      "step": 35675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5899623069644566e-05,
+      "loss": 2.7372,
+      "step": 35680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.588948816814486e-05,
+      "loss": 2.7164,
+      "step": 35685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.587935302125626e-05,
+      "loss": 2.8207,
+      "step": 35690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.586921762940107e-05,
+      "loss": 2.7983,
+      "step": 35695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.585908199300155e-05,
+      "loss": 2.6789,
+      "step": 35700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.584894611248004e-05,
+      "loss": 2.6685,
+      "step": 35705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5838809988258814e-05,
+      "loss": 2.667,
+      "step": 35710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.582867362076024e-05,
+      "loss": 2.7457,
+      "step": 35715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5818537010406635e-05,
+      "loss": 2.7271,
+      "step": 35720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.580840015762037e-05,
+      "loss": 2.5823,
+      "step": 35725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.579826306282377e-05,
+      "loss": 2.6773,
+      "step": 35730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.578812572643923e-05,
+      "loss": 2.6761,
+      "step": 35735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5777988148889114e-05,
+      "loss": 2.7552,
+      "step": 35740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5767850330595816e-05,
+      "loss": 2.5895,
+      "step": 35745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.575771227198172e-05,
+      "loss": 2.6037,
+      "step": 35750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.574757397346926e-05,
+      "loss": 2.7708,
+      "step": 35755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5737435435480836e-05,
+      "loss": 2.7463,
+      "step": 35760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5727296658438886e-05,
+      "loss": 2.7052,
+      "step": 35765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5717157642765814e-05,
+      "loss": 2.7247,
+      "step": 35770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5707018388884125e-05,
+      "loss": 2.6701,
+      "step": 35775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.569687889721623e-05,
+      "loss": 2.6049,
+      "step": 35780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.568673916818462e-05,
+      "loss": 2.5696,
+      "step": 35785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.567659920221177e-05,
+      "loss": 2.6983,
+      "step": 35790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.566645899972016e-05,
+      "loss": 2.7707,
+      "step": 35795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5656318561132284e-05,
+      "loss": 2.7926,
+      "step": 35800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.564617788687067e-05,
+      "loss": 2.6356,
+      "step": 35805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.563603697735782e-05,
+      "loss": 2.7599,
+      "step": 35810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.562589583301624e-05,
+      "loss": 2.6963,
+      "step": 35815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.561575445426851e-05,
+      "loss": 2.6124,
+      "step": 35820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5605612841537155e-05,
+      "loss": 2.6798,
+      "step": 35825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.5595470995244724e-05,
+      "loss": 2.726,
+      "step": 35830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.558532891581379e-05,
+      "loss": 2.7737,
+      "step": 35835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.557518660366694e-05,
+      "loss": 2.6737,
+      "step": 35840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5565044059226734e-05,
+      "loss": 2.7641,
+      "step": 35845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.555490128291578e-05,
+      "loss": 2.7186,
+      "step": 35850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.55447582751567e-05,
+      "loss": 2.7758,
+      "step": 35855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.553461503637207e-05,
+      "loss": 2.8042,
+      "step": 35860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.552447156698455e-05,
+      "loss": 2.6913,
+      "step": 35865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.551432786741676e-05,
+      "loss": 2.8004,
+      "step": 35870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.550418393809132e-05,
+      "loss": 2.7814,
+      "step": 35875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.549403977943093e-05,
+      "loss": 2.701,
+      "step": 35880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.548389539185822e-05,
+      "loss": 2.695,
+      "step": 35885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.547375077579587e-05,
+      "loss": 2.6865,
+      "step": 35890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.546360593166655e-05,
+      "loss": 2.8029,
+      "step": 35895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.545346085989297e-05,
+      "loss": 2.6556,
+      "step": 35900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.544331556089781e-05,
+      "loss": 2.7764,
+      "step": 35905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.54331700351038e-05,
+      "loss": 2.7517,
+      "step": 35910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5423024282933657e-05,
+      "loss": 2.8769,
+      "step": 35915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.541287830481009e-05,
+      "loss": 2.6969,
+      "step": 35920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5402732101155844e-05,
+      "loss": 2.7356,
+      "step": 35925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.539258567239368e-05,
+      "loss": 2.7374,
+      "step": 35930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.538243901894635e-05,
+      "loss": 2.7181,
+      "step": 35935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.53722921412366e-05,
+      "loss": 2.7473,
+      "step": 35940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.536214503968724e-05,
+      "loss": 2.6582,
+      "step": 35945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.535199771472104e-05,
+      "loss": 2.7141,
+      "step": 35950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5341850166760766e-05,
+      "loss": 2.6036,
+      "step": 35955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.533170239622927e-05,
+      "loss": 2.8019,
+      "step": 35960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.532155440354932e-05,
+      "loss": 2.6231,
+      "step": 35965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.531140618914377e-05,
+      "loss": 2.634,
+      "step": 35970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.530125775343544e-05,
+      "loss": 2.7889,
+      "step": 35975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.529110909684718e-05,
+      "loss": 2.7398,
+      "step": 35980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5280960219801816e-05,
+      "loss": 2.5688,
+      "step": 35985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.527081112272222e-05,
+      "loss": 2.762,
+      "step": 35990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5260661806031265e-05,
+      "loss": 2.743,
+      "step": 35995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.525051227015182e-05,
+      "loss": 2.8429,
+      "step": 36000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.524036251550678e-05,
+      "loss": 2.5755,
+      "step": 36005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.523021254251903e-05,
+      "loss": 2.6912,
+      "step": 36010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.522006235161148e-05,
+      "loss": 2.72,
+      "step": 36015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.520991194320705e-05,
+      "loss": 2.7198,
+      "step": 36020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.519976131772866e-05,
+      "loss": 2.6085,
+      "step": 36025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.518961047559923e-05,
+      "loss": 2.701,
+      "step": 36030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.51794594172417e-05,
+      "loss": 2.7617,
+      "step": 36035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.516930814307903e-05,
+      "loss": 2.6993,
+      "step": 36040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5159156653534184e-05,
+      "loss": 2.9343,
+      "step": 36045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.514900494903013e-05,
+      "loss": 2.7078,
+      "step": 36050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5138853029989825e-05,
+      "loss": 2.7683,
+      "step": 36055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.512870089683626e-05,
+      "loss": 2.7957,
+      "step": 36060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.511854854999244e-05,
+      "loss": 2.5715,
+      "step": 36065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.510839598988137e-05,
+      "loss": 2.6061,
+      "step": 36070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.509824321692604e-05,
+      "loss": 2.6323,
+      "step": 36075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5088090231549505e-05,
+      "loss": 2.8246,
+      "step": 36080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5077937034174775e-05,
+      "loss": 2.6827,
+      "step": 36085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.506778362522489e-05,
+      "loss": 2.8788,
+      "step": 36090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5057630005122884e-05,
+      "loss": 2.6672,
+      "step": 36095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.504747617429185e-05,
+      "loss": 2.6908,
+      "step": 36100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.503732213315482e-05,
+      "loss": 2.623,
+      "step": 36105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5027167882134864e-05,
+      "loss": 2.5874,
+      "step": 36110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.50170134216551e-05,
+      "loss": 2.5625,
+      "step": 36115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.50068587521386e-05,
+      "loss": 2.7657,
+      "step": 36120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.499670387400846e-05,
+      "loss": 2.5999,
+      "step": 36125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4986548787687783e-05,
+      "loss": 2.5717,
+      "step": 36130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.497639349359971e-05,
+      "loss": 2.6538,
+      "step": 36135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.496623799216734e-05,
+      "loss": 2.6894,
+      "step": 36140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4956082283813816e-05,
+      "loss": 2.7571,
+      "step": 36145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4945926368962297e-05,
+      "loss": 2.5687,
+      "step": 36150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4935770248035934e-05,
+      "loss": 2.6463,
+      "step": 36155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.492561392145785e-05,
+      "loss": 2.8787,
+      "step": 36160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.491545738965126e-05,
+      "loss": 2.6363,
+      "step": 36165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.490530065303931e-05,
+      "loss": 2.7416,
+      "step": 36170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4895143712045216e-05,
+      "loss": 2.6968,
+      "step": 36175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.488498656709216e-05,
+      "loss": 2.7531,
+      "step": 36180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.487482921860333e-05,
+      "loss": 2.7158,
+      "step": 36185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4864671667001946e-05,
+      "loss": 2.6768,
+      "step": 36190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4854513912711235e-05,
+      "loss": 2.6831,
+      "step": 36195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4844355956154416e-05,
+      "loss": 2.7335,
+      "step": 36200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.483419779775474e-05,
+      "loss": 2.5516,
+      "step": 36205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.482403943793545e-05,
+      "loss": 2.544,
+      "step": 36210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4813880877119794e-05,
+      "loss": 2.8023,
+      "step": 36215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4803722115731025e-05,
+      "loss": 2.8026,
+      "step": 36220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4793563154192416e-05,
+      "loss": 2.7231,
+      "step": 36225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.478340399292726e-05,
+      "loss": 2.71,
+      "step": 36230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.477324463235886e-05,
+      "loss": 2.6788,
+      "step": 36235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.476308507291047e-05,
+      "loss": 2.7985,
+      "step": 36240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.475292531500541e-05,
+      "loss": 2.6417,
+      "step": 36245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.474276535906701e-05,
+      "loss": 2.6948,
+      "step": 36250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4732605205518564e-05,
+      "loss": 2.6939,
+      "step": 36255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.47224448547834e-05,
+      "loss": 2.7505,
+      "step": 36260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.471228430728489e-05,
+      "loss": 2.8248,
+      "step": 36265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.470212356344635e-05,
+      "loss": 2.7045,
+      "step": 36270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.469196262369114e-05,
+      "loss": 2.6586,
+      "step": 36275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.468180148844262e-05,
+      "loss": 2.6926,
+      "step": 36280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.467164015812415e-05,
+      "loss": 2.6099,
+      "step": 36285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.466147863315911e-05,
+      "loss": 2.7346,
+      "step": 36290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.465131691397092e-05,
+      "loss": 2.6401,
+      "step": 36295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4641155000982936e-05,
+      "loss": 2.7194,
+      "step": 36300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.463099289461857e-05,
+      "loss": 2.5858,
+      "step": 36305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.462083059530124e-05,
+      "loss": 2.6381,
+      "step": 36310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.461066810345434e-05,
+      "loss": 2.7144,
+      "step": 36315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4600505419501335e-05,
+      "loss": 2.7383,
+      "step": 36320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4590342543865625e-05,
+      "loss": 2.5883,
+      "step": 36325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4580179476970675e-05,
+      "loss": 2.5702,
+      "step": 36330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4570016219239914e-05,
+      "loss": 2.6268,
+      "step": 36335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.455985277109682e-05,
+      "loss": 2.7196,
+      "step": 36340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.454968913296484e-05,
+      "loss": 2.7639,
+      "step": 36345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.453952530526746e-05,
+      "loss": 2.7602,
+      "step": 36350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.452936128842815e-05,
+      "loss": 2.6001,
+      "step": 36355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.451919708287042e-05,
+      "loss": 2.6459,
+      "step": 36360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4509032689017755e-05,
+      "loss": 2.7111,
+      "step": 36365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4498868107293663e-05,
+      "loss": 2.6079,
+      "step": 36370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.448870333812166e-05,
+      "loss": 2.6682,
+      "step": 36375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4478538381925246e-05,
+      "loss": 2.7783,
+      "step": 36380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4468373239127967e-05,
+      "loss": 2.5903,
+      "step": 36385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.445820791015337e-05,
+      "loss": 2.749,
+      "step": 36390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.444804239542498e-05,
+      "loss": 2.837,
+      "step": 36395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.443787669536636e-05,
+      "loss": 2.5542,
+      "step": 36400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.442771081040108e-05,
+      "loss": 2.7124,
+      "step": 36405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4417544740952674e-05,
+      "loss": 2.7184,
+      "step": 36410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4407378487444723e-05,
+      "loss": 2.7195,
+      "step": 36415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.439721205030086e-05,
+      "loss": 2.7093,
+      "step": 36420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4387045429944616e-05,
+      "loss": 2.6185,
+      "step": 36425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4376878626799635e-05,
+      "loss": 2.6571,
+      "step": 36430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4366711641289484e-05,
+      "loss": 2.8991,
+      "step": 36435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.435654447383779e-05,
+      "loss": 2.6952,
+      "step": 36440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4346377124868176e-05,
+      "loss": 2.7355,
+      "step": 36445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.433620959480426e-05,
+      "loss": 2.6227,
+      "step": 36450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.43260418840697e-05,
+      "loss": 2.7031,
+      "step": 36455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.431587399308814e-05,
+      "loss": 2.6726,
+      "step": 36460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.43057059222832e-05,
+      "loss": 2.5585,
+      "step": 36465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.429553767207858e-05,
+      "loss": 2.7738,
+      "step": 36470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.42853692428979e-05,
+      "loss": 2.6926,
+      "step": 36475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4275200635164845e-05,
+      "loss": 2.6872,
+      "step": 36480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.426503184930314e-05,
+      "loss": 2.7696,
+      "step": 36485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.425486288573642e-05,
+      "loss": 2.7645,
+      "step": 36490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.424469374488842e-05,
+      "loss": 2.584,
+      "step": 36495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.423452442718282e-05,
+      "loss": 2.7226,
+      "step": 36500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.422435493304332e-05,
+      "loss": 2.5905,
+      "step": 36505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4214185262893656e-05,
+      "loss": 2.7286,
+      "step": 36510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.420401541715756e-05,
+      "loss": 2.6099,
+      "step": 36515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.419384539625875e-05,
+      "loss": 2.735,
+      "step": 36520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.418367520062097e-05,
+      "loss": 2.6803,
+      "step": 36525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4173504830667976e-05,
+      "loss": 2.6893,
+      "step": 36530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.41633342868235e-05,
+      "loss": 2.7521,
+      "step": 36535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4153163569511324e-05,
+      "loss": 2.6386,
+      "step": 36540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.41429926791552e-05,
+      "loss": 2.8829,
+      "step": 36545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.413282161617893e-05,
+      "loss": 2.8229,
+      "step": 36550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.412265038100627e-05,
+      "loss": 2.7065,
+      "step": 36555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.411247897406103e-05,
+      "loss": 2.572,
+      "step": 36560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4102307395766994e-05,
+      "loss": 2.769,
+      "step": 36565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.409213564654796e-05,
+      "loss": 2.7118,
+      "step": 36570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.408196372682776e-05,
+      "loss": 2.64,
+      "step": 36575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.407179163703021e-05,
+      "loss": 2.7125,
+      "step": 36580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.406161937757912e-05,
+      "loss": 2.8226,
+      "step": 36585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4051446948898344e-05,
+      "loss": 2.5265,
+      "step": 36590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4041274351411706e-05,
+      "loss": 2.7094,
+      "step": 36595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4031101585543056e-05,
+      "loss": 2.7207,
+      "step": 36600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.402092865171624e-05,
+      "loss": 2.6579,
+      "step": 36605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.401075555035515e-05,
+      "loss": 2.7113,
+      "step": 36610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.400058228188363e-05,
+      "loss": 2.5916,
+      "step": 36615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.399040884672555e-05,
+      "loss": 2.6352,
+      "step": 36620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.398023524530481e-05,
+      "loss": 2.6948,
+      "step": 36625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.397006147804528e-05,
+      "loss": 2.7841,
+      "step": 36630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3959887545370866e-05,
+      "loss": 2.8554,
+      "step": 36635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.394971344770546e-05,
+      "loss": 2.7061,
+      "step": 36640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3939539185472996e-05,
+      "loss": 2.8277,
+      "step": 36645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3929364759097375e-05,
+      "loss": 2.6149,
+      "step": 36650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.391919016900253e-05,
+      "loss": 2.7447,
+      "step": 36655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3909015415612365e-05,
+      "loss": 2.665,
+      "step": 36660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.389884049935084e-05,
+      "loss": 2.7289,
+      "step": 36665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.388866542064189e-05,
+      "loss": 2.7507,
+      "step": 36670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.387849017990948e-05,
+      "loss": 2.6533,
+      "step": 36675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.386831477757755e-05,
+      "loss": 2.7029,
+      "step": 36680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.385813921407008e-05,
+      "loss": 2.7986,
+      "step": 36685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3847963489811014e-05,
+      "loss": 2.7284,
+      "step": 36690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.383778760522435e-05,
+      "loss": 2.6689,
+      "step": 36695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.382761156073406e-05,
+      "loss": 2.7399,
+      "step": 36700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.381743535676416e-05,
+      "loss": 2.7282,
+      "step": 36705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.380725899373863e-05,
+      "loss": 2.5582,
+      "step": 36710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3797082472081464e-05,
+      "loss": 2.7263,
+      "step": 36715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.378690579221668e-05,
+      "loss": 2.8268,
+      "step": 36720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.377672895456831e-05,
+      "loss": 2.6039,
+      "step": 36725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.376655195956035e-05,
+      "loss": 2.724,
+      "step": 36730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.375637480761686e-05,
+      "loss": 2.7406,
+      "step": 36735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.374619749916185e-05,
+      "loss": 2.7586,
+      "step": 36740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.373602003461939e-05,
+      "loss": 2.7115,
+      "step": 36745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.372584241441351e-05,
+      "loss": 2.6209,
+      "step": 36750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.371566463896827e-05,
+      "loss": 2.589,
+      "step": 36755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.370548670870774e-05,
+      "loss": 2.619,
+      "step": 36760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.369530862405597e-05,
+      "loss": 2.7107,
+      "step": 36765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3685130385437065e-05,
+      "loss": 2.7865,
+      "step": 36770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.36749519932751e-05,
+      "loss": 2.7078,
+      "step": 36775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.366477344799414e-05,
+      "loss": 2.7271,
+      "step": 36780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3654594750018296e-05,
+      "loss": 2.6296,
+      "step": 36785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3644415899771684e-05,
+      "loss": 2.6154,
+      "step": 36790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.363423689767837e-05,
+      "loss": 2.6477,
+      "step": 36795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.362405774416253e-05,
+      "loss": 2.6092,
+      "step": 36800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.361387843964822e-05,
+      "loss": 2.6449,
+      "step": 36805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.360369898455961e-05,
+      "loss": 2.7612,
+      "step": 36810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.359351937932081e-05,
+      "loss": 2.7371,
+      "step": 36815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3583339624355974e-05,
+      "loss": 2.5569,
+      "step": 36820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.357315972008923e-05,
+      "loss": 2.6833,
+      "step": 36825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3562979666944755e-05,
+      "loss": 2.6794,
+      "step": 36830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.355279946534669e-05,
+      "loss": 2.7132,
+      "step": 36835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.354261911571919e-05,
+      "loss": 2.6889,
+      "step": 36840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.353243861848645e-05,
+      "loss": 2.6243,
+      "step": 36845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.352225797407262e-05,
+      "loss": 2.6821,
+      "step": 36850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.351207718290191e-05,
+      "loss": 2.6254,
+      "step": 36855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3501896245398475e-05,
+      "loss": 2.6829,
+      "step": 36860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3491715161986536e-05,
+      "loss": 2.6725,
+      "step": 36865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.348153393309029e-05,
+      "loss": 2.6594,
+      "step": 36870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.347135255913392e-05,
+      "loss": 2.687,
+      "step": 36875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3461171040541677e-05,
+      "loss": 2.6044,
+      "step": 36880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.345098937773776e-05,
+      "loss": 2.6441,
+      "step": 36885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.344080757114639e-05,
+      "loss": 2.6461,
+      "step": 36890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3430625621191796e-05,
+      "loss": 2.573,
+      "step": 36895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3420443528298236e-05,
+      "loss": 2.6969,
+      "step": 36900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.341026129288993e-05,
+      "loss": 2.6925,
+      "step": 36905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.340007891539114e-05,
+      "loss": 2.6206,
+      "step": 36910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.33898963962261e-05,
+      "loss": 2.7264,
+      "step": 36915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3379713735819096e-05,
+      "loss": 2.7634,
+      "step": 36920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.336953093459438e-05,
+      "loss": 2.7394,
+      "step": 36925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.335934799297624e-05,
+      "loss": 2.701,
+      "step": 36930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.334916491138893e-05,
+      "loss": 2.5104,
+      "step": 36935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.333898169025675e-05,
+      "loss": 2.773,
+      "step": 36940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3328798330003993e-05,
+      "loss": 2.7282,
+      "step": 36945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.331861483105494e-05,
+      "loss": 2.7263,
+      "step": 36950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.33084311938339e-05,
+      "loss": 2.7668,
+      "step": 36955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.329824741876518e-05,
+      "loss": 2.6943,
+      "step": 36960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.328806350627308e-05,
+      "loss": 2.7589,
+      "step": 36965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3277879456781954e-05,
+      "loss": 2.6214,
+      "step": 36970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.326769527071609e-05,
+      "loss": 2.6373,
+      "step": 36975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.325751094849985e-05,
+      "loss": 2.6177,
+      "step": 36980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.324732649055752e-05,
+      "loss": 2.7076,
+      "step": 36985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.323714189731349e-05,
+      "loss": 2.5949,
+      "step": 36990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.322695716919209e-05,
+      "loss": 2.6413,
+      "step": 36995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3216772306617665e-05,
+      "loss": 2.6559,
+      "step": 37000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.320658731001459e-05,
+      "loss": 2.7687,
+      "step": 37005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3196402179807215e-05,
+      "loss": 2.8514,
+      "step": 37010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.318621691641991e-05,
+      "loss": 2.7459,
+      "step": 37015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.317603152027705e-05,
+      "loss": 2.701,
+      "step": 37020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.316584599180303e-05,
+      "loss": 2.6394,
+      "step": 37025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.315566033142221e-05,
+      "loss": 2.6309,
+      "step": 37030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3145474539559e-05,
+      "loss": 2.6573,
+      "step": 37035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3135288616637803e-05,
+      "loss": 2.6059,
+      "step": 37040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3125102563082996e-05,
+      "loss": 2.7335,
+      "step": 37045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3114916379319005e-05,
+      "loss": 2.6604,
+      "step": 37050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.310473006577025e-05,
+      "loss": 2.5106,
+      "step": 37055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.309454362286111e-05,
+      "loss": 2.7188,
+      "step": 37060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.308435705101607e-05,
+      "loss": 2.684,
+      "step": 37065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3074170350659516e-05,
+      "loss": 2.5846,
+      "step": 37070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3063983522215896e-05,
+      "loss": 2.5764,
+      "step": 37075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.305379656610964e-05,
+      "loss": 2.6508,
+      "step": 37080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.304360948276521e-05,
+      "loss": 2.6984,
+      "step": 37085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.303342227260704e-05,
+      "loss": 2.6279,
+      "step": 37090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.30232349360596e-05,
+      "loss": 2.6513,
+      "step": 37095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.301304747354735e-05,
+      "loss": 2.7678,
+      "step": 37100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.300285988549475e-05,
+      "loss": 2.4848,
+      "step": 37105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2992672172326266e-05,
+      "loss": 2.855,
+      "step": 37110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.298248433446639e-05,
+      "loss": 2.6104,
+      "step": 37115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.297229637233958e-05,
+      "loss": 2.6548,
+      "step": 37120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.296210828637036e-05,
+      "loss": 2.525,
+      "step": 37125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.29519200769832e-05,
+      "loss": 2.5863,
+      "step": 37130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2941731744602594e-05,
+      "loss": 2.7746,
+      "step": 37135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.293154328965304e-05,
+      "loss": 2.7338,
+      "step": 37140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2921354712559066e-05,
+      "loss": 2.7068,
+      "step": 37145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.291116601374517e-05,
+      "loss": 2.7062,
+      "step": 37150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.290097719363589e-05,
+      "loss": 2.6478,
+      "step": 37155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.289078825265573e-05,
+      "loss": 2.6508,
+      "step": 37160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.288059919122922e-05,
+      "loss": 2.7985,
+      "step": 37165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.287041000978089e-05,
+      "loss": 2.6301,
+      "step": 37170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.286022070873528e-05,
+      "loss": 2.7073,
+      "step": 37175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.285003128851694e-05,
+      "loss": 2.707,
+      "step": 37180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.283984174955042e-05,
+      "loss": 2.7112,
+      "step": 37185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.282965209226026e-05,
+      "loss": 2.6552,
+      "step": 37190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.281946231707104e-05,
+      "loss": 2.5915,
+      "step": 37195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2809272424407294e-05,
+      "loss": 2.7298,
+      "step": 37200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2799082414693614e-05,
+      "loss": 2.815,
+      "step": 37205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.278889228835455e-05,
+      "loss": 2.7071,
+      "step": 37210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.27787020458147e-05,
+      "loss": 2.6405,
+      "step": 37215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.276851168749863e-05,
+      "loss": 2.8384,
+      "step": 37220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.275832121383095e-05,
+      "loss": 2.7139,
+      "step": 37225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.274813062523624e-05,
+      "loss": 2.6242,
+      "step": 37230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.273793992213907e-05,
+      "loss": 2.6166,
+      "step": 37235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.272774910496407e-05,
+      "loss": 2.6849,
+      "step": 37240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.271755817413585e-05,
+      "loss": 2.675,
+      "step": 37245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.270736713007902e-05,
+      "loss": 2.6569,
+      "step": 37250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.269717597321818e-05,
+      "loss": 2.6294,
+      "step": 37255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.268698470397797e-05,
+      "loss": 2.6985,
+      "step": 37260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.267679332278299e-05,
+      "loss": 2.677,
+      "step": 37265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.266660183005788e-05,
+      "loss": 2.6843,
+      "step": 37270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.265641022622728e-05,
+      "loss": 2.6891,
+      "step": 37275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2646218511715826e-05,
+      "loss": 2.636,
+      "step": 37280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.263602668694818e-05,
+      "loss": 2.7771,
+      "step": 37285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.262583475234897e-05,
+      "loss": 2.6466,
+      "step": 37290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2615642708342835e-05,
+      "loss": 2.6702,
+      "step": 37295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2605450555354455e-05,
+      "loss": 2.7454,
+      "step": 37300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.259525829380848e-05,
+      "loss": 2.6008,
+      "step": 37305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2585065924129585e-05,
+      "loss": 2.5979,
+      "step": 37310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2574873446742446e-05,
+      "loss": 2.6881,
+      "step": 37315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.256468086207173e-05,
+      "loss": 2.7827,
+      "step": 37320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2554488170542115e-05,
+      "loss": 2.6691,
+      "step": 37325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.254429537257829e-05,
+      "loss": 2.7034,
+      "step": 37330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2534102468604926e-05,
+      "loss": 2.7124,
+      "step": 37335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.252390945904675e-05,
+      "loss": 2.5887,
+      "step": 37340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2513716344328436e-05,
+      "loss": 2.6579,
+      "step": 37345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2503523124874685e-05,
+      "loss": 2.6404,
+      "step": 37350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.249332980111022e-05,
+      "loss": 2.6627,
+      "step": 37355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.248313637345974e-05,
+      "loss": 2.5959,
+      "step": 37360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.247294284234795e-05,
+      "loss": 2.6819,
+      "step": 37365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.246274920819958e-05,
+      "loss": 2.694,
+      "step": 37370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.245255547143937e-05,
+      "loss": 2.6769,
+      "step": 37375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.244236163249202e-05,
+      "loss": 2.7223,
+      "step": 37380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.243216769178228e-05,
+      "loss": 2.6466,
+      "step": 37385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.242197364973489e-05,
+      "loss": 2.7569,
+      "step": 37390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.241177950677456e-05,
+      "loss": 2.7272,
+      "step": 37395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.240158526332606e-05,
+      "loss": 2.8114,
+      "step": 37400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2391390919814145e-05,
+      "loss": 2.6251,
+      "step": 37405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.238119647666355e-05,
+      "loss": 2.6485,
+      "step": 37410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2371001934299045e-05,
+      "loss": 2.729,
+      "step": 37415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2360807293145385e-05,
+      "loss": 2.661,
+      "step": 37420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2350612553627334e-05,
+      "loss": 2.7313,
+      "step": 37425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.234041771616967e-05,
+      "loss": 2.7317,
+      "step": 37430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.233022278119716e-05,
+      "loss": 2.6801,
+      "step": 37435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.232002774913458e-05,
+      "loss": 2.6827,
+      "step": 37440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.230983262040672e-05,
+      "loss": 2.6422,
+      "step": 37445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.229963739543837e-05,
+      "loss": 2.6281,
+      "step": 37450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2289442074654295e-05,
+      "loss": 2.72,
+      "step": 37455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2279246658479305e-05,
+      "loss": 2.6604,
+      "step": 37460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.22690511473382e-05,
+      "loss": 2.6917,
+      "step": 37465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2258855541655797e-05,
+      "loss": 2.7116,
+      "step": 37470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.224865984185686e-05,
+      "loss": 2.786,
+      "step": 37475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2238464048366245e-05,
+      "loss": 2.6721,
+      "step": 37480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.222826816160874e-05,
+      "loss": 2.7782,
+      "step": 37485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.221807218200915e-05,
+      "loss": 2.6885,
+      "step": 37490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.220787610999232e-05,
+      "loss": 2.74,
+      "step": 37495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.219767994598307e-05,
+      "loss": 2.6778,
+      "step": 37500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.218748369040622e-05,
+      "loss": 2.6712,
+      "step": 37505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.217728734368662e-05,
+      "loss": 2.7639,
+      "step": 37510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.216709090624909e-05,
+      "loss": 2.7048,
+      "step": 37515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2156894378518484e-05,
+      "loss": 2.6719,
+      "step": 37520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.214669776091962e-05,
+      "loss": 2.6254,
+      "step": 37525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.213650105387738e-05,
+      "loss": 2.6847,
+      "step": 37530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.21263042578166e-05,
+      "loss": 2.6299,
+      "step": 37535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.211610737316214e-05,
+      "loss": 2.73,
+      "step": 37540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.210591040033885e-05,
+      "loss": 2.5954,
+      "step": 37545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.209571333977159e-05,
+      "loss": 2.6013,
+      "step": 37550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.208551619188524e-05,
+      "loss": 2.7027,
+      "step": 37555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.207531895710466e-05,
+      "loss": 2.7361,
+      "step": 37560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.206512163585473e-05,
+      "loss": 2.7356,
+      "step": 37565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.205492422856033e-05,
+      "loss": 2.5809,
+      "step": 37570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.204472673564633e-05,
+      "loss": 2.7091,
+      "step": 37575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2034529157537615e-05,
+      "loss": 2.6607,
+      "step": 37580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.202433149465908e-05,
+      "loss": 2.7305,
+      "step": 37585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.201413374743561e-05,
+      "loss": 2.6701,
+      "step": 37590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.200393591629212e-05,
+      "loss": 2.736,
+      "step": 37595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1993738001653483e-05,
+      "loss": 2.6559,
+      "step": 37600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.198354000394462e-05,
+      "loss": 2.6841,
+      "step": 37605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.19733419235904e-05,
+      "loss": 2.7214,
+      "step": 37610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1963143761015784e-05,
+      "loss": 2.651,
+      "step": 37615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1952945516645644e-05,
+      "loss": 2.6052,
+      "step": 37620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.194274719090493e-05,
+      "loss": 2.7444,
+      "step": 37625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1932548784218525e-05,
+      "loss": 2.5733,
+      "step": 37630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.192235029701138e-05,
+      "loss": 2.7121,
+      "step": 37635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.19121517297084e-05,
+      "loss": 2.7552,
+      "step": 37640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.190195308273452e-05,
+      "loss": 2.6449,
+      "step": 37645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1891754356514674e-05,
+      "loss": 2.6563,
+      "step": 37650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1881555551473816e-05,
+      "loss": 2.8096,
+      "step": 37655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1871356668036866e-05,
+      "loss": 2.6631,
+      "step": 37660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.186115770662876e-05,
+      "loss": 2.7162,
+      "step": 37665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1850958667674467e-05,
+      "loss": 2.6822,
+      "step": 37670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.184075955159891e-05,
+      "loss": 2.636,
+      "step": 37675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.183056035882704e-05,
+      "loss": 2.6795,
+      "step": 37680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.182036108978383e-05,
+      "loss": 2.7255,
+      "step": 37685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1810161744894246e-05,
+      "loss": 2.7308,
+      "step": 37690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.179996232458323e-05,
+      "loss": 2.6788,
+      "step": 37695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1789762829275755e-05,
+      "loss": 2.6177,
+      "step": 37700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.177956325939678e-05,
+      "loss": 2.6695,
+      "step": 37705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.176936361537129e-05,
+      "loss": 2.5963,
+      "step": 37710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.175916389762424e-05,
+      "loss": 2.7125,
+      "step": 37715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1748964106580635e-05,
+      "loss": 2.7103,
+      "step": 37720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.173876424266544e-05,
+      "loss": 2.6682,
+      "step": 37725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.172856430630363e-05,
+      "loss": 2.8003,
+      "step": 37730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1718364297920194e-05,
+      "loss": 2.6488,
+      "step": 37735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1708164217940136e-05,
+      "loss": 2.6563,
+      "step": 37740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.169796406678842e-05,
+      "loss": 2.6754,
+      "step": 37745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.168776384489007e-05,
+      "loss": 2.6933,
+      "step": 37750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1677563552670086e-05,
+      "loss": 2.5513,
+      "step": 37755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.166736319055344e-05,
+      "loss": 2.6758,
+      "step": 37760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.165716275896515e-05,
+      "loss": 2.8872,
+      "step": 37765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1646962258330234e-05,
+      "loss": 2.6556,
+      "step": 37770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.163676168907369e-05,
+      "loss": 2.6714,
+      "step": 37775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1626561051620524e-05,
+      "loss": 2.6812,
+      "step": 37780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1616360346395764e-05,
+      "loss": 2.8256,
+      "step": 37785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1606159573824444e-05,
+      "loss": 2.7161,
+      "step": 37790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.159595873433154e-05,
+      "loss": 2.6982,
+      "step": 37795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.158575782834211e-05,
+      "loss": 2.6956,
+      "step": 37800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.157555685628118e-05,
+      "loss": 2.5337,
+      "step": 37805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1565355818573754e-05,
+      "loss": 2.7376,
+      "step": 37810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1555154715644894e-05,
+      "loss": 2.7339,
+      "step": 37815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.154495354791963e-05,
+      "loss": 2.7892,
+      "step": 37820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.153475231582298e-05,
+      "loss": 2.7899,
+      "step": 37825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.152455101978e-05,
+      "loss": 2.6405,
+      "step": 37830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1514349660215734e-05,
+      "loss": 2.6078,
+      "step": 37835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1504148237555206e-05,
+      "loss": 2.7157,
+      "step": 37840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.149394675222349e-05,
+      "loss": 2.7186,
+      "step": 37845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1483745204645626e-05,
+      "loss": 2.5827,
+      "step": 37850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1473543595246676e-05,
+      "loss": 2.6247,
+      "step": 37855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.146334192445168e-05,
+      "loss": 2.6775,
+      "step": 37860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.145314019268571e-05,
+      "loss": 2.7314,
+      "step": 37865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.144293840037382e-05,
+      "loss": 2.8529,
+      "step": 37870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1432736547941065e-05,
+      "loss": 2.6678,
+      "step": 37875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.142253463581253e-05,
+      "loss": 2.6765,
+      "step": 37880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1412332664413284e-05,
+      "loss": 2.8318,
+      "step": 37885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.140213063416838e-05,
+      "loss": 2.6918,
+      "step": 37890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.13919285455029e-05,
+      "loss": 2.6327,
+      "step": 37895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1381726398841924e-05,
+      "loss": 2.6245,
+      "step": 37900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.137152419461051e-05,
+      "loss": 2.7926,
+      "step": 37905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.136132193323377e-05,
+      "loss": 2.6607,
+      "step": 37910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1351119615136776e-05,
+      "loss": 2.744,
+      "step": 37915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.13409172407446e-05,
+      "loss": 2.6028,
+      "step": 37920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.133071481048234e-05,
+      "loss": 2.7107,
+      "step": 37925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1320512324775096e-05,
+      "loss": 2.652,
+      "step": 37930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.131030978404794e-05,
+      "loss": 2.5905,
+      "step": 37935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1300107188725974e-05,
+      "loss": 2.7288,
+      "step": 37940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.128990453923431e-05,
+      "loss": 2.6038,
+      "step": 37945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.127970183599802e-05,
+      "loss": 2.6915,
+      "step": 37950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.126949907944223e-05,
+      "loss": 2.6838,
+      "step": 37955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.125929626999204e-05,
+      "loss": 2.5977,
+      "step": 37960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.124909340807253e-05,
+      "loss": 2.7352,
+      "step": 37965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.123889049410885e-05,
+      "loss": 2.8251,
+      "step": 37970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.122868752852607e-05,
+      "loss": 2.7498,
+      "step": 37975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.121848451174932e-05,
+      "loss": 2.5785,
+      "step": 37980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.120828144420373e-05,
+      "loss": 2.6768,
+      "step": 37985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.11980783263144e-05,
+      "loss": 2.6025,
+      "step": 37990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.118787515850645e-05,
+      "loss": 2.5983,
+      "step": 37995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.117767194120498e-05,
+      "loss": 2.7355,
+      "step": 38000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.116746867483516e-05,
+      "loss": 2.7684,
+      "step": 38005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.115726535982208e-05,
+      "loss": 2.8468,
+      "step": 38010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1147061996590875e-05,
+      "loss": 2.6494,
+      "step": 38015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1136858585566684e-05,
+      "loss": 2.6312,
+      "step": 38020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.112665512717463e-05,
+      "loss": 2.8275,
+      "step": 38025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.111645162183982e-05,
+      "loss": 2.722,
+      "step": 38030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.110624806998744e-05,
+      "loss": 2.6139,
+      "step": 38035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.109604447204258e-05,
+      "loss": 2.8009,
+      "step": 38040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.108584082843041e-05,
+      "loss": 2.7305,
+      "step": 38045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.107563713957607e-05,
+      "loss": 2.6555,
+      "step": 38050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1065433405904684e-05,
+      "loss": 2.7723,
+      "step": 38055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.105522962784139e-05,
+      "loss": 2.6352,
+      "step": 38060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1045025805811376e-05,
+      "loss": 2.7826,
+      "step": 38065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1034821940239754e-05,
+      "loss": 2.7608,
+      "step": 38070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1024618031551674e-05,
+      "loss": 2.665,
+      "step": 38075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1014414080172315e-05,
+      "loss": 2.7116,
+      "step": 38080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.10042100865268e-05,
+      "loss": 2.6588,
+      "step": 38085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.09940060510403e-05,
+      "loss": 2.6652,
+      "step": 38090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.098380197413796e-05,
+      "loss": 2.8051,
+      "step": 38095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.097359785624496e-05,
+      "loss": 2.7577,
+      "step": 38100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0963393697786435e-05,
+      "loss": 2.585,
+      "step": 38105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.095318949918757e-05,
+      "loss": 2.6788,
+      "step": 38110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.094298526087351e-05,
+      "loss": 2.7436,
+      "step": 38115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.093278098326945e-05,
+      "loss": 2.6444,
+      "step": 38120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0922576666800505e-05,
+      "loss": 2.6731,
+      "step": 38125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.09123723118919e-05,
+      "loss": 2.7697,
+      "step": 38130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.090216791896876e-05,
+      "loss": 2.8212,
+      "step": 38135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.089196348845629e-05,
+      "loss": 2.8321,
+      "step": 38140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.088175902077965e-05,
+      "loss": 2.6358,
+      "step": 38145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0871554516364014e-05,
+      "loss": 2.6506,
+      "step": 38150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0861349975634566e-05,
+      "loss": 2.7291,
+      "step": 38155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0851145399016474e-05,
+      "loss": 2.5713,
+      "step": 38160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.084094078693491e-05,
+      "loss": 2.7689,
+      "step": 38165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.083073613981508e-05,
+      "loss": 2.585,
+      "step": 38170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.082053145808216e-05,
+      "loss": 2.7016,
+      "step": 38175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0810326742161305e-05,
+      "loss": 2.7516,
+      "step": 38180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0800121992477746e-05,
+      "loss": 2.6533,
+      "step": 38185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.078991720945663e-05,
+      "loss": 2.5706,
+      "step": 38190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.077971239352315e-05,
+      "loss": 2.7523,
+      "step": 38195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0769507545102525e-05,
+      "loss": 2.5973,
+      "step": 38200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0759302664619936e-05,
+      "loss": 2.7408,
+      "step": 38205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.074909775250055e-05,
+      "loss": 2.6497,
+      "step": 38210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.073889280916959e-05,
+      "loss": 2.7125,
+      "step": 38215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.072868783505223e-05,
+      "loss": 2.7749,
+      "step": 38220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.071848283057366e-05,
+      "loss": 2.6858,
+      "step": 38225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.070827779615911e-05,
+      "loss": 2.6116,
+      "step": 38230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.069807273223376e-05,
+      "loss": 2.7341,
+      "step": 38235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.068786763922281e-05,
+      "loss": 2.6501,
+      "step": 38240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.067766251755146e-05,
+      "loss": 2.8152,
+      "step": 38245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.06674573676449e-05,
+      "loss": 2.7195,
+      "step": 38250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.065725218992835e-05,
+      "loss": 2.6479,
+      "step": 38255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.064704698482702e-05,
+      "loss": 2.6632,
+      "step": 38260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.06368417527661e-05,
+      "loss": 2.8818,
+      "step": 38265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.062663649417082e-05,
+      "loss": 2.7276,
+      "step": 38270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0616431209466366e-05,
+      "loss": 2.7497,
+      "step": 38275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.060622589907793e-05,
+      "loss": 2.7594,
+      "step": 38280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.059602056343077e-05,
+      "loss": 2.604,
+      "step": 38285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.058581520295006e-05,
+      "loss": 2.6551,
+      "step": 38290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0575609818061034e-05,
+      "loss": 2.7389,
+      "step": 38295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.056540440918889e-05,
+      "loss": 2.5367,
+      "step": 38300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.055519897675884e-05,
+      "loss": 2.8034,
+      "step": 38305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.054499352119613e-05,
+      "loss": 2.6726,
+      "step": 38310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.053478804292592e-05,
+      "loss": 2.6769,
+      "step": 38315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.052458254237347e-05,
+      "loss": 2.7851,
+      "step": 38320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0514377019964e-05,
+      "loss": 2.555,
+      "step": 38325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0504171476122706e-05,
+      "loss": 2.6843,
+      "step": 38330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.049396591127483e-05,
+      "loss": 2.6774,
+      "step": 38335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0483760325845575e-05,
+      "loss": 2.5473,
+      "step": 38340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.047355472026016e-05,
+      "loss": 2.8586,
+      "step": 38345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0463349094943824e-05,
+      "loss": 2.6875,
+      "step": 38350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0453143450321784e-05,
+      "loss": 2.7946,
+      "step": 38355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0442937786819265e-05,
+      "loss": 2.7215,
+      "step": 38360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.043273210486149e-05,
+      "loss": 2.5926,
+      "step": 38365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.042252640487369e-05,
+      "loss": 2.694,
+      "step": 38370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.041232068728107e-05,
+      "loss": 2.7268,
+      "step": 38375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0402114952508886e-05,
+      "loss": 2.6709,
+      "step": 38380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.039190920098234e-05,
+      "loss": 2.7173,
+      "step": 38385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.038170343312668e-05,
+      "loss": 2.5513,
+      "step": 38390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0371497649367125e-05,
+      "loss": 2.6936,
+      "step": 38395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.036129185012891e-05,
+      "loss": 2.6433,
+      "step": 38400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.035108603583727e-05,
+      "loss": 2.7055,
+      "step": 38405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.034088020691742e-05,
+      "loss": 2.8189,
+      "step": 38410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0330674363794584e-05,
+      "loss": 2.7501,
+      "step": 38415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.032046850689404e-05,
+      "loss": 2.7823,
+      "step": 38420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.031026263664098e-05,
+      "loss": 2.7451,
+      "step": 38425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.030005675346066e-05,
+      "loss": 2.7863,
+      "step": 38430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0289850857778296e-05,
+      "loss": 2.7643,
+      "step": 38435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.027964495001912e-05,
+      "loss": 2.6055,
+      "step": 38440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.026943903060838e-05,
+      "loss": 2.7162,
+      "step": 38445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.025923309997131e-05,
+      "loss": 2.6225,
+      "step": 38450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.024902715853316e-05,
+      "loss": 2.7966,
+      "step": 38455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0238821206719145e-05,
+      "loss": 2.5821,
+      "step": 38460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0228615244954505e-05,
+      "loss": 2.6732,
+      "step": 38465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.021840927366447e-05,
+      "loss": 2.6522,
+      "step": 38470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.020820329327429e-05,
+      "loss": 2.6877,
+      "step": 38475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.019799730420921e-05,
+      "loss": 2.7225,
+      "step": 38480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0187791306894446e-05,
+      "loss": 2.6968,
+      "step": 38485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.017758530175527e-05,
+      "loss": 2.6295,
+      "step": 38490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0167379289216885e-05,
+      "loss": 2.7436,
+      "step": 38495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.015717326970456e-05,
+      "loss": 2.725,
+      "step": 38500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.014696724364351e-05,
+      "loss": 2.7202,
+      "step": 38505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0136761211458974e-05,
+      "loss": 2.6836,
+      "step": 38510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.012655517357622e-05,
+      "loss": 2.7418,
+      "step": 38515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.011634913042047e-05,
+      "loss": 2.7408,
+      "step": 38520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0106143082416965e-05,
+      "loss": 2.7329,
+      "step": 38525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.009593702999096e-05,
+      "loss": 2.7459,
+      "step": 38530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.008573097356766e-05,
+      "loss": 2.7073,
+      "step": 38535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0075524913572336e-05,
+      "loss": 2.615,
+      "step": 38540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.006531885043023e-05,
+      "loss": 2.6978,
+      "step": 38545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0055112784566586e-05,
+      "loss": 2.6891,
+      "step": 38550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0044906716406625e-05,
+      "loss": 2.6141,
+      "step": 38555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0034700646375596e-05,
+      "loss": 2.7707,
+      "step": 38560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.002449457489875e-05,
+      "loss": 2.7145,
+      "step": 38565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.001428850240131e-05,
+      "loss": 2.6845,
+      "step": 38570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0004082429308555e-05,
+      "loss": 2.7916,
+      "step": 38575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9993876356045675e-05,
+      "loss": 2.714,
+      "step": 38580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9983670283037975e-05,
+      "loss": 2.7267,
+      "step": 38585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.997346421071063e-05,
+      "loss": 2.6187,
+      "step": 38590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.996325813948895e-05,
+      "loss": 2.7085,
+      "step": 38595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9953052069798113e-05,
+      "loss": 2.6969,
+      "step": 38600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.99428460020634e-05,
+      "loss": 2.5692,
+      "step": 38605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.993263993671002e-05,
+      "loss": 2.6609,
+      "step": 38610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.992243387416325e-05,
+      "loss": 2.7414,
+      "step": 38615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.991222781484833e-05,
+      "loss": 2.6481,
+      "step": 38620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.990202175919047e-05,
+      "loss": 2.6563,
+      "step": 38625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.989181570761494e-05,
+      "loss": 2.5918,
+      "step": 38630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.988160966054696e-05,
+      "loss": 2.5846,
+      "step": 38635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9871403618411775e-05,
+      "loss": 2.7353,
+      "step": 38640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.986119758163467e-05,
+      "loss": 2.5478,
+      "step": 38645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.985099155064081e-05,
+      "loss": 2.8408,
+      "step": 38650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.98407855258555e-05,
+      "loss": 2.649,
+      "step": 38655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.983057950770393e-05,
+      "loss": 2.7371,
+      "step": 38660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9820373496611366e-05,
+      "loss": 2.7922,
+      "step": 38665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.981016749300304e-05,
+      "loss": 2.7034,
+      "step": 38670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9799961497304195e-05,
+      "loss": 2.7976,
+      "step": 38675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9789755509940085e-05,
+      "loss": 2.7142,
+      "step": 38680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9779549531335915e-05,
+      "loss": 2.5769,
+      "step": 38685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.976934356191695e-05,
+      "loss": 2.8241,
+      "step": 38690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9759137602108414e-05,
+      "loss": 2.6464,
+      "step": 38695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.974893165233554e-05,
+      "loss": 2.6646,
+      "step": 38700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.973872571302358e-05,
+      "loss": 2.5765,
+      "step": 38705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.972851978459776e-05,
+      "loss": 2.7475,
+      "step": 38710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.971831386748333e-05,
+      "loss": 2.6976,
+      "step": 38715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.97081079621055e-05,
+      "loss": 2.733,
+      "step": 38720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9697902068889524e-05,
+      "loss": 2.6273,
+      "step": 38725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9687696188260626e-05,
+      "loss": 2.5719,
+      "step": 38730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9677490320644044e-05,
+      "loss": 2.6941,
+      "step": 38735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.966728446646503e-05,
+      "loss": 2.6819,
+      "step": 38740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.965707862614878e-05,
+      "loss": 2.6486,
+      "step": 38745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.964687280012057e-05,
+      "loss": 2.702,
+      "step": 38750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.963666698880558e-05,
+      "loss": 2.6679,
+      "step": 38755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.962646119262908e-05,
+      "loss": 2.7114,
+      "step": 38760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.961625541201628e-05,
+      "loss": 2.6558,
+      "step": 38765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.960604964739242e-05,
+      "loss": 2.6225,
+      "step": 38770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.959584389918274e-05,
+      "loss": 2.6118,
+      "step": 38775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9585638167812445e-05,
+      "loss": 2.7509,
+      "step": 38780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9575432453706775e-05,
+      "loss": 2.7273,
+      "step": 38785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.956522675729095e-05,
+      "loss": 2.6575,
+      "step": 38790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9555021078990206e-05,
+      "loss": 2.6126,
+      "step": 38795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.954481541922977e-05,
+      "loss": 2.6557,
+      "step": 38800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.953460977843484e-05,
+      "loss": 2.7182,
+      "step": 38805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.952440415703069e-05,
+      "loss": 2.7319,
+      "step": 38810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.951419855544249e-05,
+      "loss": 2.7143,
+      "step": 38815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.950399297409549e-05,
+      "loss": 2.7436,
+      "step": 38820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.94937874134149e-05,
+      "loss": 2.781,
+      "step": 38825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.948358187382595e-05,
+      "loss": 2.6824,
+      "step": 38830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.947337635575388e-05,
+      "loss": 2.6812,
+      "step": 38835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9463170859623864e-05,
+      "loss": 2.6196,
+      "step": 38840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.945296538586116e-05,
+      "loss": 2.6604,
+      "step": 38845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.944275993489094e-05,
+      "loss": 2.7465,
+      "step": 38850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.943255450713846e-05,
+      "loss": 2.7565,
+      "step": 38855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.942234910302892e-05,
+      "loss": 2.7041,
+      "step": 38860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9412143722987525e-05,
+      "loss": 2.6415,
+      "step": 38865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.940193836743953e-05,
+      "loss": 2.7499,
+      "step": 38870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.939173303681009e-05,
+      "loss": 2.7289,
+      "step": 38875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.938152773152444e-05,
+      "loss": 2.6289,
+      "step": 38880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937132245200779e-05,
+      "loss": 2.5719,
+      "step": 38885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9361117198685344e-05,
+      "loss": 2.6441,
+      "step": 38890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935091197198235e-05,
+      "loss": 2.6883,
+      "step": 38895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934070677232395e-05,
+      "loss": 2.6481,
+      "step": 38900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.93305016001354e-05,
+      "loss": 2.7323,
+      "step": 38905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.932029645584187e-05,
+      "loss": 2.7615,
+      "step": 38910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931009133986857e-05,
+      "loss": 2.7794,
+      "step": 38915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.929988625264072e-05,
+      "loss": 2.7732,
+      "step": 38920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.928968119458351e-05,
+      "loss": 2.7215,
+      "step": 38925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.927947616612215e-05,
+      "loss": 2.8087,
+      "step": 38930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9269271167681814e-05,
+      "loss": 2.6532,
+      "step": 38935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.925906619968773e-05,
+      "loss": 2.6419,
+      "step": 38940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.924886126256506e-05,
+      "loss": 2.7539,
+      "step": 38945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.923865635673901e-05,
+      "loss": 2.5701,
+      "step": 38950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.922845148263479e-05,
+      "loss": 2.5613,
+      "step": 38955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9218246640677576e-05,
+      "loss": 2.7355,
+      "step": 38960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.920804183129258e-05,
+      "loss": 2.7582,
+      "step": 38965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9197837054904946e-05,
+      "loss": 2.703,
+      "step": 38970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918763231193991e-05,
+      "loss": 2.6963,
+      "step": 38975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.917742760282263e-05,
+      "loss": 2.802,
+      "step": 38980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9167222927978295e-05,
+      "loss": 2.5852,
+      "step": 38985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9157018287832115e-05,
+      "loss": 2.6794,
+      "step": 38990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9146813682809235e-05,
+      "loss": 2.7219,
+      "step": 38995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.913660911333487e-05,
+      "loss": 2.6465,
+      "step": 39000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9126404579834156e-05,
+      "loss": 2.6534,
+      "step": 39005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.91162000827323e-05,
+      "loss": 2.491,
+      "step": 39010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.910599562245448e-05,
+      "loss": 2.6458,
+      "step": 39015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.909579119942586e-05,
+      "loss": 2.5638,
+      "step": 39020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9085586814071634e-05,
+      "loss": 2.7585,
+      "step": 39025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.907538246681693e-05,
+      "loss": 2.6742,
+      "step": 39030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9065178158086976e-05,
+      "loss": 2.6951,
+      "step": 39035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.905497388830689e-05,
+      "loss": 2.7549,
+      "step": 39040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9044769657901844e-05,
+      "loss": 2.7632,
+      "step": 39045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9034565467297036e-05,
+      "loss": 2.7597,
+      "step": 39050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9024361316917605e-05,
+      "loss": 2.7866,
+      "step": 39055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.901415720718873e-05,
+      "loss": 2.6893,
+      "step": 39060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9003953138535544e-05,
+      "loss": 2.573,
+      "step": 39065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8993749111383225e-05,
+      "loss": 2.6458,
+      "step": 39070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8983545126156924e-05,
+      "loss": 2.6303,
+      "step": 39075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8973341183281795e-05,
+      "loss": 2.6687,
+      "step": 39080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8963137283183016e-05,
+      "loss": 2.5801,
+      "step": 39085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.895293342628569e-05,
+      "loss": 2.5899,
+      "step": 39090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.894272961301502e-05,
+      "loss": 2.7859,
+      "step": 39095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8932525843796104e-05,
+      "loss": 2.6987,
+      "step": 39100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8922322119054104e-05,
+      "loss": 2.7837,
+      "step": 39105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.891211843921418e-05,
+      "loss": 2.5807,
+      "step": 39110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.890191480470145e-05,
+      "loss": 2.6006,
+      "step": 39115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8891711215941104e-05,
+      "loss": 2.6711,
+      "step": 39120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8881507673358215e-05,
+      "loss": 2.5849,
+      "step": 39125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.887130417737795e-05,
+      "loss": 2.6004,
+      "step": 39130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8861100728425464e-05,
+      "loss": 2.7231,
+      "step": 39135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8850897326925846e-05,
+      "loss": 2.6798,
+      "step": 39140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8840693973304264e-05,
+      "loss": 2.5543,
+      "step": 39145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8830490667985814e-05,
+      "loss": 2.6806,
+      "step": 39150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8820287411395664e-05,
+      "loss": 2.6946,
+      "step": 39155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.881008420395889e-05,
+      "loss": 2.7002,
+      "step": 39160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.879988104610065e-05,
+      "loss": 2.6518,
+      "step": 39165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.878967793824605e-05,
+      "loss": 2.4773,
+      "step": 39170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.877947488082021e-05,
+      "loss": 2.7526,
+      "step": 39175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8769271874248266e-05,
+      "loss": 2.6137,
+      "step": 39180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.87590689189553e-05,
+      "loss": 2.7416,
+      "step": 39185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.874886601536645e-05,
+      "loss": 2.7898,
+      "step": 39190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.87386631639068e-05,
+      "loss": 2.7209,
+      "step": 39195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.872846036500146e-05,
+      "loss": 2.7061,
+      "step": 39200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.871825761907557e-05,
+      "loss": 2.6792,
+      "step": 39205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.870805492655419e-05,
+      "loss": 2.7236,
+      "step": 39210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.869785228786247e-05,
+      "loss": 2.627,
+      "step": 39215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8687649703425455e-05,
+      "loss": 2.7577,
+      "step": 39220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.867744717366827e-05,
+      "loss": 2.6196,
+      "step": 39225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.866724469901602e-05,
+      "loss": 2.7417,
+      "step": 39230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8657042279893764e-05,
+      "loss": 2.5952,
+      "step": 39235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8646839916726625e-05,
+      "loss": 2.6364,
+      "step": 39240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.863663760993966e-05,
+      "loss": 2.7538,
+      "step": 39245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8626435359957996e-05,
+      "loss": 2.6361,
+      "step": 39250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.861623316720666e-05,
+      "loss": 2.6544,
+      "step": 39255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.860603103211077e-05,
+      "loss": 2.6813,
+      "step": 39260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.859582895509541e-05,
+      "loss": 2.6977,
+      "step": 39265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.858562693658562e-05,
+      "loss": 2.6119,
+      "step": 39270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8575424977006516e-05,
+      "loss": 2.8372,
+      "step": 39275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.856522307678313e-05,
+      "loss": 2.6706,
+      "step": 39280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8555021236340545e-05,
+      "loss": 2.6933,
+      "step": 39285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.854481945610384e-05,
+      "loss": 2.7037,
+      "step": 39290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.853461773649805e-05,
+      "loss": 2.6198,
+      "step": 39295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8524416077948264e-05,
+      "loss": 2.6487,
+      "step": 39300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.851421448087952e-05,
+      "loss": 2.6519,
+      "step": 39305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8504012945716895e-05,
+      "loss": 2.6985,
+      "step": 39310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8493811472885406e-05,
+      "loss": 2.7341,
+      "step": 39315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.848361006281013e-05,
+      "loss": 2.7534,
+      "step": 39320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.847340871591612e-05,
+      "loss": 2.6679,
+      "step": 39325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.846320743262839e-05,
+      "loss": 2.6473,
+      "step": 39330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8453006213372026e-05,
+      "loss": 2.8296,
+      "step": 39335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.844280505857202e-05,
+      "loss": 2.7738,
+      "step": 39340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.843260396865344e-05,
+      "loss": 2.7497,
+      "step": 39345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.842240294404133e-05,
+      "loss": 2.6845,
+      "step": 39350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.841220198516069e-05,
+      "loss": 2.7556,
+      "step": 39355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.840200109243657e-05,
+      "loss": 2.6467,
+      "step": 39360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.839180026629398e-05,
+      "loss": 2.608,
+      "step": 39365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.838159950715797e-05,
+      "loss": 2.749,
+      "step": 39370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.837139881545352e-05,
+      "loss": 2.6548,
+      "step": 39375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.836119819160568e-05,
+      "loss": 2.7194,
+      "step": 39380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8350997636039476e-05,
+      "loss": 2.7513,
+      "step": 39385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.834079714917987e-05,
+      "loss": 2.6889,
+      "step": 39390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.833059673145192e-05,
+      "loss": 2.7705,
+      "step": 39395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8320396383280594e-05,
+      "loss": 2.7525,
+      "step": 39400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8310196105090944e-05,
+      "loss": 2.5449,
+      "step": 39405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.829999589730791e-05,
+      "loss": 2.7291,
+      "step": 39410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.828979576035653e-05,
+      "loss": 2.6686,
+      "step": 39415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8279595694661814e-05,
+      "loss": 2.6854,
+      "step": 39420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8269395700648695e-05,
+      "loss": 2.6537,
+      "step": 39425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.825919577874221e-05,
+      "loss": 2.5738,
+      "step": 39430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.824899592936732e-05,
+      "loss": 2.6471,
+      "step": 39435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8238796152949016e-05,
+      "loss": 2.7424,
+      "step": 39440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.822859644991231e-05,
+      "loss": 2.6584,
+      "step": 39445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.821839682068211e-05,
+      "loss": 2.7549,
+      "step": 39450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.820819726568344e-05,
+      "loss": 2.6984,
+      "step": 39455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8197997785341254e-05,
+      "loss": 2.7239,
+      "step": 39460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.818779838008054e-05,
+      "loss": 2.6879,
+      "step": 39465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.817759905032622e-05,
+      "loss": 2.7938,
+      "step": 39470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8167399796503284e-05,
+      "loss": 2.6615,
+      "step": 39475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.81572006190367e-05,
+      "loss": 2.7661,
+      "step": 39480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8147001518351395e-05,
+      "loss": 2.5856,
+      "step": 39485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8136802494872333e-05,
+      "loss": 2.6381,
+      "step": 39490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8126603549024465e-05,
+      "loss": 2.7233,
+      "step": 39495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8116404681232716e-05,
+      "loss": 2.6345,
+      "step": 39500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.810620589192208e-05,
+      "loss": 2.675,
+      "step": 39505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8096007181517433e-05,
+      "loss": 2.7,
+      "step": 39510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8085808550443746e-05,
+      "loss": 2.7056,
+      "step": 39515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.807560999912594e-05,
+      "loss": 2.7507,
+      "step": 39520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8065411527988954e-05,
+      "loss": 2.6622,
+      "step": 39525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.805521313745768e-05,
+      "loss": 2.7322,
+      "step": 39530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.804501482795708e-05,
+      "loss": 2.6875,
+      "step": 39535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.803481659991207e-05,
+      "loss": 2.7302,
+      "step": 39540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.802461845374753e-05,
+      "loss": 2.6607,
+      "step": 39545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8014420389888405e-05,
+      "loss": 2.8398,
+      "step": 39550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.800422240875958e-05,
+      "loss": 2.5871,
+      "step": 39555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.799402451078597e-05,
+      "loss": 2.7232,
+      "step": 39560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7983826696392506e-05,
+      "loss": 2.6988,
+      "step": 39565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7973628966004025e-05,
+      "loss": 2.721,
+      "step": 39570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7963431320045484e-05,
+      "loss": 2.8313,
+      "step": 39575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7953233758941714e-05,
+      "loss": 2.7553,
+      "step": 39580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.794303628311764e-05,
+      "loss": 2.5696,
+      "step": 39585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.793283889299814e-05,
+      "loss": 2.7562,
+      "step": 39590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7922641589008077e-05,
+      "loss": 2.7156,
+      "step": 39595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7912444371572374e-05,
+      "loss": 2.7297,
+      "step": 39600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.790224724111584e-05,
+      "loss": 2.7465,
+      "step": 39605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.789205019806339e-05,
+      "loss": 2.6843,
+      "step": 39610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.788185324283987e-05,
+      "loss": 2.6429,
+      "step": 39615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.787165637587017e-05,
+      "loss": 2.6737,
+      "step": 39620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.78614595975791e-05,
+      "loss": 2.6515,
+      "step": 39625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7851262908391536e-05,
+      "loss": 2.6839,
+      "step": 39630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.784106630873236e-05,
+      "loss": 2.7118,
+      "step": 39635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.783086979902637e-05,
+      "loss": 2.5615,
+      "step": 39640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.782067337969844e-05,
+      "loss": 2.6645,
+      "step": 39645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.78104770511734e-05,
+      "loss": 2.6448,
+      "step": 39650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.780028081387608e-05,
+      "loss": 2.7114,
+      "step": 39655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.779008466823134e-05,
+      "loss": 2.5256,
+      "step": 39660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.777988861466397e-05,
+      "loss": 2.5989,
+      "step": 39665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.776969265359883e-05,
+      "loss": 2.5597,
+      "step": 39670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7759496785460704e-05,
+      "loss": 2.5873,
+      "step": 39675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7749301010674434e-05,
+      "loss": 2.6725,
+      "step": 39680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7739105329664815e-05,
+      "loss": 2.6648,
+      "step": 39685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.772890974285668e-05,
+      "loss": 2.5495,
+      "step": 39690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7718714250674826e-05,
+      "loss": 2.645,
+      "step": 39695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.770851885354404e-05,
+      "loss": 2.6867,
+      "step": 39700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.769832355188913e-05,
+      "loss": 2.7296,
+      "step": 39705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.768812834613488e-05,
+      "loss": 2.6623,
+      "step": 39710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.76779332367061e-05,
+      "loss": 2.7191,
+      "step": 39715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.766773822402755e-05,
+      "loss": 2.7271,
+      "step": 39720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.765754330852402e-05,
+      "loss": 2.5381,
+      "step": 39725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.764734849062031e-05,
+      "loss": 2.8971,
+      "step": 39730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7637153770741154e-05,
+      "loss": 2.7293,
+      "step": 39735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.762695914931135e-05,
+      "loss": 2.66,
+      "step": 39740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7616764626755635e-05,
+      "loss": 2.7836,
+      "step": 39745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.760657020349879e-05,
+      "loss": 2.6861,
+      "step": 39750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.759637587996559e-05,
+      "loss": 2.7842,
+      "step": 39755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.758618165658074e-05,
+      "loss": 2.864,
+      "step": 39760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.757598753376904e-05,
+      "loss": 2.6285,
+      "step": 39765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.756579351195518e-05,
+      "loss": 2.7557,
+      "step": 39770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7555599591563934e-05,
+      "loss": 2.7017,
+      "step": 39775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7545405773020036e-05,
+      "loss": 2.6739,
+      "step": 39780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7535212056748204e-05,
+      "loss": 2.6856,
+      "step": 39785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7525018443173194e-05,
+      "loss": 2.5697,
+      "step": 39790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.751482493271968e-05,
+      "loss": 2.7223,
+      "step": 39795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.750463152581242e-05,
+      "loss": 2.7321,
+      "step": 39800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.749443822287612e-05,
+      "loss": 2.8539,
+      "step": 39805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.748424502433547e-05,
+      "loss": 2.7094,
+      "step": 39810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7474051930615224e-05,
+      "loss": 2.6827,
+      "step": 39815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7463858942140026e-05,
+      "loss": 2.7676,
+      "step": 39820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.745366605933462e-05,
+      "loss": 2.7871,
+      "step": 39825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.744347328262365e-05,
+      "loss": 2.7194,
+      "step": 39830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743328061243184e-05,
+      "loss": 2.5988,
+      "step": 39835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.742308804918385e-05,
+      "loss": 2.7756,
+      "step": 39840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.741289559330438e-05,
+      "loss": 2.6891,
+      "step": 39845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.740270324521812e-05,
+      "loss": 2.7489,
+      "step": 39850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7392511005349685e-05,
+      "loss": 2.7184,
+      "step": 39855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7382318874123796e-05,
+      "loss": 2.6973,
+      "step": 39860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.737212685196507e-05,
+      "loss": 2.5369,
+      "step": 39865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.736193493929818e-05,
+      "loss": 2.7326,
+      "step": 39870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7351743136547796e-05,
+      "loss": 2.6707,
+      "step": 39875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.734155144413854e-05,
+      "loss": 2.6817,
+      "step": 39880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7331359862495087e-05,
+      "loss": 2.6429,
+      "step": 39885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.732116839204203e-05,
+      "loss": 2.7062,
+      "step": 39890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.731097703320404e-05,
+      "loss": 2.7657,
+      "step": 39895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.730078578640572e-05,
+      "loss": 2.7521,
+      "step": 39900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.72905946520717e-05,
+      "loss": 2.7661,
+      "step": 39905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.728040363062664e-05,
+      "loss": 2.7454,
+      "step": 39910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.727021272249508e-05,
+      "loss": 2.6592,
+      "step": 39915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.726002192810171e-05,
+      "loss": 2.6537,
+      "step": 39920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7249831247871054e-05,
+      "loss": 2.8062,
+      "step": 39925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.723964068222777e-05,
+      "loss": 2.733,
+      "step": 39930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.722945023159644e-05,
+      "loss": 2.5844,
+      "step": 39935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7219259896401644e-05,
+      "loss": 2.62,
+      "step": 39940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7209069677068004e-05,
+      "loss": 2.5994,
+      "step": 39945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.719887957402004e-05,
+      "loss": 2.7632,
+      "step": 39950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.71886895876824e-05,
+      "loss": 2.6671,
+      "step": 39955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.717849971847958e-05,
+      "loss": 2.7033,
+      "step": 39960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.71683099668362e-05,
+      "loss": 2.6918,
+      "step": 39965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.715812033317681e-05,
+      "loss": 2.6607,
+      "step": 39970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.714793081792595e-05,
+      "loss": 2.681,
+      "step": 39975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7137741421508206e-05,
+      "loss": 2.7192,
+      "step": 39980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.712755214434809e-05,
+      "loss": 2.6156,
+      "step": 39985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.711736298687016e-05,
+      "loss": 2.7131,
+      "step": 39990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.710717394949896e-05,
+      "loss": 2.7092,
+      "step": 39995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7096985032658994e-05,
+      "loss": 2.6775,
+      "step": 40000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.708679623677484e-05,
+      "loss": 2.7368,
+      "step": 40005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.707660756227096e-05,
+      "loss": 2.7369,
+      "step": 40010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.706641900957193e-05,
+      "loss": 2.6542,
+      "step": 40015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.70562305791022e-05,
+      "loss": 2.6564,
+      "step": 40020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.704604227128632e-05,
+      "loss": 2.6547,
+      "step": 40025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7035854086548785e-05,
+      "loss": 2.6673,
+      "step": 40030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.702566602531408e-05,
+      "loss": 2.6975,
+      "step": 40035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.701547808800672e-05,
+      "loss": 2.7033,
+      "step": 40040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7005290275051156e-05,
+      "loss": 2.7461,
+      "step": 40045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6995102586871906e-05,
+      "loss": 2.6213,
+      "step": 40050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.69849150238934e-05,
+      "loss": 2.6894,
+      "step": 40055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.697472758654015e-05,
+      "loss": 2.58,
+      "step": 40060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.696454027523661e-05,
+      "loss": 2.5824,
+      "step": 40065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.695435309040723e-05,
+      "loss": 2.7713,
+      "step": 40070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6944166032476494e-05,
+      "loss": 2.6192,
+      "step": 40075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.69339791018688e-05,
+      "loss": 2.6221,
+      "step": 40080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.692379229900863e-05,
+      "loss": 2.7146,
+      "step": 40085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.691360562432043e-05,
+      "loss": 2.8088,
+      "step": 40090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.69034190782286e-05,
+      "loss": 2.752,
+      "step": 40095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.689323266115761e-05,
+      "loss": 2.6765,
+      "step": 40100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6883046373531844e-05,
+      "loss": 2.7838,
+      "step": 40105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6872860215775755e-05,
+      "loss": 2.6048,
+      "step": 40110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.686267418831371e-05,
+      "loss": 2.7978,
+      "step": 40115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6852488291570146e-05,
+      "loss": 2.5563,
+      "step": 40120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.684230252596947e-05,
+      "loss": 2.5787,
+      "step": 40125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.683211689193605e-05,
+      "loss": 2.7429,
+      "step": 40130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6821931389894325e-05,
+      "loss": 2.6992,
+      "step": 40135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.681174602026862e-05,
+      "loss": 2.5686,
+      "step": 40140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6801560783483345e-05,
+      "loss": 2.608,
+      "step": 40145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.679137567996289e-05,
+      "loss": 2.7194,
+      "step": 40150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6781190710131584e-05,
+      "loss": 2.5845,
+      "step": 40155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6771005874413826e-05,
+      "loss": 2.6366,
+      "step": 40160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.676082117323394e-05,
+      "loss": 2.6763,
+      "step": 40165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.675063660701632e-05,
+      "loss": 2.775,
+      "step": 40170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6740452176185265e-05,
+      "loss": 2.7237,
+      "step": 40175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6730267881165134e-05,
+      "loss": 2.6799,
+      "step": 40180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.672008372238028e-05,
+      "loss": 2.5874,
+      "step": 40185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6709899700254994e-05,
+      "loss": 2.6124,
+      "step": 40190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.669971581521365e-05,
+      "loss": 2.8744,
+      "step": 40195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6689532067680516e-05,
+      "loss": 2.7269,
+      "step": 40200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.667934845807992e-05,
+      "loss": 2.5925,
+      "step": 40205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6669164986836194e-05,
+      "loss": 2.6083,
+      "step": 40210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.66589816543736e-05,
+      "loss": 2.5713,
+      "step": 40215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.664879846111646e-05,
+      "loss": 2.7413,
+      "step": 40220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.663861540748903e-05,
+      "loss": 2.774,
+      "step": 40225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.662843249391564e-05,
+      "loss": 2.7985,
+      "step": 40230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.661824972082052e-05,
+      "loss": 2.757,
+      "step": 40235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.660806708862796e-05,
+      "loss": 2.6154,
+      "step": 40240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6597884597762257e-05,
+      "loss": 2.5326,
+      "step": 40245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6587702248647606e-05,
+      "loss": 2.726,
+      "step": 40250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.657752004170831e-05,
+      "loss": 2.5921,
+      "step": 40255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.656733797736859e-05,
+      "loss": 2.6501,
+      "step": 40260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6557156056052725e-05,
+      "loss": 2.5733,
+      "step": 40265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6546974278184894e-05,
+      "loss": 2.6611,
+      "step": 40270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.653679264418937e-05,
+      "loss": 2.6546,
+      "step": 40275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6526611154490366e-05,
+      "loss": 2.7065,
+      "step": 40280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6516429809512085e-05,
+      "loss": 2.6891,
+      "step": 40285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6506248609678774e-05,
+      "loss": 2.6424,
+      "step": 40290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6496067555414594e-05,
+      "loss": 2.6887,
+      "step": 40295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6485886647143766e-05,
+      "loss": 2.5798,
+      "step": 40300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6475705885290505e-05,
+      "loss": 2.6035,
+      "step": 40305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6465525270278956e-05,
+      "loss": 2.5614,
+      "step": 40310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.645534480253333e-05,
+      "loss": 2.7489,
+      "step": 40315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6445164482477775e-05,
+      "loss": 2.7394,
+      "step": 40320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.64349843105365e-05,
+      "loss": 2.7012,
+      "step": 40325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.642480428713363e-05,
+      "loss": 2.8228,
+      "step": 40330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6414624412693344e-05,
+      "loss": 2.6948,
+      "step": 40335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6404444687639784e-05,
+      "loss": 2.5926,
+      "step": 40340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.639426511239709e-05,
+      "loss": 2.7412,
+      "step": 40345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6384085687389416e-05,
+      "loss": 2.6779,
+      "step": 40350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.637390641304087e-05,
+      "loss": 2.7995,
+      "step": 40355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.636372728977558e-05,
+      "loss": 2.7528,
+      "step": 40360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.635354831801771e-05,
+      "loss": 2.6721,
+      "step": 40365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.634336949819131e-05,
+      "loss": 2.6845,
+      "step": 40370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.633319083072052e-05,
+      "loss": 2.7778,
+      "step": 40375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6323012316029425e-05,
+      "loss": 2.7218,
+      "step": 40380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.631283395454215e-05,
+      "loss": 2.5613,
+      "step": 40385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.630265574668274e-05,
+      "loss": 2.4549,
+      "step": 40390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.629247769287529e-05,
+      "loss": 2.6217,
+      "step": 40395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.628229979354391e-05,
+      "loss": 2.6309,
+      "step": 40400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.627212204911261e-05,
+      "loss": 2.7107,
+      "step": 40405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.626194446000549e-05,
+      "loss": 2.7924,
+      "step": 40410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6251767026646585e-05,
+      "loss": 2.6402,
+      "step": 40415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6241589749459954e-05,
+      "loss": 2.6008,
+      "step": 40420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6231412628869656e-05,
+      "loss": 2.605,
+      "step": 40425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.622123566529969e-05,
+      "loss": 2.5791,
+      "step": 40430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6211058859174115e-05,
+      "loss": 2.6575,
+      "step": 40435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.620088221091694e-05,
+      "loss": 2.6898,
+      "step": 40440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6190705720952186e-05,
+      "loss": 2.7462,
+      "step": 40445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.618052938970385e-05,
+      "loss": 2.5989,
+      "step": 40450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.617035321759593e-05,
+      "loss": 2.6453,
+      "step": 40455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6160177205052476e-05,
+      "loss": 2.6598,
+      "step": 40460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.615000135249741e-05,
+      "loss": 2.614,
+      "step": 40465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.613982566035475e-05,
+      "loss": 2.7671,
+      "step": 40470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6129650129048454e-05,
+      "loss": 2.5393,
+      "step": 40475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.611947475900252e-05,
+      "loss": 2.6073,
+      "step": 40480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.610929955064088e-05,
+      "loss": 2.7175,
+      "step": 40485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.609912450438749e-05,
+      "loss": 2.7094,
+      "step": 40490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6088949620666346e-05,
+      "loss": 2.6814,
+      "step": 40495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.607877489990132e-05,
+      "loss": 2.7023,
+      "step": 40500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.60686003425164e-05,
+      "loss": 2.6123,
+      "step": 40505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.60584259489355e-05,
+      "loss": 2.4918,
+      "step": 40510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6048251719582526e-05,
+      "loss": 2.7112,
+      "step": 40515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.603807765488143e-05,
+      "loss": 2.7587,
+      "step": 40520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6027903755256075e-05,
+      "loss": 2.6649,
+      "step": 40525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6017730021130396e-05,
+      "loss": 2.6851,
+      "step": 40530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6007556452928266e-05,
+      "loss": 2.7335,
+      "step": 40535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.59973830510736e-05,
+      "loss": 2.6296,
+      "step": 40540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.598720981599024e-05,
+      "loss": 2.8007,
+      "step": 40545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.597703674810209e-05,
+      "loss": 2.7114,
+      "step": 40550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.596686384783302e-05,
+      "loss": 2.6398,
+      "step": 40555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5956691115606853e-05,
+      "loss": 2.6261,
+      "step": 40560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.594651855184748e-05,
+      "loss": 2.4754,
+      "step": 40565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.593634615697873e-05,
+      "loss": 2.6351,
+      "step": 40570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.592617393142443e-05,
+      "loss": 2.7685,
+      "step": 40575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.591600187560846e-05,
+      "loss": 2.7382,
+      "step": 40580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.590582998995459e-05,
+      "loss": 2.7213,
+      "step": 40585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.589565827488667e-05,
+      "loss": 2.7542,
+      "step": 40590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.588548673082848e-05,
+      "loss": 2.5797,
+      "step": 40595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.587531535820385e-05,
+      "loss": 2.7495,
+      "step": 40600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.586514415743656e-05,
+      "loss": 2.6377,
+      "step": 40605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.58549731289504e-05,
+      "loss": 2.5726,
+      "step": 40610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.584480227316919e-05,
+      "loss": 2.5697,
+      "step": 40615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5834631590516644e-05,
+      "loss": 2.5537,
+      "step": 40620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5824461081416564e-05,
+      "loss": 2.6218,
+      "step": 40625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5814290746292695e-05,
+      "loss": 2.6364,
+      "step": 40630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.58041205855688e-05,
+      "loss": 2.7491,
+      "step": 40635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5793950599668656e-05,
+      "loss": 2.7782,
+      "step": 40640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.578378078901593e-05,
+      "loss": 2.7174,
+      "step": 40645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.577361115403443e-05,
+      "loss": 2.6202,
+      "step": 40650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5763441695147805e-05,
+      "loss": 2.7283,
+      "step": 40655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.575327241277983e-05,
+      "loss": 2.7012,
+      "step": 40660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.574310330735417e-05,
+      "loss": 2.6148,
+      "step": 40665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5732934379294554e-05,
+      "loss": 2.6064,
+      "step": 40670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.57227656290247e-05,
+      "loss": 2.6548,
+      "step": 40675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.571259705696823e-05,
+      "loss": 2.5215,
+      "step": 40680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.570242866354889e-05,
+      "loss": 2.7455,
+      "step": 40685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.569226044919029e-05,
+      "loss": 2.7104,
+      "step": 40690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.568209241431614e-05,
+      "loss": 2.576,
+      "step": 40695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5671924559350077e-05,
+      "loss": 2.7575,
+      "step": 40700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5661756884715746e-05,
+      "loss": 2.7038,
+      "step": 40705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.565158939083682e-05,
+      "loss": 2.6997,
+      "step": 40710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.56414220781369e-05,
+      "loss": 2.6585,
+      "step": 40715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.563125494703963e-05,
+      "loss": 2.7509,
+      "step": 40720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5621087997968615e-05,
+      "loss": 2.6382,
+      "step": 40725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.561092123134747e-05,
+      "loss": 2.7011,
+      "step": 40730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.560075464759983e-05,
+      "loss": 2.7158,
+      "step": 40735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.559058824714925e-05,
+      "loss": 2.7685,
+      "step": 40740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5580422030419354e-05,
+      "loss": 2.6806,
+      "step": 40745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5570255997833675e-05,
+      "loss": 2.6326,
+      "step": 40750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.556009014981584e-05,
+      "loss": 2.6758,
+      "step": 40755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5549924486789375e-05,
+      "loss": 2.7549,
+      "step": 40760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5539759009177855e-05,
+      "loss": 2.6203,
+      "step": 40765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.552959371740485e-05,
+      "loss": 2.7931,
+      "step": 40770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5519428611893857e-05,
+      "loss": 2.777,
+      "step": 40775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.550926369306847e-05,
+      "loss": 2.6594,
+      "step": 40780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5499098961352147e-05,
+      "loss": 2.6297,
+      "step": 40785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.548893441716845e-05,
+      "loss": 2.6884,
+      "step": 40790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.547877006094089e-05,
+      "loss": 2.696,
+      "step": 40795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.546860589309296e-05,
+      "loss": 2.7932,
+      "step": 40800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.545844191404818e-05,
+      "loss": 2.626,
+      "step": 40805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5448278124229995e-05,
+      "loss": 2.6387,
+      "step": 40810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.543811452406191e-05,
+      "loss": 2.5528,
+      "step": 40815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.54279511139674e-05,
+      "loss": 2.6858,
+      "step": 40820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.541778789436992e-05,
+      "loss": 2.6983,
+      "step": 40825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.540762486569295e-05,
+      "loss": 2.5872,
+      "step": 40830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.53974620283599e-05,
+      "loss": 2.5728,
+      "step": 40835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.538729938279425e-05,
+      "loss": 2.6256,
+      "step": 40840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.53771369294194e-05,
+      "loss": 2.684,
+      "step": 40845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5366974668658776e-05,
+      "loss": 2.7631,
+      "step": 40850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.535681260093582e-05,
+      "loss": 2.638,
+      "step": 40855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.534665072667391e-05,
+      "loss": 2.554,
+      "step": 40860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.533648904629649e-05,
+      "loss": 2.6516,
+      "step": 40865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5326327560226894e-05,
+      "loss": 2.6708,
+      "step": 40870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.531616626888857e-05,
+      "loss": 2.7192,
+      "step": 40875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.530600517270483e-05,
+      "loss": 2.6444,
+      "step": 40880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5295844272099076e-05,
+      "loss": 2.7129,
+      "step": 40885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5285683567494666e-05,
+      "loss": 2.5254,
+      "step": 40890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.527552305931495e-05,
+      "loss": 2.7848,
+      "step": 40895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.526536274798329e-05,
+      "loss": 2.7228,
+      "step": 40900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.525520263392298e-05,
+      "loss": 2.5427,
+      "step": 40905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.524504271755738e-05,
+      "loss": 2.5527,
+      "step": 40910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5234882999309786e-05,
+      "loss": 2.5974,
+      "step": 40915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.522472347960352e-05,
+      "loss": 2.7232,
+      "step": 40920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5214564158861904e-05,
+      "loss": 2.7166,
+      "step": 40925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5204405037508195e-05,
+      "loss": 2.6182,
+      "step": 40930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5194246115965716e-05,
+      "loss": 2.7191,
+      "step": 40935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5184087394657706e-05,
+      "loss": 2.7243,
+      "step": 40940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.517392887400745e-05,
+      "loss": 2.6187,
+      "step": 40945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.516377055443822e-05,
+      "loss": 2.6953,
+      "step": 40950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.515361243637325e-05,
+      "loss": 2.6172,
+      "step": 40955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.514345452023581e-05,
+      "loss": 2.7962,
+      "step": 40960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.51332968064491e-05,
+      "loss": 2.633,
+      "step": 40965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.512313929543639e-05,
+      "loss": 2.8299,
+      "step": 40970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.511298198762085e-05,
+      "loss": 2.6574,
+      "step": 40975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.510282488342571e-05,
+      "loss": 2.7156,
+      "step": 40980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.509266798327419e-05,
+      "loss": 2.5773,
+      "step": 40985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5082511287589455e-05,
+      "loss": 2.6529,
+      "step": 40990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.507235479679473e-05,
+      "loss": 2.6867,
+      "step": 40995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.506219851131314e-05,
+      "loss": 2.7165,
+      "step": 41000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.505204243156787e-05,
+      "loss": 2.6918,
+      "step": 41005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.50418865579821e-05,
+      "loss": 2.7659,
+      "step": 41010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5031730890978954e-05,
+      "loss": 2.5957,
+      "step": 41015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.502157543098161e-05,
+      "loss": 2.7767,
+      "step": 41020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5011420178413154e-05,
+      "loss": 2.6512,
+      "step": 41025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5001265133696754e-05,
+      "loss": 2.7066,
+      "step": 41030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4991110297255484e-05,
+      "loss": 2.7906,
+      "step": 41035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.498095566951247e-05,
+      "loss": 2.8326,
+      "step": 41040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.497080125089082e-05,
+      "loss": 2.7052,
+      "step": 41045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4960647041813605e-05,
+      "loss": 2.6656,
+      "step": 41050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.495049304270394e-05,
+      "loss": 2.6027,
+      "step": 41055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4940339253984856e-05,
+      "loss": 2.6056,
+      "step": 41060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4930185676079436e-05,
+      "loss": 2.5981,
+      "step": 41065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4920032309410745e-05,
+      "loss": 2.8947,
+      "step": 41070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4909879154401805e-05,
+      "loss": 2.796,
+      "step": 41075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.489972621147567e-05,
+      "loss": 2.6942,
+      "step": 41080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4889573481055345e-05,
+      "loss": 2.6942,
+      "step": 41085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.48794209635639e-05,
+      "loss": 2.741,
+      "step": 41090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4869268659424284e-05,
+      "loss": 2.7757,
+      "step": 41095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.485911656905953e-05,
+      "loss": 2.7165,
+      "step": 41100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4848964692892634e-05,
+      "loss": 2.6141,
+      "step": 41105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.483881303134657e-05,
+      "loss": 2.5563,
+      "step": 41110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.482866158484434e-05,
+      "loss": 2.6249,
+      "step": 41115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.481851035380886e-05,
+      "loss": 2.5946,
+      "step": 41120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.480835933866313e-05,
+      "loss": 2.7015,
+      "step": 41125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.479820853983006e-05,
+      "loss": 2.6571,
+      "step": 41130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.478805795773261e-05,
+      "loss": 2.6157,
+      "step": 41135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.477790759279373e-05,
+      "loss": 2.6612,
+      "step": 41140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.47677574454363e-05,
+      "loss": 2.7108,
+      "step": 41145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4757607516083276e-05,
+      "loss": 2.6447,
+      "step": 41150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4747457805157516e-05,
+      "loss": 2.6721,
+      "step": 41155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4737308313081935e-05,
+      "loss": 2.6606,
+      "step": 41160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.472715904027943e-05,
+      "loss": 2.6724,
+      "step": 41165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.471700998717285e-05,
+      "loss": 2.6824,
+      "step": 41170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4706861154185086e-05,
+      "loss": 2.7929,
+      "step": 41175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.469671254173897e-05,
+      "loss": 2.7808,
+      "step": 41180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.468656415025738e-05,
+      "loss": 2.587,
+      "step": 41185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.467641598016312e-05,
+      "loss": 2.7419,
+      "step": 41190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4666268031879036e-05,
+      "loss": 2.6942,
+      "step": 41195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4656120305827956e-05,
+      "loss": 2.5492,
+      "step": 41200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.464597280243268e-05,
+      "loss": 2.6506,
+      "step": 41205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.463582552211603e-05,
+      "loss": 2.7334,
+      "step": 41210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.462567846530075e-05,
+      "loss": 2.8211,
+      "step": 41215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.461553163240967e-05,
+      "loss": 2.7183,
+      "step": 41220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.460538502386556e-05,
+      "loss": 2.5912,
+      "step": 41225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.459523864009115e-05,
+      "loss": 2.7729,
+      "step": 41230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4585092481509234e-05,
+      "loss": 2.6763,
+      "step": 41235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4574946548542526e-05,
+      "loss": 2.5306,
+      "step": 41240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.45648008416138e-05,
+      "loss": 2.6049,
+      "step": 41245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.455465536114573e-05,
+      "loss": 2.6687,
+      "step": 41250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.454451010756107e-05,
+      "loss": 2.6032,
+      "step": 41255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4534365081282524e-05,
+      "loss": 2.632,
+      "step": 41260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.452422028273279e-05,
+      "loss": 2.6981,
+      "step": 41265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4514075712334556e-05,
+      "loss": 2.5791,
+      "step": 41270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.450393137051049e-05,
+      "loss": 2.6479,
+      "step": 41275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.449378725768326e-05,
+      "loss": 2.725,
+      "step": 41280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.448364337427557e-05,
+      "loss": 2.566,
+      "step": 41285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.447349972071e-05,
+      "loss": 2.7091,
+      "step": 41290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.446335629740926e-05,
+      "loss": 2.6204,
+      "step": 41295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4453213104795925e-05,
+      "loss": 2.677,
+      "step": 41300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.444307014329267e-05,
+      "loss": 2.7037,
+      "step": 41305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.443292741332206e-05,
+      "loss": 2.5997,
+      "step": 41310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.442278491530671e-05,
+      "loss": 2.7214,
+      "step": 41315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4412642649669246e-05,
+      "loss": 2.7388,
+      "step": 41320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.44025006168322e-05,
+      "loss": 2.7127,
+      "step": 41325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4392358817218184e-05,
+      "loss": 2.5858,
+      "step": 41330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4382217251249735e-05,
+      "loss": 2.7211,
+      "step": 41335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.437207591934945e-05,
+      "loss": 2.7083,
+      "step": 41340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.436193482193981e-05,
+      "loss": 2.6656,
+      "step": 41345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.435179395944339e-05,
+      "loss": 2.5763,
+      "step": 41350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.434165333228273e-05,
+      "loss": 2.5754,
+      "step": 41355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.43315129408803e-05,
+      "loss": 2.5515,
+      "step": 41360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4321372785658663e-05,
+      "loss": 2.6897,
+      "step": 41365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.431123286704025e-05,
+      "loss": 2.674,
+      "step": 41370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.430109318544757e-05,
+      "loss": 2.6266,
+      "step": 41375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4290953741303135e-05,
+      "loss": 2.7047,
+      "step": 41380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.428081453502936e-05,
+      "loss": 2.588,
+      "step": 41385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4270675567048735e-05,
+      "loss": 2.6875,
+      "step": 41390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4260536837783686e-05,
+      "loss": 2.7401,
+      "step": 41395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.425039834765667e-05,
+      "loss": 2.7142,
+      "step": 41400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.424026009709008e-05,
+      "loss": 2.724,
+      "step": 41405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4230122086506345e-05,
+      "loss": 2.7098,
+      "step": 41410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.421998431632791e-05,
+      "loss": 2.7422,
+      "step": 41415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.420984678697711e-05,
+      "loss": 2.7671,
+      "step": 41420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4199709498876366e-05,
+      "loss": 2.6105,
+      "step": 41425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.418957245244805e-05,
+      "loss": 2.6927,
+      "step": 41430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.417943564811451e-05,
+      "loss": 2.8575,
+      "step": 41435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4169299086298144e-05,
+      "loss": 2.7772,
+      "step": 41440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.415916276742125e-05,
+      "loss": 2.5957,
+      "step": 41445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.414902669190619e-05,
+      "loss": 2.735,
+      "step": 41450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.413889086017528e-05,
+      "loss": 2.6303,
+      "step": 41455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.412875527265085e-05,
+      "loss": 2.591,
+      "step": 41460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.411861992975518e-05,
+      "loss": 2.5904,
+      "step": 41465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.410848483191057e-05,
+      "loss": 2.6445,
+      "step": 41470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4098349979539344e-05,
+      "loss": 2.6316,
+      "step": 41475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.408821537306372e-05,
+      "loss": 2.7254,
+      "step": 41480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4078081012906e-05,
+      "loss": 2.7047,
+      "step": 41485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.406794689948841e-05,
+      "loss": 2.6262,
+      "step": 41490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.405781303323321e-05,
+      "loss": 2.7006,
+      "step": 41495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4047679414562655e-05,
+      "loss": 2.6944,
+      "step": 41500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4037546043898925e-05,
+      "loss": 2.5966,
+      "step": 41505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.402741292166427e-05,
+      "loss": 2.6593,
+      "step": 41510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.401728004828086e-05,
+      "loss": 2.6276,
+      "step": 41515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4007147424170914e-05,
+      "loss": 2.6731,
+      "step": 41520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.399701504975659e-05,
+      "loss": 2.6494,
+      "step": 41525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.398688292546007e-05,
+      "loss": 2.6298,
+      "step": 41530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3976751051703536e-05,
+      "loss": 2.5927,
+      "step": 41535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.39666194289091e-05,
+      "loss": 2.6427,
+      "step": 41540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3956488057498926e-05,
+      "loss": 2.6652,
+      "step": 41545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.394635693789513e-05,
+      "loss": 2.6369,
+      "step": 41550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3936226070519856e-05,
+      "loss": 2.7424,
+      "step": 41555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.392609545579518e-05,
+      "loss": 2.5895,
+      "step": 41560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3915965094143215e-05,
+      "loss": 2.6968,
+      "step": 41565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.390583498598607e-05,
+      "loss": 2.6828,
+      "step": 41570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.389570513174577e-05,
+      "loss": 2.5917,
+      "step": 41575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.388557553184442e-05,
+      "loss": 2.5274,
+      "step": 41580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3875446186704076e-05,
+      "loss": 2.6242,
+      "step": 41585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3865317096746756e-05,
+      "loss": 2.7336,
+      "step": 41590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.385518826239454e-05,
+      "loss": 2.6914,
+      "step": 41595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.38450596840694e-05,
+      "loss": 2.593,
+      "step": 41600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.383493136219339e-05,
+      "loss": 2.5011,
+      "step": 41605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.382480329718848e-05,
+      "loss": 2.5461,
+      "step": 41610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.381467548947668e-05,
+      "loss": 2.6048,
+      "step": 41615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.380454793947996e-05,
+      "loss": 2.604,
+      "step": 41620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.37944206476203e-05,
+      "loss": 2.6749,
+      "step": 41625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.378429361431968e-05,
+      "loss": 2.7018,
+      "step": 41630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.377416684e-05,
+      "loss": 2.6353,
+      "step": 41635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3764040325083235e-05,
+      "loss": 2.7719,
+      "step": 41640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.375391406999129e-05,
+      "loss": 2.681,
+      "step": 41645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3743788075146086e-05,
+      "loss": 2.6171,
+      "step": 41650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.373366234096956e-05,
+      "loss": 2.8173,
+      "step": 41655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3723536867883565e-05,
+      "loss": 2.5896,
+      "step": 41660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.371341165631002e-05,
+      "loss": 2.6485,
+      "step": 41665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.370328670667075e-05,
+      "loss": 2.7477,
+      "step": 41670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.369316201938766e-05,
+      "loss": 2.5796,
+      "step": 41675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.368303759488258e-05,
+      "loss": 2.6341,
+      "step": 41680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3672913433577343e-05,
+      "loss": 2.7072,
+      "step": 41685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.366278953589382e-05,
+      "loss": 2.6457,
+      "step": 41690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3652665902253784e-05,
+      "loss": 2.7301,
+      "step": 41695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.364254253307907e-05,
+      "loss": 2.6138,
+      "step": 41700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.363241942879144e-05,
+      "loss": 2.7254,
+      "step": 41705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3622296589812704e-05,
+      "loss": 2.7293,
+      "step": 41710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.361217401656464e-05,
+      "loss": 2.5087,
+      "step": 41715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3602051709468996e-05,
+      "loss": 2.7388,
+      "step": 41720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.359192966894755e-05,
+      "loss": 2.6208,
+      "step": 41725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3581807895422004e-05,
+      "loss": 2.7625,
+      "step": 41730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.357168638931412e-05,
+      "loss": 2.6906,
+      "step": 41735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.356156515104559e-05,
+      "loss": 2.6958,
+      "step": 41740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.355144418103814e-05,
+      "loss": 2.7241,
+      "step": 41745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.354132347971347e-05,
+      "loss": 2.6952,
+      "step": 41750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.353120304749324e-05,
+      "loss": 2.6403,
+      "step": 41755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.352108288479916e-05,
+      "loss": 2.5504,
+      "step": 41760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3510962992052855e-05,
+      "loss": 2.7047,
+      "step": 41765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3500843369675996e-05,
+      "loss": 2.7096,
+      "step": 41770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.349072401809021e-05,
+      "loss": 2.6773,
+      "step": 41775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.348060493771714e-05,
+      "loss": 2.6749,
+      "step": 41780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.347048612897842e-05,
+      "loss": 2.6176,
+      "step": 41785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3460367592295606e-05,
+      "loss": 2.5853,
+      "step": 41790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3450249328090345e-05,
+      "loss": 2.5731,
+      "step": 41795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.344013133678418e-05,
+      "loss": 2.6455,
+      "step": 41800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.343001361879869e-05,
+      "loss": 2.7288,
+      "step": 41805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.341989617455546e-05,
+      "loss": 2.6679,
+      "step": 41810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3409779004476015e-05,
+      "loss": 2.6582,
+      "step": 41815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.339966210898192e-05,
+      "loss": 2.6354,
+      "step": 41820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.338954548849466e-05,
+      "loss": 2.5939,
+      "step": 41825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3379429143435786e-05,
+      "loss": 2.6572,
+      "step": 41830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.336931307422678e-05,
+      "loss": 2.6666,
+      "step": 41835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3359197281289133e-05,
+      "loss": 2.6595,
+      "step": 41840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3349081765044355e-05,
+      "loss": 2.6398,
+      "step": 41845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.333896652591388e-05,
+      "loss": 2.7505,
+      "step": 41850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.33288515643192e-05,
+      "loss": 2.6517,
+      "step": 41855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.331873688068171e-05,
+      "loss": 2.696,
+      "step": 41860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.330862247542288e-05,
+      "loss": 2.5552,
+      "step": 41865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.329850834896413e-05,
+      "loss": 2.6313,
+      "step": 41870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.328839450172686e-05,
+      "loss": 2.745,
+      "step": 41875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3278280934132506e-05,
+      "loss": 2.664,
+      "step": 41880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.326816764660239e-05,
+      "loss": 2.5958,
+      "step": 41885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3258054639557945e-05,
+      "loss": 2.6747,
+      "step": 41890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.324794191342051e-05,
+      "loss": 2.6438,
+      "step": 41895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.323782946861143e-05,
+      "loss": 2.6052,
+      "step": 41900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3227717305552065e-05,
+      "loss": 2.7105,
+      "step": 41905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3217605424663734e-05,
+      "loss": 2.7743,
+      "step": 41910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.320749382636777e-05,
+      "loss": 2.6259,
+      "step": 41915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.319738251108545e-05,
+      "loss": 2.7443,
+      "step": 41920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.318727147923808e-05,
+      "loss": 2.7158,
+      "step": 41925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.317716073124695e-05,
+      "loss": 2.5448,
+      "step": 41930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.316705026753332e-05,
+      "loss": 2.7171,
+      "step": 41935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3156940088518476e-05,
+      "loss": 2.5327,
+      "step": 41940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3146830194623614e-05,
+      "loss": 2.5917,
+      "step": 41945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3136720586270024e-05,
+      "loss": 2.7198,
+      "step": 41950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.312661126387888e-05,
+      "loss": 2.6827,
+      "step": 41955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.311650222787141e-05,
+      "loss": 2.5714,
+      "step": 41960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.310639347866882e-05,
+      "loss": 2.5231,
+      "step": 41965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.30962850166923e-05,
+      "loss": 2.6556,
+      "step": 41970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.308617684236302e-05,
+      "loss": 2.7043,
+      "step": 41975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3076068956102125e-05,
+      "loss": 2.5993,
+      "step": 41980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.30659613583308e-05,
+      "loss": 2.7446,
+      "step": 41985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3055854049470156e-05,
+      "loss": 2.6648,
+      "step": 41990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.304574702994132e-05,
+      "loss": 2.5644,
+      "step": 41995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.303564030016544e-05,
+      "loss": 2.7633,
+      "step": 42000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.302553386056357e-05,
+      "loss": 2.5739,
+      "step": 42005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3015427711556846e-05,
+      "loss": 2.7535,
+      "step": 42010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.300532185356631e-05,
+      "loss": 2.6985,
+      "step": 42015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.299521628701304e-05,
+      "loss": 2.8325,
+      "step": 42020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.298511101231811e-05,
+      "loss": 2.6707,
+      "step": 42025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.297500602990253e-05,
+      "loss": 2.6649,
+      "step": 42030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2964901340187367e-05,
+      "loss": 2.7225,
+      "step": 42035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2954796943593605e-05,
+      "loss": 2.6104,
+      "step": 42040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2944692840542277e-05,
+      "loss": 2.6337,
+      "step": 42045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.293458903145434e-05,
+      "loss": 2.641,
+      "step": 42050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2924485516750804e-05,
+      "loss": 2.8449,
+      "step": 42055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.291438229685263e-05,
+      "loss": 2.7018,
+      "step": 42060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.290427937218078e-05,
+      "loss": 2.6495,
+      "step": 42065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.28941767431562e-05,
+      "loss": 2.5972,
+      "step": 42070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.28840744101998e-05,
+      "loss": 2.604,
+      "step": 42075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2873972373732505e-05,
+      "loss": 2.7225,
+      "step": 42080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2863870634175254e-05,
+      "loss": 2.7364,
+      "step": 42085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.285376919194891e-05,
+      "loss": 2.4938,
+      "step": 42090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.284366804747437e-05,
+      "loss": 2.7811,
+      "step": 42095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2833567201172486e-05,
+      "loss": 2.6304,
+      "step": 42100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.282346665346415e-05,
+      "loss": 2.739,
+      "step": 42105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2813366404770163e-05,
+      "loss": 2.6888,
+      "step": 42110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.280326645551138e-05,
+      "loss": 2.6878,
+      "step": 42115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2793166806108634e-05,
+      "loss": 2.4685,
+      "step": 42120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.278306745698271e-05,
+      "loss": 2.5795,
+      "step": 42125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2772968408554434e-05,
+      "loss": 2.6637,
+      "step": 42130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.276286966124454e-05,
+      "loss": 2.7002,
+      "step": 42135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.275277121547383e-05,
+      "loss": 2.7207,
+      "step": 42140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.274267307166307e-05,
+      "loss": 2.7844,
+      "step": 42145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2732575230232984e-05,
+      "loss": 2.7036,
+      "step": 42150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.272247769160432e-05,
+      "loss": 2.679,
+      "step": 42155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2712380456197776e-05,
+      "loss": 2.7084,
+      "step": 42160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.270228352443409e-05,
+      "loss": 2.5545,
+      "step": 42165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2692186896733926e-05,
+      "loss": 2.7477,
+      "step": 42170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2682090573517974e-05,
+      "loss": 2.6689,
+      "step": 42175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2671994555206926e-05,
+      "loss": 2.6244,
+      "step": 42180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.266189884222141e-05,
+      "loss": 2.6883,
+      "step": 42185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.265180343498209e-05,
+      "loss": 2.6574,
+      "step": 42190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2641708333909574e-05,
+      "loss": 2.6466,
+      "step": 42195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.263161353942451e-05,
+      "loss": 2.6935,
+      "step": 42200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.262151905194748e-05,
+      "loss": 2.7143,
+      "step": 42205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.261142487189907e-05,
+      "loss": 2.5814,
+      "step": 42210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2601330999699886e-05,
+      "loss": 2.7351,
+      "step": 42215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.259123743577047e-05,
+      "loss": 2.7051,
+      "step": 42220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.258114418053142e-05,
+      "loss": 2.7713,
+      "step": 42225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.257105123440321e-05,
+      "loss": 2.8506,
+      "step": 42230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.256095859780642e-05,
+      "loss": 2.605,
+      "step": 42235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.255086627116156e-05,
+      "loss": 2.7516,
+      "step": 42240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.254077425488911e-05,
+      "loss": 2.7034,
+      "step": 42245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2530682549409564e-05,
+      "loss": 2.621,
+      "step": 42250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.252059115514341e-05,
+      "loss": 2.6892,
+      "step": 42255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.251050007251113e-05,
+      "loss": 2.5736,
+      "step": 42260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.250040930193313e-05,
+      "loss": 2.6548,
+      "step": 42265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2490318843829866e-05,
+      "loss": 2.6175,
+      "step": 42270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.248022869862178e-05,
+      "loss": 2.7767,
+      "step": 42275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.247013886672927e-05,
+      "loss": 2.5824,
+      "step": 42280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2460049348572744e-05,
+      "loss": 2.7995,
+      "step": 42285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2449960144572565e-05,
+      "loss": 2.5688,
+      "step": 42290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.243987125514911e-05,
+      "loss": 2.7917,
+      "step": 42295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.242978268072279e-05,
+      "loss": 2.5308,
+      "step": 42300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2419694421713866e-05,
+      "loss": 2.7571,
+      "step": 42305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.240960647854274e-05,
+      "loss": 2.6273,
+      "step": 42310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.239951885162969e-05,
+      "loss": 2.6253,
+      "step": 42315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2389431541395056e-05,
+      "loss": 2.64,
+      "step": 42320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.23793445482591e-05,
+      "loss": 2.7663,
+      "step": 42325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.236925787264212e-05,
+      "loss": 2.5407,
+      "step": 42330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2359171514964393e-05,
+      "loss": 2.6439,
+      "step": 42335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.234908547564614e-05,
+      "loss": 2.6926,
+      "step": 42340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.233899975510763e-05,
+      "loss": 2.6977,
+      "step": 42345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2328914353769076e-05,
+      "loss": 2.7058,
+      "step": 42350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.23188292720507e-05,
+      "loss": 2.7736,
+      "step": 42355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.230874451037271e-05,
+      "loss": 2.8351,
+      "step": 42360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.229866006915528e-05,
+      "loss": 2.7208,
+      "step": 42365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2288575948818584e-05,
+      "loss": 2.7611,
+      "step": 42370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2278492149782784e-05,
+      "loss": 2.7341,
+      "step": 42375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2268408672468054e-05,
+      "loss": 2.7165,
+      "step": 42380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.225832551729447e-05,
+      "loss": 2.7151,
+      "step": 42385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.224824268468219e-05,
+      "loss": 2.681,
+      "step": 42390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.223816017505135e-05,
+      "loss": 2.608,
+      "step": 42395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.222807798882198e-05,
+      "loss": 2.7503,
+      "step": 42400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.22179961264142e-05,
+      "loss": 2.7428,
+      "step": 42405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.220791458824806e-05,
+      "loss": 2.7358,
+      "step": 42410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.219783337474365e-05,
+      "loss": 2.6816,
+      "step": 42415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.218775248632094e-05,
+      "loss": 2.6037,
+      "step": 42420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.217767192340001e-05,
+      "loss": 2.6228,
+      "step": 42425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2167591686400875e-05,
+      "loss": 2.5204,
+      "step": 42430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.21575117757435e-05,
+      "loss": 2.6217,
+      "step": 42435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.214743219184789e-05,
+      "loss": 2.7433,
+      "step": 42440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.213735293513401e-05,
+      "loss": 2.572,
+      "step": 42445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.212727400602182e-05,
+      "loss": 2.5112,
+      "step": 42450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.211719540493129e-05,
+      "loss": 2.6427,
+      "step": 42455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.21071171322823e-05,
+      "loss": 2.5963,
+      "step": 42460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.209703918849481e-05,
+      "loss": 2.7837,
+      "step": 42465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.208696157398869e-05,
+      "loss": 2.8299,
+      "step": 42470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.207688428918388e-05,
+      "loss": 2.7339,
+      "step": 42475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.206680733450019e-05,
+      "loss": 2.7889,
+      "step": 42480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.205673071035752e-05,
+      "loss": 2.7018,
+      "step": 42485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.204665441717573e-05,
+      "loss": 2.7223,
+      "step": 42490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2036578455374626e-05,
+      "loss": 2.6498,
+      "step": 42495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.202650282537405e-05,
+      "loss": 2.7093,
+      "step": 42500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2016427527593784e-05,
+      "loss": 2.6916,
+      "step": 42505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2006352562453635e-05,
+      "loss": 2.5849,
+      "step": 42510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.199627793037341e-05,
+      "loss": 2.8024,
+      "step": 42515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.198620363177282e-05,
+      "loss": 2.7105,
+      "step": 42520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1976129667071676e-05,
+      "loss": 2.6032,
+      "step": 42525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.196605603668965e-05,
+      "loss": 2.7523,
+      "step": 42530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.195598274104652e-05,
+      "loss": 2.7099,
+      "step": 42535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1945909780561954e-05,
+      "loss": 2.756,
+      "step": 42540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1935837155655674e-05,
+      "loss": 2.6485,
+      "step": 42545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.192576486674739e-05,
+      "loss": 2.711,
+      "step": 42550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.19156929142567e-05,
+      "loss": 2.5143,
+      "step": 42555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.19056212986033e-05,
+      "loss": 2.6015,
+      "step": 42560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.189555002020683e-05,
+      "loss": 2.6804,
+      "step": 42565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.188547907948689e-05,
+      "loss": 2.7313,
+      "step": 42570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.187540847686315e-05,
+      "loss": 2.6503,
+      "step": 42575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.186533821275513e-05,
+      "loss": 2.7184,
+      "step": 42580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.185526828758248e-05,
+      "loss": 2.6606,
+      "step": 42585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.184519870176473e-05,
+      "loss": 2.5641,
+      "step": 42590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.183512945572145e-05,
+      "loss": 2.7652,
+      "step": 42595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.182506054987215e-05,
+      "loss": 2.6496,
+      "step": 42600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.18149919846364e-05,
+      "loss": 2.7291,
+      "step": 42605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1804923760433715e-05,
+      "loss": 2.642,
+      "step": 42610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1794855877683545e-05,
+      "loss": 2.6231,
+      "step": 42615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1784788336805434e-05,
+      "loss": 2.6565,
+      "step": 42620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1774721138218785e-05,
+      "loss": 2.7379,
+      "step": 42625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1764654282343116e-05,
+      "loss": 2.5884,
+      "step": 42630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1754587769597825e-05,
+      "loss": 2.6765,
+      "step": 42635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1744521600402354e-05,
+      "loss": 2.6314,
+      "step": 42640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.173445577517614e-05,
+      "loss": 2.5926,
+      "step": 42645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.172439029433854e-05,
+      "loss": 2.6517,
+      "step": 42650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1714325158308963e-05,
+      "loss": 2.6644,
+      "step": 42655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.170426036750676e-05,
+      "loss": 2.5314,
+      "step": 42660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.169419592235131e-05,
+      "loss": 2.7281,
+      "step": 42665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.168413182326196e-05,
+      "loss": 2.6566,
+      "step": 42670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1674068070658e-05,
+      "loss": 2.7281,
+      "step": 42675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1664004664958774e-05,
+      "loss": 2.7116,
+      "step": 42680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1653941606583546e-05,
+      "loss": 2.6335,
+      "step": 42685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.164387889595164e-05,
+      "loss": 2.5048,
+      "step": 42690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.163381653348229e-05,
+      "loss": 2.6846,
+      "step": 42695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.162375451959476e-05,
+      "loss": 2.5746,
+      "step": 42700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.161369285470832e-05,
+      "loss": 2.7107,
+      "step": 42705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.160363153924214e-05,
+      "loss": 2.6695,
+      "step": 42710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1593570573615495e-05,
+      "loss": 2.7799,
+      "step": 42715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.158350995824751e-05,
+      "loss": 2.5279,
+      "step": 42720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1573449693557406e-05,
+      "loss": 2.6782,
+      "step": 42725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.156338977996436e-05,
+      "loss": 2.5683,
+      "step": 42730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.15533302178875e-05,
+      "loss": 2.6905,
+      "step": 42735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.154327100774599e-05,
+      "loss": 2.4991,
+      "step": 42740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.153321214995892e-05,
+      "loss": 2.5355,
+      "step": 42745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1523153644945415e-05,
+      "loss": 2.7428,
+      "step": 42750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1513095493124564e-05,
+      "loss": 2.6629,
+      "step": 42755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1503037694915446e-05,
+      "loss": 2.6447,
+      "step": 42760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.149298025073715e-05,
+      "loss": 2.7863,
+      "step": 42765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.148292316100868e-05,
+      "loss": 2.7236,
+      "step": 42770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.147286642614912e-05,
+      "loss": 2.6128,
+      "step": 42775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.146281004657744e-05,
+      "loss": 2.7139,
+      "step": 42780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1452754022712665e-05,
+      "loss": 2.697,
+      "step": 42785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.14426983549738e-05,
+      "loss": 2.5546,
+      "step": 42790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.14326430437798e-05,
+      "loss": 2.5483,
+      "step": 42795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1422588089549644e-05,
+      "loss": 2.6928,
+      "step": 42800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1412533492702254e-05,
+      "loss": 2.6933,
+      "step": 42805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.140247925365658e-05,
+      "loss": 2.6467,
+      "step": 42810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1392425372831535e-05,
+      "loss": 2.7061,
+      "step": 42815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1382371850646e-05,
+      "loss": 2.6106,
+      "step": 42820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.137231868751889e-05,
+      "loss": 2.4924,
+      "step": 42825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.136226588386905e-05,
+      "loss": 2.5968,
+      "step": 42830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1352213440115374e-05,
+      "loss": 2.701,
+      "step": 42835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.134216135667665e-05,
+      "loss": 2.5245,
+      "step": 42840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1332109633971745e-05,
+      "loss": 2.7801,
+      "step": 42845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.132205827241944e-05,
+      "loss": 2.6275,
+      "step": 42850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1312007272438545e-05,
+      "loss": 2.6484,
+      "step": 42855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.130195663444787e-05,
+      "loss": 2.6842,
+      "step": 42860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.129190635886613e-05,
+      "loss": 2.715,
+      "step": 42865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1281856446112116e-05,
+      "loss": 2.6111,
+      "step": 42870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1271806896604524e-05,
+      "loss": 2.6296,
+      "step": 42875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.126175771076209e-05,
+      "loss": 2.7602,
+      "step": 42880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1251708889003544e-05,
+      "loss": 2.7139,
+      "step": 42885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.124166043174754e-05,
+      "loss": 2.7446,
+      "step": 42890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1231612339412796e-05,
+      "loss": 2.6773,
+      "step": 42895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.122156461241792e-05,
+      "loss": 2.7001,
+      "step": 42900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.121151725118159e-05,
+      "loss": 2.6008,
+      "step": 42905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.120147025612242e-05,
+      "loss": 2.6959,
+      "step": 42910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.119142362765904e-05,
+      "loss": 2.6304,
+      "step": 42915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1181377366210035e-05,
+      "loss": 2.6672,
+      "step": 42920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.117133147219398e-05,
+      "loss": 2.6447,
+      "step": 42925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.116128594602948e-05,
+      "loss": 2.6824,
+      "step": 42930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1151240788135036e-05,
+      "loss": 2.728,
+      "step": 42935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1141195998929216e-05,
+      "loss": 2.5223,
+      "step": 42940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1131151578830545e-05,
+      "loss": 2.6629,
+      "step": 42945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1121107528257506e-05,
+      "loss": 2.7394,
+      "step": 42950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.111106384762864e-05,
+      "loss": 2.5895,
+      "step": 42955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.110102053736236e-05,
+      "loss": 2.723,
+      "step": 42960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.109097759787717e-05,
+      "loss": 2.5494,
+      "step": 42965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1080935029591486e-05,
+      "loss": 2.5109,
+      "step": 42970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1070892832923755e-05,
+      "loss": 2.66,
+      "step": 42975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.106085100829239e-05,
+      "loss": 2.6387,
+      "step": 42980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1050809556115776e-05,
+      "loss": 2.6627,
+      "step": 42985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.104076847681233e-05,
+      "loss": 2.7311,
+      "step": 42990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1030727770800376e-05,
+      "loss": 2.7423,
+      "step": 42995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.102068743849828e-05,
+      "loss": 2.7063,
+      "step": 43000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.10106474803244e-05,
+      "loss": 2.7827,
+      "step": 43005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1000607896697024e-05,
+      "loss": 2.6636,
+      "step": 43010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0990568688034495e-05,
+      "loss": 2.6445,
+      "step": 43015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.098052985475506e-05,
+      "loss": 2.5455,
+      "step": 43020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.097049139727703e-05,
+      "loss": 2.7012,
+      "step": 43025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0960453316018635e-05,
+      "loss": 2.7044,
+      "step": 43030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.095041561139813e-05,
+      "loss": 2.7633,
+      "step": 43035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.094037828383374e-05,
+      "loss": 2.612,
+      "step": 43040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.093034133374367e-05,
+      "loss": 2.724,
+      "step": 43045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.092030476154614e-05,
+      "loss": 2.6864,
+      "step": 43050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0910268567659295e-05,
+      "loss": 2.6471,
+      "step": 43055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.090023275250133e-05,
+      "loss": 2.7032,
+      "step": 43060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0890197316490366e-05,
+      "loss": 2.7247,
+      "step": 43065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0880162260044534e-05,
+      "loss": 2.5644,
+      "step": 43070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0870127583581985e-05,
+      "loss": 2.6127,
+      "step": 43075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0860093287520787e-05,
+      "loss": 2.7092,
+      "step": 43080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.085005937227905e-05,
+      "loss": 2.7124,
+      "step": 43085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0840025838274807e-05,
+      "loss": 2.7287,
+      "step": 43090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.082999268592614e-05,
+      "loss": 2.6141,
+      "step": 43095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0819959915651086e-05,
+      "loss": 2.6513,
+      "step": 43100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.080992752786765e-05,
+      "loss": 2.6799,
+      "step": 43105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.079989552299386e-05,
+      "loss": 2.6916,
+      "step": 43110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.078986390144768e-05,
+      "loss": 2.6782,
+      "step": 43115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07798326636471e-05,
+      "loss": 2.8544,
+      "step": 43120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0769801810010065e-05,
+      "loss": 2.5594,
+      "step": 43125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0759771340954515e-05,
+      "loss": 2.7018,
+      "step": 43130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07497412568984e-05,
+      "loss": 2.6992,
+      "step": 43135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07397115582596e-05,
+      "loss": 2.6083,
+      "step": 43140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0729682245456035e-05,
+      "loss": 2.5662,
+      "step": 43145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071965331890555e-05,
+      "loss": 2.5628,
+      "step": 43150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070962477902602e-05,
+      "loss": 2.7709,
+      "step": 43155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069959662623532e-05,
+      "loss": 2.6453,
+      "step": 43160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0689568860951233e-05,
+      "loss": 2.7703,
+      "step": 43165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0679541483591595e-05,
+      "loss": 2.7061,
+      "step": 43170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066951449457419e-05,
+      "loss": 2.6831,
+      "step": 43175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.065948789431683e-05,
+      "loss": 2.4856,
+      "step": 43180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064946168323723e-05,
+      "loss": 2.6068,
+      "step": 43185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063943586175316e-05,
+      "loss": 2.6766,
+      "step": 43190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062941043028237e-05,
+      "loss": 2.601,
+      "step": 43195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.061938538924255e-05,
+      "loss": 2.6326,
+      "step": 43200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.060936073905143e-05,
+      "loss": 2.7546,
+      "step": 43205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.059933648012665e-05,
+      "loss": 2.5498,
+      "step": 43210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0589312612885896e-05,
+      "loss": 2.6151,
+      "step": 43215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0579289137746845e-05,
+      "loss": 2.5474,
+      "step": 43220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.056926605512708e-05,
+      "loss": 2.5605,
+      "step": 43225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.055924336544426e-05,
+      "loss": 2.7308,
+      "step": 43230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.054922106911596e-05,
+      "loss": 2.6473,
+      "step": 43235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.05391991665598e-05,
+      "loss": 2.7675,
+      "step": 43240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0529177658193296e-05,
+      "loss": 2.687,
+      "step": 43245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.051915654443403e-05,
+      "loss": 2.6142,
+      "step": 43250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.050913582569956e-05,
+      "loss": 2.7069,
+      "step": 43255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0499115502407355e-05,
+      "loss": 2.6323,
+      "step": 43260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.048909557497495e-05,
+      "loss": 2.6741,
+      "step": 43265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.047907604381982e-05,
+      "loss": 2.6443,
+      "step": 43270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.046905690935946e-05,
+      "loss": 2.5964,
+      "step": 43275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0459038172011274e-05,
+      "loss": 2.6954,
+      "step": 43280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.044901983219273e-05,
+      "loss": 2.6415,
+      "step": 43285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.043900189032126e-05,
+      "loss": 2.711,
+      "step": 43290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.042898434681422e-05,
+      "loss": 2.7106,
+      "step": 43295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.041896720208907e-05,
+      "loss": 2.6244,
+      "step": 43300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.040895045656311e-05,
+      "loss": 2.6595,
+      "step": 43305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.039893411065372e-05,
+      "loss": 2.7031,
+      "step": 43310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.038891816477826e-05,
+      "loss": 2.7204,
+      "step": 43315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0378902619354e-05,
+      "loss": 2.7042,
+      "step": 43320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.036888747479829e-05,
+      "loss": 2.672,
+      "step": 43325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.035887273152838e-05,
+      "loss": 2.7198,
+      "step": 43330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.034885838996158e-05,
+      "loss": 2.7057,
+      "step": 43335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0338844450515115e-05,
+      "loss": 2.6394,
+      "step": 43340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0328830913606215e-05,
+      "loss": 2.6608,
+      "step": 43345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.031881777965214e-05,
+      "loss": 2.7652,
+      "step": 43350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.030880504907003e-05,
+      "loss": 2.641,
+      "step": 43355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0298792722277136e-05,
+      "loss": 2.5848,
+      "step": 43360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.028878079969057e-05,
+      "loss": 2.7065,
+      "step": 43365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.027876928172752e-05,
+      "loss": 2.6712,
+      "step": 43370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0268758168805143e-05,
+      "loss": 2.6327,
+      "step": 43375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.02587474613405e-05,
+      "loss": 2.7028,
+      "step": 43380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.024873715975074e-05,
+      "loss": 2.6708,
+      "step": 43385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0238727264452914e-05,
+      "loss": 2.6079,
+      "step": 43390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0228717775864124e-05,
+      "loss": 2.6631,
+      "step": 43395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.021870869440139e-05,
+      "loss": 2.5915,
+      "step": 43400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.020870002048175e-05,
+      "loss": 2.5792,
+      "step": 43405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.019869175452225e-05,
+      "loss": 2.5182,
+      "step": 43410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0188683896939855e-05,
+      "loss": 2.5781,
+      "step": 43415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0178676448151576e-05,
+      "loss": 2.5438,
+      "step": 43420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.016866940857435e-05,
+      "loss": 2.6476,
+      "step": 43425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0158662778625145e-05,
+      "loss": 2.6672,
+      "step": 43430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0148656558720906e-05,
+      "loss": 2.6753,
+      "step": 43435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.013865074927852e-05,
+      "loss": 2.5396,
+      "step": 43440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.012864535071492e-05,
+      "loss": 2.66,
+      "step": 43445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.011864036344694e-05,
+      "loss": 2.6876,
+      "step": 43450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.010863578789147e-05,
+      "loss": 2.7001,
+      "step": 43455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.009863162446535e-05,
+      "loss": 2.7084,
+      "step": 43460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.008862787358541e-05,
+      "loss": 2.6052,
+      "step": 43465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0078624535668486e-05,
+      "loss": 2.52,
+      "step": 43470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.006862161113133e-05,
+      "loss": 2.7668,
+      "step": 43475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.005861910039076e-05,
+      "loss": 2.6415,
+      "step": 43480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0048617003863495e-05,
+      "loss": 2.69,
+      "step": 43485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.003861532196633e-05,
+      "loss": 2.585,
+      "step": 43490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0028614055115936e-05,
+      "loss": 2.7428,
+      "step": 43495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.001861320372905e-05,
+      "loss": 2.5554,
+      "step": 43500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.000861276822238e-05,
+      "loss": 2.8154,
+      "step": 43505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.999861274901256e-05,
+      "loss": 2.6082,
+      "step": 43510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.998861314651628e-05,
+      "loss": 2.6869,
+      "step": 43515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9978613961150145e-05,
+      "loss": 2.6515,
+      "step": 43520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.996861519333081e-05,
+      "loss": 2.6782,
+      "step": 43525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.995861684347488e-05,
+      "loss": 2.8073,
+      "step": 43530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9948618911998905e-05,
+      "loss": 2.5897,
+      "step": 43535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.993862139931951e-05,
+      "loss": 2.6258,
+      "step": 43540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9928624305853184e-05,
+      "loss": 2.5809,
+      "step": 43545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.991862763201651e-05,
+      "loss": 2.5906,
+      "step": 43550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.990863137822597e-05,
+      "loss": 2.5392,
+      "step": 43555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.989863554489809e-05,
+      "loss": 2.6101,
+      "step": 43560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.988864013244937e-05,
+      "loss": 2.5749,
+      "step": 43565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.987864514129622e-05,
+      "loss": 2.5946,
+      "step": 43570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.986865057185512e-05,
+      "loss": 2.7068,
+      "step": 43575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.985865642454249e-05,
+      "loss": 2.6932,
+      "step": 43580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.984866269977475e-05,
+      "loss": 2.5528,
+      "step": 43585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.983866939796831e-05,
+      "loss": 2.8389,
+      "step": 43590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.982867651953951e-05,
+      "loss": 2.6065,
+      "step": 43595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.981868406490474e-05,
+      "loss": 2.7513,
+      "step": 43600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9808692034480315e-05,
+      "loss": 2.6135,
+      "step": 43605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9798700428682584e-05,
+      "loss": 2.6878,
+      "step": 43610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.978870924792782e-05,
+      "loss": 2.8018,
+      "step": 43615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.977871849263234e-05,
+      "loss": 2.725,
+      "step": 43620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.976872816321243e-05,
+      "loss": 2.6016,
+      "step": 43625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.975873826008429e-05,
+      "loss": 2.7241,
+      "step": 43630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.97487487836642e-05,
+      "loss": 2.4895,
+      "step": 43635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9738759734368364e-05,
+      "loss": 2.6895,
+      "step": 43640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.972877111261296e-05,
+      "loss": 2.6483,
+      "step": 43645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.971878291881421e-05,
+      "loss": 2.658,
+      "step": 43650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.970879515338824e-05,
+      "loss": 2.5949,
+      "step": 43655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.969880781675124e-05,
+      "loss": 2.7087,
+      "step": 43660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.968882090931928e-05,
+      "loss": 2.6828,
+      "step": 43665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.967883443150852e-05,
+      "loss": 2.5762,
+      "step": 43670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9668848383735016e-05,
+      "loss": 2.7544,
+      "step": 43675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.965886276641486e-05,
+      "loss": 2.7013,
+      "step": 43680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.964887757996413e-05,
+      "loss": 2.7282,
+      "step": 43685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.963889282479883e-05,
+      "loss": 2.5898,
+      "step": 43690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.962890850133501e-05,
+      "loss": 2.5719,
+      "step": 43695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.961892460998863e-05,
+      "loss": 2.6296,
+      "step": 43700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.960894115117571e-05,
+      "loss": 2.6643,
+      "step": 43705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.95989581253122e-05,
+      "loss": 2.7212,
+      "step": 43710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9588975532814056e-05,
+      "loss": 2.6003,
+      "step": 43715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9578993374097227e-05,
+      "loss": 2.5887,
+      "step": 43720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9569011649577584e-05,
+      "loss": 2.6008,
+      "step": 43725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9559030359671054e-05,
+      "loss": 2.8394,
+      "step": 43730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.95490495047935e-05,
+      "loss": 2.6356,
+      "step": 43735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.953906908536078e-05,
+      "loss": 2.7419,
+      "step": 43740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9529089101788746e-05,
+      "loss": 2.6217,
+      "step": 43745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9519109554493196e-05,
+      "loss": 2.6161,
+      "step": 43750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.950913044388997e-05,
+      "loss": 2.6072,
+      "step": 43755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.949915177039481e-05,
+      "loss": 2.7867,
+      "step": 43760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.948917353442351e-05,
+      "loss": 2.6546,
+      "step": 43765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.947919573639181e-05,
+      "loss": 2.8814,
+      "step": 43770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9469218376715444e-05,
+      "loss": 2.698,
+      "step": 43775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9459241455810135e-05,
+      "loss": 2.7082,
+      "step": 43780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.944926497409156e-05,
+      "loss": 2.6187,
+      "step": 43785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.943928893197541e-05,
+      "loss": 2.6279,
+      "step": 43790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9429313329877314e-05,
+      "loss": 2.5286,
+      "step": 43795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.941933816821294e-05,
+      "loss": 2.6889,
+      "step": 43800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.94093634473979e-05,
+      "loss": 2.5685,
+      "step": 43805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.939938916784779e-05,
+      "loss": 2.7981,
+      "step": 43810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9389415329978226e-05,
+      "loss": 2.5865,
+      "step": 43815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9379441934204716e-05,
+      "loss": 2.657,
+      "step": 43820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.936946898094286e-05,
+      "loss": 2.7197,
+      "step": 43825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.935949647060815e-05,
+      "loss": 2.6396,
+      "step": 43830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.934952440361611e-05,
+      "loss": 2.6027,
+      "step": 43835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.933955278038226e-05,
+      "loss": 2.6046,
+      "step": 43840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.932958160132202e-05,
+      "loss": 2.6209,
+      "step": 43845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9319610866850905e-05,
+      "loss": 2.6858,
+      "step": 43850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.930964057738429e-05,
+      "loss": 2.7032,
+      "step": 43855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.929967073333762e-05,
+      "loss": 2.6574,
+      "step": 43860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.928970133512631e-05,
+      "loss": 2.711,
+      "step": 43865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.927973238316571e-05,
+      "loss": 2.5835,
+      "step": 43870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.926976387787123e-05,
+      "loss": 2.5388,
+      "step": 43875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9259795819658156e-05,
+      "loss": 2.6133,
+      "step": 43880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.924982820894186e-05,
+      "loss": 2.6769,
+      "step": 43885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9239861046137604e-05,
+      "loss": 2.7427,
+      "step": 43890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9229894331660696e-05,
+      "loss": 2.5187,
+      "step": 43895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.921992806592643e-05,
+      "loss": 2.6841,
+      "step": 43900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9209962249350015e-05,
+      "loss": 2.6257,
+      "step": 43905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9199996882346726e-05,
+      "loss": 2.6156,
+      "step": 43910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9190031965331724e-05,
+      "loss": 2.6037,
+      "step": 43915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9180067498720244e-05,
+      "loss": 2.651,
+      "step": 43920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9170103482927433e-05,
+      "loss": 2.716,
+      "step": 43925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.916013991836847e-05,
+      "loss": 2.6424,
+      "step": 43930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9150176805458494e-05,
+      "loss": 2.7034,
+      "step": 43935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.91402141446126e-05,
+      "loss": 2.7971,
+      "step": 43940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.913025193624592e-05,
+      "loss": 2.7117,
+      "step": 43945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.91202901807735e-05,
+      "loss": 2.7325,
+      "step": 43950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.911032887861042e-05,
+      "loss": 2.6165,
+      "step": 43955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.910036803017173e-05,
+      "loss": 2.5788,
+      "step": 43960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.909040763587243e-05,
+      "loss": 2.7465,
+      "step": 43965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.908044769612758e-05,
+      "loss": 2.6389,
+      "step": 43970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.907048821135209e-05,
+      "loss": 2.6741,
+      "step": 43975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9060529181961e-05,
+      "loss": 2.6651,
+      "step": 43980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.90505706083692e-05,
+      "loss": 2.6537,
+      "step": 43985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9040612490991646e-05,
+      "loss": 2.7115,
+      "step": 43990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.903065483024326e-05,
+      "loss": 2.6837,
+      "step": 43995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.902069762653891e-05,
+      "loss": 2.6335,
+      "step": 44000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9010740880293504e-05,
+      "loss": 2.6487,
+      "step": 44005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.900078459192185e-05,
+      "loss": 2.6096,
+      "step": 44010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8990828761838796e-05,
+      "loss": 2.6562,
+      "step": 44015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.898087339045918e-05,
+      "loss": 2.6666,
+      "step": 44020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8970918478197776e-05,
+      "loss": 2.7363,
+      "step": 44025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.896096402546939e-05,
+      "loss": 2.7326,
+      "step": 44030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.895101003268874e-05,
+      "loss": 2.6553,
+      "step": 44035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.894105650027059e-05,
+      "loss": 2.5634,
+      "step": 44040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.893110342862965e-05,
+      "loss": 2.6437,
+      "step": 44045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.892115081818062e-05,
+      "loss": 2.6794,
+      "step": 44050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.891119866933819e-05,
+      "loss": 2.6716,
+      "step": 44055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8901246982517004e-05,
+      "loss": 2.736,
+      "step": 44060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.889129575813175e-05,
+      "loss": 2.6348,
+      "step": 44065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.888134499659699e-05,
+      "loss": 2.5375,
+      "step": 44070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8871394698327354e-05,
+      "loss": 2.6454,
+      "step": 44075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8861444863737454e-05,
+      "loss": 2.538,
+      "step": 44080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.885149549324181e-05,
+      "loss": 2.6278,
+      "step": 44085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8841546587255006e-05,
+      "loss": 2.7429,
+      "step": 44090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.883159814619154e-05,
+      "loss": 2.6446,
+      "step": 44095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.882165017046595e-05,
+      "loss": 2.7568,
+      "step": 44100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.881170266049269e-05,
+      "loss": 2.5268,
+      "step": 44105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.880175561668624e-05,
+      "loss": 2.7277,
+      "step": 44110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.879180903946107e-05,
+      "loss": 2.5468,
+      "step": 44115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.878186292923158e-05,
+      "loss": 2.7468,
+      "step": 44120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.877191728641221e-05,
+      "loss": 2.6724,
+      "step": 44125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.876197211141732e-05,
+      "loss": 2.5913,
+      "step": 44130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.875202740466132e-05,
+      "loss": 2.7249,
+      "step": 44135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8742083166558515e-05,
+      "loss": 2.603,
+      "step": 44140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.873213939752326e-05,
+      "loss": 2.6509,
+      "step": 44145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8722196097969886e-05,
+      "loss": 2.7929,
+      "step": 44150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.871225326831265e-05,
+      "loss": 2.5953,
+      "step": 44155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.870231090896588e-05,
+      "loss": 2.5723,
+      "step": 44160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8692369020343764e-05,
+      "loss": 2.6866,
+      "step": 44165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8682427602860563e-05,
+      "loss": 2.6002,
+      "step": 44170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8672486656930525e-05,
+      "loss": 2.5884,
+      "step": 44175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.866254618296779e-05,
+      "loss": 2.7172,
+      "step": 44180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8652606181386576e-05,
+      "loss": 2.5487,
+      "step": 44185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.864266665260101e-05,
+      "loss": 2.6133,
+      "step": 44190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.863272759702526e-05,
+      "loss": 2.6791,
+      "step": 44195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.86227890150734e-05,
+      "loss": 2.8209,
+      "step": 44200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.861285090715956e-05,
+      "loss": 2.5952,
+      "step": 44205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.860291327369781e-05,
+      "loss": 2.7057,
+      "step": 44210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.85929761151022e-05,
+      "loss": 2.7615,
+      "step": 44215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.858303943178678e-05,
+      "loss": 2.737,
+      "step": 44220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8573103224165555e-05,
+      "loss": 2.4807,
+      "step": 44225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8563167492652516e-05,
+      "loss": 2.6496,
+      "step": 44230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.855323223766167e-05,
+      "loss": 2.8157,
+      "step": 44235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.854329745960695e-05,
+      "loss": 2.7181,
+      "step": 44240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.85333631589023e-05,
+      "loss": 2.673,
+      "step": 44245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8523429335961636e-05,
+      "loss": 2.6522,
+      "step": 44250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.851349599119888e-05,
+      "loss": 2.6765,
+      "step": 44255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8503563125027864e-05,
+      "loss": 2.7232,
+      "step": 44260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.849363073786248e-05,
+      "loss": 2.7303,
+      "step": 44265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8483698830116574e-05,
+      "loss": 2.8023,
+      "step": 44270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.847376740220394e-05,
+      "loss": 2.7494,
+      "step": 44275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.846383645453839e-05,
+      "loss": 2.646,
+      "step": 44280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.845390598753369e-05,
+      "loss": 2.6988,
+      "step": 44285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.844397600160361e-05,
+      "loss": 2.72,
+      "step": 44290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8434046497161907e-05,
+      "loss": 2.6823,
+      "step": 44295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.842411747462226e-05,
+      "loss": 2.7725,
+      "step": 44300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8414188934398395e-05,
+      "loss": 2.6274,
+      "step": 44305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.840426087690397e-05,
+      "loss": 2.7439,
+      "step": 44310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.839433330255268e-05,
+      "loss": 2.6892,
+      "step": 44315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8384406211758116e-05,
+      "loss": 2.5677,
+      "step": 44320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8374479604933915e-05,
+      "loss": 2.5655,
+      "step": 44325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8364553482493704e-05,
+      "loss": 2.8256,
+      "step": 44330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.835462784485101e-05,
+      "loss": 2.7283,
+      "step": 44335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.834470269241943e-05,
+      "loss": 2.58,
+      "step": 44340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.833477802561247e-05,
+      "loss": 2.7477,
+      "step": 44345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.832485384484369e-05,
+      "loss": 2.5971,
+      "step": 44350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.831493015052653e-05,
+      "loss": 2.6011,
+      "step": 44355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.83050069430745e-05,
+      "loss": 2.7692,
+      "step": 44360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.829508422290108e-05,
+      "loss": 2.7164,
+      "step": 44365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.828516199041965e-05,
+      "loss": 2.6749,
+      "step": 44370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.827524024604366e-05,
+      "loss": 2.7326,
+      "step": 44375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.826531899018649e-05,
+      "loss": 2.6548,
+      "step": 44380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8255398223261535e-05,
+      "loss": 2.6504,
+      "step": 44385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8245477945682154e-05,
+      "loss": 2.5033,
+      "step": 44390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.823555815786164e-05,
+      "loss": 2.58,
+      "step": 44395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.822563886021334e-05,
+      "loss": 2.6299,
+      "step": 44400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8215720053150534e-05,
+      "loss": 2.7631,
+      "step": 44405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8205801737086514e-05,
+      "loss": 2.675,
+      "step": 44410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.819588391243451e-05,
+      "loss": 2.7462,
+      "step": 44415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.818596657960774e-05,
+      "loss": 2.6864,
+      "step": 44420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.817604973901948e-05,
+      "loss": 2.661,
+      "step": 44425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.816613339108285e-05,
+      "loss": 2.7153,
+      "step": 44430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8156217536211055e-05,
+      "loss": 2.7036,
+      "step": 44435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.814630217481723e-05,
+      "loss": 2.6261,
+      "step": 44440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.813638730731451e-05,
+      "loss": 2.6714,
+      "step": 44445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8126472934116023e-05,
+      "loss": 2.617,
+      "step": 44450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8116559055634826e-05,
+      "loss": 2.6116,
+      "step": 44455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8106645672284014e-05,
+      "loss": 2.7434,
+      "step": 44460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8096732784476615e-05,
+      "loss": 2.7878,
+      "step": 44465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8086820392625654e-05,
+      "loss": 2.5977,
+      "step": 44470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8076908497144145e-05,
+      "loss": 2.5084,
+      "step": 44475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.806699709844506e-05,
+      "loss": 2.5443,
+      "step": 44480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.805708619694141e-05,
+      "loss": 2.6036,
+      "step": 44485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8047175793046065e-05,
+      "loss": 2.6402,
+      "step": 44490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8037265887171996e-05,
+      "loss": 2.7855,
+      "step": 44495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.802735647973209e-05,
+      "loss": 2.5785,
+      "step": 44500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8017447571139234e-05,
+      "loss": 2.7213,
+      "step": 44505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8007539161806297e-05,
+      "loss": 2.5952,
+      "step": 44510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.799763125214609e-05,
+      "loss": 2.6159,
+      "step": 44515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7987723842571474e-05,
+      "loss": 2.5906,
+      "step": 44520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.79778169334952e-05,
+      "loss": 2.5635,
+      "step": 44525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.796791052533008e-05,
+      "loss": 2.6495,
+      "step": 44530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.795800461848884e-05,
+      "loss": 2.5929,
+      "step": 44535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.794809921338424e-05,
+      "loss": 2.5455,
+      "step": 44540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.793819431042901e-05,
+      "loss": 2.5747,
+      "step": 44545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.792828991003579e-05,
+      "loss": 2.7304,
+      "step": 44550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.79183860126173e-05,
+      "loss": 2.7233,
+      "step": 44555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.790848261858616e-05,
+      "loss": 2.6912,
+      "step": 44560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7898579728355045e-05,
+      "loss": 2.4823,
+      "step": 44565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.788867734233651e-05,
+      "loss": 2.5739,
+      "step": 44570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.787877546094316e-05,
+      "loss": 2.7755,
+      "step": 44575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7868874084587605e-05,
+      "loss": 2.6965,
+      "step": 44580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.785897321368233e-05,
+      "loss": 2.5846,
+      "step": 44585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.784907284863989e-05,
+      "loss": 2.6672,
+      "step": 44590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.783917298987279e-05,
+      "loss": 2.6824,
+      "step": 44595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.782927363779351e-05,
+      "loss": 2.5489,
+      "step": 44600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7819374792814526e-05,
+      "loss": 2.6364,
+      "step": 44605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.780947645534825e-05,
+      "loss": 2.6498,
+      "step": 44610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7799578625807135e-05,
+      "loss": 2.5958,
+      "step": 44615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.778968130460354e-05,
+      "loss": 2.6151,
+      "step": 44620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.777978449214988e-05,
+      "loss": 2.6645,
+      "step": 44625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.776988818885849e-05,
+      "loss": 2.5576,
+      "step": 44630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7759992395141706e-05,
+      "loss": 2.6998,
+      "step": 44635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.775009711141187e-05,
+      "loss": 2.6601,
+      "step": 44640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7740202338081234e-05,
+      "loss": 2.7405,
+      "step": 44645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.773030807556209e-05,
+      "loss": 2.6572,
+      "step": 44650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.772041432426668e-05,
+      "loss": 2.7158,
+      "step": 44655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7710521084607236e-05,
+      "loss": 2.8055,
+      "step": 44660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7700628356995994e-05,
+      "loss": 2.5442,
+      "step": 44665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7690736141845086e-05,
+      "loss": 2.6996,
+      "step": 44670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.768084443956673e-05,
+      "loss": 2.548,
+      "step": 44675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7670953250573024e-05,
+      "loss": 2.6351,
+      "step": 44680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.766106257527612e-05,
+      "loss": 2.6297,
+      "step": 44685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7651172414088095e-05,
+      "loss": 2.5843,
+      "step": 44690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.764128276742105e-05,
+      "loss": 2.7591,
+      "step": 44695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7631393635687054e-05,
+      "loss": 2.6996,
+      "step": 44700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.76215050192981e-05,
+      "loss": 2.6695,
+      "step": 44705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.761161691866625e-05,
+      "loss": 2.6896,
+      "step": 44710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.760172933420345e-05,
+      "loss": 2.6356,
+      "step": 44715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.75918422663217e-05,
+      "loss": 2.7454,
+      "step": 44720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.758195571543295e-05,
+      "loss": 2.5589,
+      "step": 44725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.757206968194913e-05,
+      "loss": 2.7154,
+      "step": 44730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7562184166282156e-05,
+      "loss": 2.6274,
+      "step": 44735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7552299168843875e-05,
+      "loss": 2.5739,
+      "step": 44740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.754241469004619e-05,
+      "loss": 2.6199,
+      "step": 44745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.753253073030093e-05,
+      "loss": 2.5808,
+      "step": 44750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.75226472900199e-05,
+      "loss": 2.6739,
+      "step": 44755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.751276436961494e-05,
+      "loss": 2.7993,
+      "step": 44760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.750288196949779e-05,
+      "loss": 2.6194,
+      "step": 44765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7493000090080234e-05,
+      "loss": 2.589,
+      "step": 44770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.748311873177397e-05,
+      "loss": 2.6221,
+      "step": 44775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.747323789499073e-05,
+      "loss": 2.7094,
+      "step": 44780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7463357580142224e-05,
+      "loss": 2.5629,
+      "step": 44785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.745347778764008e-05,
+      "loss": 2.5609,
+      "step": 44790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7443598517896e-05,
+      "loss": 2.6933,
+      "step": 44795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.743371977132154e-05,
+      "loss": 2.7171,
+      "step": 44800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.742384154832838e-05,
+      "loss": 2.6712,
+      "step": 44805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.741396384932804e-05,
+      "loss": 2.6434,
+      "step": 44810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.74040866747321e-05,
+      "loss": 2.7416,
+      "step": 44815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.73942100249521e-05,
+      "loss": 2.6867,
+      "step": 44820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.738433390039956e-05,
+      "loss": 2.5683,
+      "step": 44825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.737445830148599e-05,
+      "loss": 2.7952,
+      "step": 44830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7364583228622824e-05,
+      "loss": 2.6736,
+      "step": 44835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.735470868222154e-05,
+      "loss": 2.7589,
+      "step": 44840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.734483466269355e-05,
+      "loss": 2.5981,
+      "step": 44845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.733496117045028e-05,
+      "loss": 2.5506,
+      "step": 44850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.732508820590311e-05,
+      "loss": 2.5799,
+      "step": 44855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7315215769463383e-05,
+      "loss": 2.6243,
+      "step": 44860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.730534386154248e-05,
+      "loss": 2.6642,
+      "step": 44865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.729547248255166e-05,
+      "loss": 2.7138,
+      "step": 44870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.728560163290227e-05,
+      "loss": 2.6466,
+      "step": 44875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.727573131300557e-05,
+      "loss": 2.6549,
+      "step": 44880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.72658615232728e-05,
+      "loss": 2.6113,
+      "step": 44885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.725599226411522e-05,
+      "loss": 2.6691,
+      "step": 44890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7246123535944e-05,
+      "loss": 2.5161,
+      "step": 44895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.723625533917038e-05,
+      "loss": 2.7929,
+      "step": 44900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7226387674205454e-05,
+      "loss": 2.6048,
+      "step": 44905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.72165205414604e-05,
+      "loss": 2.5884,
+      "step": 44910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.720665394134635e-05,
+      "loss": 2.7147,
+      "step": 44915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.719678787427438e-05,
+      "loss": 2.5927,
+      "step": 44920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.718692234065559e-05,
+      "loss": 2.6847,
+      "step": 44925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7177057340901e-05,
+      "loss": 2.7053,
+      "step": 44930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.716719287542167e-05,
+      "loss": 2.6134,
+      "step": 44935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7157328944628586e-05,
+      "loss": 2.6188,
+      "step": 44940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7147465548932746e-05,
+      "loss": 2.7856,
+      "step": 44945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7137602688745136e-05,
+      "loss": 2.5784,
+      "step": 44950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7127740364476654e-05,
+      "loss": 2.6265,
+      "step": 44955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.711787857653825e-05,
+      "loss": 2.7394,
+      "step": 44960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7108017325340806e-05,
+      "loss": 2.6071,
+      "step": 44965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.70981566112952e-05,
+      "loss": 2.6334,
+      "step": 44970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7088296434812286e-05,
+      "loss": 2.7621,
+      "step": 44975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7078436796302896e-05,
+      "loss": 2.7611,
+      "step": 44980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7068577696177846e-05,
+      "loss": 2.5791,
+      "step": 44985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.70587191348479e-05,
+      "loss": 2.6194,
+      "step": 44990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.704886111272382e-05,
+      "loss": 2.61,
+      "step": 44995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.703900363021638e-05,
+      "loss": 2.8425,
+      "step": 45000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.702914668773626e-05,
+      "loss": 2.6177,
+      "step": 45005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.701929028569418e-05,
+      "loss": 2.5676,
+      "step": 45010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.700943442450079e-05,
+      "loss": 2.6572,
+      "step": 45015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.699957910456678e-05,
+      "loss": 2.7024,
+      "step": 45020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.698972432630272e-05,
+      "loss": 2.5845,
+      "step": 45025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.697987009011925e-05,
+      "loss": 2.5996,
+      "step": 45030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.697001639642697e-05,
+      "loss": 2.7432,
+      "step": 45035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.6960163245636395e-05,
+      "loss": 2.8824,
+      "step": 45040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.6950310638158104e-05,
+      "loss": 2.5934,
+      "step": 45045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.694045857440257e-05,
+      "loss": 2.5386,
+      "step": 45050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.693060705478034e-05,
+      "loss": 2.8408,
+      "step": 45055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.6920756079701816e-05,
+      "loss": 2.7518,
+      "step": 45060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.6910905649577485e-05,
+      "loss": 2.7021,
+      "step": 45065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.690105576481778e-05,
+      "loss": 2.6492,
+      "step": 45070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.689120642583307e-05,
+      "loss": 2.6539,
+      "step": 45075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.688135763303378e-05,
+      "loss": 2.761,
+      "step": 45080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.687150938683021e-05,
+      "loss": 2.7207,
+      "step": 45085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.686166168763271e-05,
+      "loss": 2.7757,
+      "step": 45090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.685181453585163e-05,
+      "loss": 2.7322,
+      "step": 45095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6841967931897206e-05,
+      "loss": 2.5919,
+      "step": 45100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6832121876179724e-05,
+      "loss": 2.7135,
+      "step": 45105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.682227636910942e-05,
+      "loss": 2.7268,
+      "step": 45110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6812431411096536e-05,
+      "loss": 2.6636,
+      "step": 45115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6802587002551226e-05,
+      "loss": 2.6758,
+      "step": 45120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.679274314388369e-05,
+      "loss": 2.6855,
+      "step": 45125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6782899835504084e-05,
+      "loss": 2.579,
+      "step": 45130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6773057077822505e-05,
+      "loss": 2.6415,
+      "step": 45135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.676321487124911e-05,
+      "loss": 2.5533,
+      "step": 45140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.675337321619391e-05,
+      "loss": 2.672,
+      "step": 45145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.674353211306701e-05,
+      "loss": 2.6451,
+      "step": 45150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6733691562278454e-05,
+      "loss": 2.5598,
+      "step": 45155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.672385156423822e-05,
+      "loss": 2.6997,
+      "step": 45160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6714012119356315e-05,
+      "loss": 2.6367,
+      "step": 45165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6704173228042695e-05,
+      "loss": 2.66,
+      "step": 45170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.669433489070733e-05,
+      "loss": 2.6727,
+      "step": 45175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.668449710776011e-05,
+      "loss": 2.6656,
+      "step": 45180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.667465987961093e-05,
+      "loss": 2.576,
+      "step": 45185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6664823206669694e-05,
+      "loss": 2.6385,
+      "step": 45190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6654987089346226e-05,
+      "loss": 2.6651,
+      "step": 45195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.664515152805037e-05,
+      "loss": 2.5388,
+      "step": 45200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.663531652319191e-05,
+      "loss": 2.6429,
+      "step": 45205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.662548207518063e-05,
+      "loss": 2.7273,
+      "step": 45210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6615648184426334e-05,
+      "loss": 2.9253,
+      "step": 45215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.660581485133869e-05,
+      "loss": 2.6589,
+      "step": 45220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.659598207632745e-05,
+      "loss": 2.7437,
+      "step": 45225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.658614985980229e-05,
+      "loss": 2.4894,
+      "step": 45230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.657631820217289e-05,
+      "loss": 2.5984,
+      "step": 45235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6566487103848856e-05,
+      "loss": 2.6187,
+      "step": 45240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.655665656523983e-05,
+      "loss": 2.641,
+      "step": 45245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6546826586755435e-05,
+      "loss": 2.7112,
+      "step": 45250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.653699716880519e-05,
+      "loss": 2.6134,
+      "step": 45255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.652716831179867e-05,
+      "loss": 2.6166,
+      "step": 45260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.651734001614539e-05,
+      "loss": 2.818,
+      "step": 45265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.650751228225488e-05,
+      "loss": 2.6602,
+      "step": 45270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.649768511053657e-05,
+      "loss": 2.703,
+      "step": 45275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.648785850139995e-05,
+      "loss": 2.7736,
+      "step": 45280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.647803245525445e-05,
+      "loss": 2.5521,
+      "step": 45285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6468206972509475e-05,
+      "loss": 2.6156,
+      "step": 45290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6458382053574394e-05,
+      "loss": 2.6449,
+      "step": 45295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.644855769885858e-05,
+      "loss": 2.6095,
+      "step": 45300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.643873390877136e-05,
+      "loss": 2.685,
+      "step": 45305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.642891068372208e-05,
+      "loss": 2.4958,
+      "step": 45310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.641908802411999e-05,
+      "loss": 2.5561,
+      "step": 45315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.640926593037438e-05,
+      "loss": 2.6652,
+      "step": 45320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.639944440289448e-05,
+      "loss": 2.6322,
+      "step": 45325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6389623442089525e-05,
+      "loss": 2.6638,
+      "step": 45330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.63798030483687e-05,
+      "loss": 2.7177,
+      "step": 45335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6369983222141166e-05,
+      "loss": 2.7122,
+      "step": 45340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6360163963816114e-05,
+      "loss": 2.7392,
+      "step": 45345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6350345273802614e-05,
+      "loss": 2.6541,
+      "step": 45350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.634052715250981e-05,
+      "loss": 2.595,
+      "step": 45355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.633070960034675e-05,
+      "loss": 2.6585,
+      "step": 45360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.63208926177225e-05,
+      "loss": 2.6784,
+      "step": 45365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.631107620504612e-05,
+      "loss": 2.558,
+      "step": 45370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.630126036272656e-05,
+      "loss": 2.6974,
+      "step": 45375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.629144509117285e-05,
+      "loss": 2.6798,
+      "step": 45380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.628163039079391e-05,
+      "loss": 2.7005,
+      "step": 45385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.627181626199871e-05,
+      "loss": 2.6386,
+      "step": 45390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.626200270519614e-05,
+      "loss": 2.6951,
+      "step": 45395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.625218972079509e-05,
+      "loss": 2.6295,
+      "step": 45400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6242377309204445e-05,
+      "loss": 2.7093,
+      "step": 45405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6232565470833e-05,
+      "loss": 2.6404,
+      "step": 45410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6222754206089624e-05,
+      "loss": 2.5253,
+      "step": 45415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6212943515383056e-05,
+      "loss": 2.5321,
+      "step": 45420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6203133399122104e-05,
+      "loss": 2.7761,
+      "step": 45425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.619332385771551e-05,
+      "loss": 2.6204,
+      "step": 45430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.618351489157197e-05,
+      "loss": 2.6872,
+      "step": 45435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.617370650110021e-05,
+      "loss": 2.6131,
+      "step": 45440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6163898686708864e-05,
+      "loss": 2.5648,
+      "step": 45445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6154091448806616e-05,
+      "loss": 2.5215,
+      "step": 45450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.614428478780206e-05,
+      "loss": 2.7717,
+      "step": 45455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.613447870410381e-05,
+      "loss": 2.809,
+      "step": 45460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6124673198120474e-05,
+      "loss": 2.6658,
+      "step": 45465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.611486827026055e-05,
+      "loss": 2.6821,
+      "step": 45470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.61050639209326e-05,
+      "loss": 2.6615,
+      "step": 45475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.609526015054511e-05,
+      "loss": 2.7351,
+      "step": 45480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.608545695950658e-05,
+      "loss": 2.4531,
+      "step": 45485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.607565434822544e-05,
+      "loss": 2.5723,
+      "step": 45490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.606585231711013e-05,
+      "loss": 2.6752,
+      "step": 45495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.605605086656909e-05,
+      "loss": 2.7098,
+      "step": 45500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.604624999701065e-05,
+      "loss": 2.6857,
+      "step": 45505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.60364497088432e-05,
+      "loss": 2.7324,
+      "step": 45510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.602665000247506e-05,
+      "loss": 2.6073,
+      "step": 45515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.601685087831455e-05,
+      "loss": 2.6627,
+      "step": 45520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.600705233676998e-05,
+      "loss": 2.6131,
+      "step": 45525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.599725437824956e-05,
+      "loss": 2.5883,
+      "step": 45530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.598745700316157e-05,
+      "loss": 2.5695,
+      "step": 45535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5977660211914196e-05,
+      "loss": 2.6698,
+      "step": 45540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.596786400491565e-05,
+      "loss": 2.5669,
+      "step": 45545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5958068382574075e-05,
+      "loss": 2.5022,
+      "step": 45550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.594827334529761e-05,
+      "loss": 2.605,
+      "step": 45555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.593847889349441e-05,
+      "loss": 2.6381,
+      "step": 45560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5928685027572514e-05,
+      "loss": 2.8011,
+      "step": 45565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.591889174794003e-05,
+      "loss": 2.6337,
+      "step": 45570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.590909905500497e-05,
+      "loss": 2.5976,
+      "step": 45575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.589930694917537e-05,
+      "loss": 2.7259,
+      "step": 45580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.588951543085924e-05,
+      "loss": 2.7316,
+      "step": 45585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5879724500464504e-05,
+      "loss": 2.6609,
+      "step": 45590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.586993415839915e-05,
+      "loss": 2.6935,
+      "step": 45595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.586014440507107e-05,
+      "loss": 2.5631,
+      "step": 45600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5850355240888176e-05,
+      "loss": 2.7744,
+      "step": 45605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5840566666258326e-05,
+      "loss": 2.7104,
+      "step": 45610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.583077868158936e-05,
+      "loss": 2.6219,
+      "step": 45615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.582099128728915e-05,
+      "loss": 2.587,
+      "step": 45620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5811204483765424e-05,
+      "loss": 2.6326,
+      "step": 45625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.580141827142602e-05,
+      "loss": 2.7824,
+      "step": 45630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.579163265067862e-05,
+      "loss": 2.5846,
+      "step": 45635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.578184762193099e-05,
+      "loss": 2.57,
+      "step": 45640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.577206318559083e-05,
+      "loss": 2.5545,
+      "step": 45645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.576227934206579e-05,
+      "loss": 2.7831,
+      "step": 45650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5752496091763554e-05,
+      "loss": 2.6837,
+      "step": 45655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5742713435091716e-05,
+      "loss": 2.689,
+      "step": 45660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5732931372457886e-05,
+      "loss": 2.7337,
+      "step": 45665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.572314990426963e-05,
+      "loss": 2.576,
+      "step": 45670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.571336903093451e-05,
+      "loss": 2.6642,
+      "step": 45675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.570358875286007e-05,
+      "loss": 2.6653,
+      "step": 45680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.569380907045378e-05,
+      "loss": 2.6338,
+      "step": 45685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5684029984123134e-05,
+      "loss": 2.6979,
+      "step": 45690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.567425149427556e-05,
+      "loss": 2.6465,
+      "step": 45695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5664473601318505e-05,
+      "loss": 2.7345,
+      "step": 45700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5654696305659366e-05,
+      "loss": 2.7408,
+      "step": 45705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.564491960770551e-05,
+      "loss": 2.5938,
+      "step": 45710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5635143507864324e-05,
+      "loss": 2.5983,
+      "step": 45715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.562536800654309e-05,
+      "loss": 2.7451,
+      "step": 45720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5615593104149144e-05,
+      "loss": 2.5757,
+      "step": 45725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.560581880108973e-05,
+      "loss": 2.6026,
+      "step": 45730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.559604509777212e-05,
+      "loss": 2.6768,
+      "step": 45735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.558627199460355e-05,
+      "loss": 2.7181,
+      "step": 45740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.557649949199119e-05,
+      "loss": 2.6373,
+      "step": 45745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.556672759034227e-05,
+      "loss": 2.671,
+      "step": 45750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.555695629006388e-05,
+      "loss": 2.5641,
+      "step": 45755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.554718559156319e-05,
+      "loss": 2.3984,
+      "step": 45760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.553741549524729e-05,
+      "loss": 2.6988,
+      "step": 45765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.552764600152323e-05,
+      "loss": 2.7597,
+      "step": 45770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.551787711079813e-05,
+      "loss": 2.5979,
+      "step": 45775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5508108823478935e-05,
+      "loss": 2.7238,
+      "step": 45780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5498341139972714e-05,
+      "loss": 2.5632,
+      "step": 45785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.548857406068639e-05,
+      "loss": 2.6546,
+      "step": 45790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.547880758602693e-05,
+      "loss": 2.7192,
+      "step": 45795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5469041716401284e-05,
+      "loss": 2.7447,
+      "step": 45800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.545927645221633e-05,
+      "loss": 2.7348,
+      "step": 45805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.544951179387896e-05,
+      "loss": 2.5194,
+      "step": 45810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.543974774179599e-05,
+      "loss": 2.5899,
+      "step": 45815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5429984296374295e-05,
+      "loss": 2.6747,
+      "step": 45820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.542022145802063e-05,
+      "loss": 2.7134,
+      "step": 45825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.541045922714178e-05,
+      "loss": 2.6027,
+      "step": 45830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5400697604144515e-05,
+      "loss": 2.6763,
+      "step": 45835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.539093658943552e-05,
+      "loss": 2.6755,
+      "step": 45840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5381176183421556e-05,
+      "loss": 2.6063,
+      "step": 45845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.537141638650923e-05,
+      "loss": 2.651,
+      "step": 45850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.536165719910521e-05,
+      "loss": 2.6721,
+      "step": 45855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.535189862161614e-05,
+      "loss": 2.6958,
+      "step": 45860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.534214065444859e-05,
+      "loss": 2.6028,
+      "step": 45865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.533238329800916e-05,
+      "loss": 2.5874,
+      "step": 45870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.532262655270437e-05,
+      "loss": 2.7192,
+      "step": 45875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.531287041894076e-05,
+      "loss": 2.6595,
+      "step": 45880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.53031148971248e-05,
+      "loss": 2.6892,
+      "step": 45885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5293359987662966e-05,
+      "loss": 2.5394,
+      "step": 45890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.528360569096172e-05,
+      "loss": 2.7044,
+      "step": 45895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.527385200742746e-05,
+      "loss": 2.6173,
+      "step": 45900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5264098937466604e-05,
+      "loss": 2.6793,
+      "step": 45905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.525434648148549e-05,
+      "loss": 2.8585,
+      "step": 45910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.524459463989049e-05,
+      "loss": 2.6666,
+      "step": 45915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.523484341308789e-05,
+      "loss": 2.6638,
+      "step": 45920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.522509280148399e-05,
+      "loss": 2.8183,
+      "step": 45925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.521534280548506e-05,
+      "loss": 2.7384,
+      "step": 45930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.520559342549734e-05,
+      "loss": 2.7005,
+      "step": 45935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5195844661927056e-05,
+      "loss": 2.6773,
+      "step": 45940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.518609651518036e-05,
+      "loss": 2.6332,
+      "step": 45945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.517634898566343e-05,
+      "loss": 2.5555,
+      "step": 45950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5166602073782426e-05,
+      "loss": 2.618,
+      "step": 45955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.515685577994343e-05,
+      "loss": 2.5739,
+      "step": 45960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.514711010455256e-05,
+      "loss": 2.5699,
+      "step": 45965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.513736504801582e-05,
+      "loss": 2.6166,
+      "step": 45970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5127620610739304e-05,
+      "loss": 2.5273,
+      "step": 45975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.511787679312897e-05,
+      "loss": 2.6786,
+      "step": 45980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.510813359559082e-05,
+      "loss": 2.7139,
+      "step": 45985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5098391018530816e-05,
+      "loss": 2.6081,
+      "step": 45990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.508864906235488e-05,
+      "loss": 2.6819,
+      "step": 45995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.507890772746894e-05,
+      "loss": 2.6166,
+      "step": 46000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.506916701427883e-05,
+      "loss": 2.704,
+      "step": 46005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.505942692319043e-05,
+      "loss": 2.6277,
+      "step": 46010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.504968745460957e-05,
+      "loss": 2.6169,
+      "step": 46015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.503994860894203e-05,
+      "loss": 2.6107,
+      "step": 46020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.503021038659361e-05,
+      "loss": 2.7164,
+      "step": 46025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5020472787970036e-05,
+      "loss": 2.6469,
+      "step": 46030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5010735813477055e-05,
+      "loss": 2.4589,
+      "step": 46035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.500099946352033e-05,
+      "loss": 2.624,
+      "step": 46040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.499126373850556e-05,
+      "loss": 2.609,
+      "step": 46045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.498152863883838e-05,
+      "loss": 2.6054,
+      "step": 46050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.49717941649244e-05,
+      "loss": 2.6337,
+      "step": 46055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.496206031716925e-05,
+      "loss": 2.7555,
+      "step": 46060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4952327095978436e-05,
+      "loss": 2.6748,
+      "step": 46065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4942594501757534e-05,
+      "loss": 2.7146,
+      "step": 46070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.493286253491208e-05,
+      "loss": 2.8404,
+      "step": 46075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.492313119584751e-05,
+      "loss": 2.6151,
+      "step": 46080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.491340048496932e-05,
+      "loss": 2.6672,
+      "step": 46085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.490367040268293e-05,
+      "loss": 2.716,
+      "step": 46090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.489394094939378e-05,
+      "loss": 2.7548,
+      "step": 46095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.48842121255072e-05,
+      "loss": 2.6345,
+      "step": 46100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4874483931428586e-05,
+      "loss": 2.6237,
+      "step": 46105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4864756367563264e-05,
+      "loss": 2.6594,
+      "step": 46110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.485502943431655e-05,
+      "loss": 2.6496,
+      "step": 46115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.484530313209369e-05,
+      "loss": 2.7446,
+      "step": 46120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.483557746129994e-05,
+      "loss": 2.6719,
+      "step": 46125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4825852422340565e-05,
+      "loss": 2.8075,
+      "step": 46130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.481612801562072e-05,
+      "loss": 2.6622,
+      "step": 46135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.480640424154559e-05,
+      "loss": 2.779,
+      "step": 46140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.479668110052035e-05,
+      "loss": 2.5966,
+      "step": 46145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.478695859295007e-05,
+      "loss": 2.6766,
+      "step": 46150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4777236719239895e-05,
+      "loss": 2.6443,
+      "step": 46155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.476751547979486e-05,
+      "loss": 2.6952,
+      "step": 46160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.475779487502e-05,
+      "loss": 2.5946,
+      "step": 46165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.474807490532037e-05,
+      "loss": 2.6477,
+      "step": 46170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.473835557110091e-05,
+      "loss": 2.7933,
+      "step": 46175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.472863687276662e-05,
+      "loss": 2.7434,
+      "step": 46180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4718918810722406e-05,
+      "loss": 2.6174,
+      "step": 46185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.470920138537321e-05,
+      "loss": 2.6851,
+      "step": 46190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.469948459712388e-05,
+      "loss": 2.7422,
+      "step": 46195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.468976844637927e-05,
+      "loss": 2.6386,
+      "step": 46200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4680052933544246e-05,
+      "loss": 2.6048,
+      "step": 46205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.467033805902358e-05,
+      "loss": 2.6243,
+      "step": 46210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4660623823222066e-05,
+      "loss": 2.6166,
+      "step": 46215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.465091022654443e-05,
+      "loss": 2.6864,
+      "step": 46220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.464119726939541e-05,
+      "loss": 2.5486,
+      "step": 46225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.463148495217971e-05,
+      "loss": 2.5596,
+      "step": 46230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4621773275301985e-05,
+      "loss": 2.5816,
+      "step": 46235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4612062239166884e-05,
+      "loss": 2.7027,
+      "step": 46240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.460235184417901e-05,
+      "loss": 2.6246,
+      "step": 46245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4592642090742985e-05,
+      "loss": 2.4011,
+      "step": 46250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4582932979263326e-05,
+      "loss": 2.6195,
+      "step": 46255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4573224510144595e-05,
+      "loss": 2.8016,
+      "step": 46260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.456351668379132e-05,
+      "loss": 2.7886,
+      "step": 46265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4553809500607936e-05,
+      "loss": 2.529,
+      "step": 46270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.454410296099894e-05,
+      "loss": 2.6363,
+      "step": 46275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.453439706536873e-05,
+      "loss": 2.5634,
+      "step": 46280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4524691814121726e-05,
+      "loss": 2.6144,
+      "step": 46285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4514987207662306e-05,
+      "loss": 2.6278,
+      "step": 46290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4505283246394807e-05,
+      "loss": 2.5347,
+      "step": 46295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4495579930723555e-05,
+      "loss": 2.7512,
+      "step": 46300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4485877261052835e-05,
+      "loss": 2.7739,
+      "step": 46305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.447617523778695e-05,
+      "loss": 2.6271,
+      "step": 46310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4466473861330076e-05,
+      "loss": 2.7565,
+      "step": 46315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.445677313208647e-05,
+      "loss": 2.6252,
+      "step": 46320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.444707305046033e-05,
+      "loss": 2.5828,
+      "step": 46325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4437373616855785e-05,
+      "loss": 2.5501,
+      "step": 46330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.442767483167698e-05,
+      "loss": 2.5507,
+      "step": 46335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.441797669532802e-05,
+      "loss": 2.673,
+      "step": 46340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4408279208212994e-05,
+      "loss": 2.7084,
+      "step": 46345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.439858237073593e-05,
+      "loss": 2.6978,
+      "step": 46350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4388886183300864e-05,
+      "loss": 2.5229,
+      "step": 46355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4379190646311804e-05,
+      "loss": 2.7889,
+      "step": 46360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4369495760172697e-05,
+      "loss": 2.6145,
+      "step": 46365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.435980152528751e-05,
+      "loss": 2.628,
+      "step": 46370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.435010794206014e-05,
+      "loss": 2.8083,
+      "step": 46375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.434041501089448e-05,
+      "loss": 2.7272,
+      "step": 46380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.433072273219441e-05,
+      "loss": 2.6188,
+      "step": 46385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.432103110636374e-05,
+      "loss": 2.6805,
+      "step": 46390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.431134013380629e-05,
+      "loss": 2.5092,
+      "step": 46395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.430164981492583e-05,
+      "loss": 2.5672,
+      "step": 46400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4291960150126134e-05,
+      "loss": 2.6551,
+      "step": 46405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4282271139810895e-05,
+      "loss": 2.5483,
+      "step": 46410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4272582784383834e-05,
+      "loss": 2.696,
+      "step": 46415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4262895084248634e-05,
+      "loss": 2.5954,
+      "step": 46420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.42532080398089e-05,
+      "loss": 2.65,
+      "step": 46425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.424352165146828e-05,
+      "loss": 2.5324,
+      "step": 46430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.423383591963034e-05,
+      "loss": 2.629,
+      "step": 46435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.422415084469866e-05,
+      "loss": 2.6015,
+      "step": 46440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4214466427076786e-05,
+      "loss": 2.7588,
+      "step": 46445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.420478266716818e-05,
+      "loss": 2.5812,
+      "step": 46450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.419509956537638e-05,
+      "loss": 2.589,
+      "step": 46455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.418541712210478e-05,
+      "loss": 2.6774,
+      "step": 46460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.417573533775684e-05,
+      "loss": 2.7114,
+      "step": 46465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.416605421273594e-05,
+      "loss": 2.5864,
+      "step": 46470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.415637374744546e-05,
+      "loss": 2.7789,
+      "step": 46475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.414669394228875e-05,
+      "loss": 2.4948,
+      "step": 46480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.413701479766911e-05,
+      "loss": 2.7199,
+      "step": 46485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4127336313989834e-05,
+      "loss": 2.6319,
+      "step": 46490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.411765849165417e-05,
+      "loss": 2.7293,
+      "step": 46495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4107981331065356e-05,
+      "loss": 2.5565,
+      "step": 46500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4098304832626614e-05,
+      "loss": 2.6205,
+      "step": 46505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.40886289967411e-05,
+      "loss": 2.5547,
+      "step": 46510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.407895382381198e-05,
+      "loss": 2.705,
+      "step": 46515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.406927931424235e-05,
+      "loss": 2.6989,
+      "step": 46520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.405960546843533e-05,
+      "loss": 2.5929,
+      "step": 46525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.404993228679396e-05,
+      "loss": 2.6036,
+      "step": 46530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.40402597697213e-05,
+      "loss": 2.6217,
+      "step": 46535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4030587917620374e-05,
+      "loss": 2.6095,
+      "step": 46540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.402091673089412e-05,
+      "loss": 2.571,
+      "step": 46545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.401124620994555e-05,
+      "loss": 2.6937,
+      "step": 46550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4001576355177534e-05,
+      "loss": 2.5501,
+      "step": 46555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.399190716699301e-05,
+      "loss": 2.7776,
+      "step": 46560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.398223864579483e-05,
+      "loss": 2.7017,
+      "step": 46565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.397257079198586e-05,
+      "loss": 2.7093,
+      "step": 46570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.396290360596891e-05,
+      "loss": 2.7683,
+      "step": 46575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.395323708814675e-05,
+      "loss": 2.5779,
+      "step": 46580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.394357123892217e-05,
+      "loss": 2.7112,
+      "step": 46585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.393390605869787e-05,
+      "loss": 2.5918,
+      "step": 46590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.392424154787658e-05,
+      "loss": 2.7333,
+      "step": 46595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.391457770686098e-05,
+      "loss": 2.7408,
+      "step": 46600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.39049145360537e-05,
+      "loss": 2.518,
+      "step": 46605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.3895252035857394e-05,
+      "loss": 2.6503,
+      "step": 46610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.388559020667461e-05,
+      "loss": 2.6442,
+      "step": 46615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3875929048907945e-05,
+      "loss": 2.6816,
+      "step": 46620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.386626856295993e-05,
+      "loss": 2.6671,
+      "step": 46625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.385660874923305e-05,
+      "loss": 2.5932,
+      "step": 46630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.384694960812985e-05,
+      "loss": 2.6591,
+      "step": 46635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.383729114005272e-05,
+      "loss": 2.6766,
+      "step": 46640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3827633345404117e-05,
+      "loss": 2.5546,
+      "step": 46645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3817976224586424e-05,
+      "loss": 2.7288,
+      "step": 46650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.380831977800202e-05,
+      "loss": 2.7999,
+      "step": 46655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.379866400605325e-05,
+      "loss": 2.6076,
+      "step": 46660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3789008909142404e-05,
+      "loss": 2.6671,
+      "step": 46665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.377935448767181e-05,
+      "loss": 2.7343,
+      "step": 46670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.376970074204368e-05,
+      "loss": 2.7128,
+      "step": 46675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.376004767266027e-05,
+      "loss": 2.6491,
+      "step": 46680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.375039527992376e-05,
+      "loss": 2.7534,
+      "step": 46685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.374074356423634e-05,
+      "loss": 2.5687,
+      "step": 46690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.373109252600016e-05,
+      "loss": 2.5933,
+      "step": 46695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.372144216561732e-05,
+      "loss": 2.7656,
+      "step": 46700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3711792483489926e-05,
+      "loss": 2.5953,
+      "step": 46705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.370214348002e-05,
+      "loss": 2.5769,
+      "step": 46710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3692495155609605e-05,
+      "loss": 2.7238,
+      "step": 46715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3682847510660745e-05,
+      "loss": 2.5474,
+      "step": 46720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.367320054557538e-05,
+      "loss": 2.6495,
+      "step": 46725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3663554260755476e-05,
+      "loss": 2.6851,
+      "step": 46730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.365390865660292e-05,
+      "loss": 2.6796,
+      "step": 46735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.364426373351964e-05,
+      "loss": 2.6091,
+      "step": 46740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.363461949190746e-05,
+      "loss": 2.717,
+      "step": 46745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3624975932168224e-05,
+      "loss": 2.6646,
+      "step": 46750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.361533305470376e-05,
+      "loss": 2.5544,
+      "step": 46755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3605690859915805e-05,
+      "loss": 2.6413,
+      "step": 46760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.359604934820616e-05,
+      "loss": 2.5763,
+      "step": 46765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.358640851997649e-05,
+      "loss": 2.595,
+      "step": 46770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.357676837562851e-05,
+      "loss": 2.6565,
+      "step": 46775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.356712891556387e-05,
+      "loss": 2.6987,
+      "step": 46780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.355749014018421e-05,
+      "loss": 2.6286,
+      "step": 46785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.354785204989117e-05,
+      "loss": 2.5546,
+      "step": 46790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.353821464508626e-05,
+      "loss": 2.579,
+      "step": 46795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.35285779261711e-05,
+      "loss": 2.6357,
+      "step": 46800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.351894189354714e-05,
+      "loss": 2.6536,
+      "step": 46805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3509306547615905e-05,
+      "loss": 2.5053,
+      "step": 46810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.349967188877887e-05,
+      "loss": 2.4914,
+      "step": 46815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.349003791743744e-05,
+      "loss": 2.6823,
+      "step": 46820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.348040463399306e-05,
+      "loss": 2.6535,
+      "step": 46825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.347077203884705e-05,
+      "loss": 2.6667,
+      "step": 46830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3461140132400804e-05,
+      "loss": 2.5742,
+      "step": 46835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.345150891505562e-05,
+      "loss": 2.6244,
+      "step": 46840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3441878387212786e-05,
+      "loss": 2.4802,
+      "step": 46845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.343224854927358e-05,
+      "loss": 2.8307,
+      "step": 46850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.342261940163921e-05,
+      "loss": 2.6059,
+      "step": 46855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.341299094471092e-05,
+      "loss": 2.6558,
+      "step": 46860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.340336317888983e-05,
+      "loss": 2.654,
+      "step": 46865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.339373610457713e-05,
+      "loss": 2.6749,
+      "step": 46870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.338410972217393e-05,
+      "loss": 2.7414,
+      "step": 46875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.33744840320813e-05,
+      "loss": 2.6495,
+      "step": 46880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3364859034700335e-05,
+      "loss": 2.5558,
+      "step": 46885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3355234730432027e-05,
+      "loss": 2.6862,
+      "step": 46890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.334561111967741e-05,
+      "loss": 2.661,
+      "step": 46895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.333598820283742e-05,
+      "loss": 2.4736,
+      "step": 46900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.332636598031302e-05,
+      "loss": 2.4246,
+      "step": 46905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.331674445250515e-05,
+      "loss": 2.7412,
+      "step": 46910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.330712361981466e-05,
+      "loss": 2.624,
+      "step": 46915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.329750348264244e-05,
+      "loss": 2.5516,
+      "step": 46920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.328788404138928e-05,
+      "loss": 2.6521,
+      "step": 46925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.327826529645601e-05,
+      "loss": 2.6788,
+      "step": 46930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.326864724824339e-05,
+      "loss": 2.6651,
+      "step": 46935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.325902989715217e-05,
+      "loss": 2.6733,
+      "step": 46940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.324941324358305e-05,
+      "loss": 2.5385,
+      "step": 46945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.323979728793671e-05,
+      "loss": 2.6979,
+      "step": 46950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.323018203061384e-05,
+      "loss": 2.7156,
+      "step": 46955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.322056747201501e-05,
+      "loss": 2.7146,
+      "step": 46960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.321095361254084e-05,
+      "loss": 2.5969,
+      "step": 46965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.320134045259192e-05,
+      "loss": 2.4815,
+      "step": 46970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.319172799256876e-05,
+      "loss": 2.7367,
+      "step": 46975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3182116232871894e-05,
+      "loss": 2.7321,
+      "step": 46980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.317250517390177e-05,
+      "loss": 2.6263,
+      "step": 46985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.316289481605887e-05,
+      "loss": 2.5017,
+      "step": 46990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.315328515974357e-05,
+      "loss": 2.5836,
+      "step": 46995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.31436762053563e-05,
+      "loss": 2.5826,
+      "step": 47000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.313406795329741e-05,
+      "loss": 2.6306,
+      "step": 47005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.312446040396724e-05,
+      "loss": 2.6919,
+      "step": 47010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.311485355776611e-05,
+      "loss": 2.792,
+      "step": 47015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.310524741509425e-05,
+      "loss": 2.6478,
+      "step": 47020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3095641976351935e-05,
+      "loss": 2.6534,
+      "step": 47025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.308603724193937e-05,
+      "loss": 2.5798,
+      "step": 47030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3076433212256764e-05,
+      "loss": 2.6172,
+      "step": 47035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.306682988770426e-05,
+      "loss": 2.6696,
+      "step": 47040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.305722726868197e-05,
+      "loss": 2.6082,
+      "step": 47045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.304762535559003e-05,
+      "loss": 2.6639,
+      "step": 47050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.303802414882846e-05,
+      "loss": 2.7083,
+      "step": 47055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.302842364879733e-05,
+      "loss": 2.5135,
+      "step": 47060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.301882385589665e-05,
+      "loss": 2.7213,
+      "step": 47065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.300922477052639e-05,
+      "loss": 2.6678,
+      "step": 47070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2999626393086537e-05,
+      "loss": 2.5854,
+      "step": 47075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.299002872397696e-05,
+      "loss": 2.7379,
+      "step": 47080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.298043176359758e-05,
+      "loss": 2.6252,
+      "step": 47085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.297083551234827e-05,
+      "loss": 2.6699,
+      "step": 47090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.296123997062883e-05,
+      "loss": 2.7242,
+      "step": 47095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2951645138839096e-05,
+      "loss": 2.5529,
+      "step": 47100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2942051017378816e-05,
+      "loss": 2.6465,
+      "step": 47105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.293245760664777e-05,
+      "loss": 2.6339,
+      "step": 47110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.292286490704562e-05,
+      "loss": 2.73,
+      "step": 47115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.29132729189721e-05,
+      "loss": 2.6107,
+      "step": 47120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2903681642826845e-05,
+      "loss": 2.6133,
+      "step": 47125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.289409107900948e-05,
+      "loss": 2.7137,
+      "step": 47130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.288450122791962e-05,
+      "loss": 2.6635,
+      "step": 47135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.28749120899568e-05,
+      "loss": 2.6626,
+      "step": 47140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.286532366552058e-05,
+      "loss": 2.5956,
+      "step": 47145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.285573595501047e-05,
+      "loss": 2.8226,
+      "step": 47150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2846148958825916e-05,
+      "loss": 2.5922,
+      "step": 47155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.283656267736641e-05,
+      "loss": 2.7797,
+      "step": 47160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2826977111031335e-05,
+      "loss": 2.6875,
+      "step": 47165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.28173922602201e-05,
+      "loss": 2.7471,
+      "step": 47170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2807808125332046e-05,
+      "loss": 2.6723,
+      "step": 47175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.279822470676651e-05,
+      "loss": 2.6304,
+      "step": 47180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.278864200492281e-05,
+      "loss": 2.7324,
+      "step": 47185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.277906002020017e-05,
+      "loss": 2.7296,
+      "step": 47190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2769478752997875e-05,
+      "loss": 2.6278,
+      "step": 47195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.27598982037151e-05,
+      "loss": 2.6392,
+      "step": 47200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.275031837275105e-05,
+      "loss": 2.6502,
+      "step": 47205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.274073926050485e-05,
+      "loss": 2.6302,
+      "step": 47210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.273116086737563e-05,
+      "loss": 2.6761,
+      "step": 47215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2721583193762495e-05,
+      "loss": 2.5128,
+      "step": 47220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2712006240064476e-05,
+      "loss": 2.6655,
+      "step": 47225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.270243000668063e-05,
+      "loss": 2.6273,
+      "step": 47230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.269285449400993e-05,
+      "loss": 2.5109,
+      "step": 47235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.268327970245136e-05,
+      "loss": 2.6399,
+      "step": 47240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.267370563240387e-05,
+      "loss": 2.6388,
+      "step": 47245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2664132284266336e-05,
+      "loss": 2.5988,
+      "step": 47250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.265455965843767e-05,
+      "loss": 2.5265,
+      "step": 47255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2644987755316695e-05,
+      "loss": 2.6401,
+      "step": 47260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.263541657530227e-05,
+      "loss": 2.5553,
+      "step": 47265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2625846118793126e-05,
+      "loss": 2.5807,
+      "step": 47270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.261627638618806e-05,
+      "loss": 2.596,
+      "step": 47275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2606707377885806e-05,
+      "loss": 2.6917,
+      "step": 47280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.259713909428503e-05,
+      "loss": 2.6365,
+      "step": 47285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2587571535784436e-05,
+      "loss": 2.5375,
+      "step": 47290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.257800470278262e-05,
+      "loss": 2.5966,
+      "step": 47295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2568438595678226e-05,
+      "loss": 2.5827,
+      "step": 47300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.255887321486983e-05,
+      "loss": 2.6955,
+      "step": 47305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2549308560755956e-05,
+      "loss": 2.885,
+      "step": 47310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.253974463373513e-05,
+      "loss": 2.6495,
+      "step": 47315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.253018143420585e-05,
+      "loss": 2.6581,
+      "step": 47320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.252061896256657e-05,
+      "loss": 2.6089,
+      "step": 47325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.25110572192157e-05,
+      "loss": 2.6016,
+      "step": 47330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.250149620455164e-05,
+      "loss": 2.712,
+      "step": 47335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.249193591897278e-05,
+      "loss": 2.4563,
+      "step": 47340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.248237636287741e-05,
+      "loss": 2.7061,
+      "step": 47345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.247281753666388e-05,
+      "loss": 2.5644,
+      "step": 47350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.246325944073043e-05,
+      "loss": 2.5485,
+      "step": 47355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2453702075475314e-05,
+      "loss": 2.6011,
+      "step": 47360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.244414544129677e-05,
+      "loss": 2.6469,
+      "step": 47365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.243458953859293e-05,
+      "loss": 2.5876,
+      "step": 47370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.242503436776201e-05,
+      "loss": 2.6576,
+      "step": 47375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2415479929202066e-05,
+      "loss": 2.6885,
+      "step": 47380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2405926223311236e-05,
+      "loss": 2.6962,
+      "step": 47385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2396373250487544e-05,
+      "loss": 2.692,
+      "step": 47390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.238682101112904e-05,
+      "loss": 2.7293,
+      "step": 47395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.237726950563376e-05,
+      "loss": 2.6168,
+      "step": 47400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2367718734399596e-05,
+      "loss": 2.7604,
+      "step": 47405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2358168697824553e-05,
+      "loss": 2.7184,
+      "step": 47410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.23486193963065e-05,
+      "loss": 2.7215,
+      "step": 47415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2339070830243346e-05,
+      "loss": 2.7131,
+      "step": 47420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.23295230000329e-05,
+      "loss": 2.7395,
+      "step": 47425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.231997590607299e-05,
+      "loss": 2.6091,
+      "step": 47430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2310429548761436e-05,
+      "loss": 2.8148,
+      "step": 47435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2300883928495936e-05,
+      "loss": 2.5592,
+      "step": 47440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.229133904567425e-05,
+      "loss": 2.6603,
+      "step": 47445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.228179490069406e-05,
+      "loss": 2.681,
+      "step": 47450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.227225149395303e-05,
+      "loss": 2.6692,
+      "step": 47455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.226270882584881e-05,
+      "loss": 2.6932,
+      "step": 47460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.225316689677896e-05,
+      "loss": 2.7006,
+      "step": 47465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.22436257071411e-05,
+      "loss": 2.7706,
+      "step": 47470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.223408525733271e-05,
+      "loss": 2.6545,
+      "step": 47475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.222454554775135e-05,
+      "loss": 2.8159,
+      "step": 47480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.221500657879447e-05,
+      "loss": 2.727,
+      "step": 47485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.220546835085952e-05,
+      "loss": 2.6112,
+      "step": 47490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2195930864343946e-05,
+      "loss": 2.6488,
+      "step": 47495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.218639411964508e-05,
+      "loss": 2.7077,
+      "step": 47500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2176858117160315e-05,
+      "loss": 2.6257,
+      "step": 47505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.216732285728695e-05,
+      "loss": 2.602,
+      "step": 47510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.215778834042229e-05,
+      "loss": 2.7405,
+      "step": 47515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2148254566963624e-05,
+      "loss": 2.6731,
+      "step": 47520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.213872153730812e-05,
+      "loss": 2.5764,
+      "step": 47525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2129189251853037e-05,
+      "loss": 2.676,
+      "step": 47530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.21196577109955e-05,
+      "loss": 2.6937,
+      "step": 47535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.211012691513267e-05,
+      "loss": 2.7155,
+      "step": 47540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.210059686466163e-05,
+      "loss": 2.6145,
+      "step": 47545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.209106755997947e-05,
+      "loss": 2.6431,
+      "step": 47550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2081539001483264e-05,
+      "loss": 2.554,
+      "step": 47555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2072011189569976e-05,
+      "loss": 2.6241,
+      "step": 47560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.206248412463662e-05,
+      "loss": 2.6835,
+      "step": 47565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.205295780708012e-05,
+      "loss": 2.6897,
+      "step": 47570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.20434322372974e-05,
+      "loss": 2.5868,
+      "step": 47575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2033907415685374e-05,
+      "loss": 2.679,
+      "step": 47580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.202438334264087e-05,
+      "loss": 2.56,
+      "step": 47585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2014860018560756e-05,
+      "loss": 2.6163,
+      "step": 47590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2005337443841766e-05,
+      "loss": 2.5441,
+      "step": 47595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.199581561888071e-05,
+      "loss": 2.7047,
+      "step": 47600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.19862945440743e-05,
+      "loss": 2.648,
+      "step": 47605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.197677421981924e-05,
+      "loss": 2.6493,
+      "step": 47610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.196725464651221e-05,
+      "loss": 2.6172,
+      "step": 47615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1957735824549836e-05,
+      "loss": 2.6778,
+      "step": 47620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.194821775432875e-05,
+      "loss": 2.6382,
+      "step": 47625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.193870043624548e-05,
+      "loss": 2.7251,
+      "step": 47630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.192918387069661e-05,
+      "loss": 2.5477,
+      "step": 47635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.191966805807863e-05,
+      "loss": 2.691,
+      "step": 47640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1910152998788034e-05,
+      "loss": 2.5736,
+      "step": 47645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.190063869322129e-05,
+      "loss": 2.6378,
+      "step": 47650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.189112514177478e-05,
+      "loss": 2.5795,
+      "step": 47655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1881612344844924e-05,
+      "loss": 2.5914,
+      "step": 47660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.187210030282806e-05,
+      "loss": 2.7001,
+      "step": 47665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1862589016120506e-05,
+      "loss": 2.61,
+      "step": 47670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.185307848511858e-05,
+      "loss": 2.6152,
+      "step": 47675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.184356871021852e-05,
+      "loss": 2.563,
+      "step": 47680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1834059691816574e-05,
+      "loss": 2.749,
+      "step": 47685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1824551430308925e-05,
+      "loss": 2.5581,
+      "step": 47690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.181504392609176e-05,
+      "loss": 2.4965,
+      "step": 47695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1805537179561187e-05,
+      "loss": 2.6815,
+      "step": 47700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.179603119111332e-05,
+      "loss": 2.705,
+      "step": 47705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1786525961144263e-05,
+      "loss": 2.5713,
+      "step": 47710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.177702149005001e-05,
+      "loss": 2.6669,
+      "step": 47715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.176751777822662e-05,
+      "loss": 2.8035,
+      "step": 47720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.175801482607001e-05,
+      "loss": 2.6718,
+      "step": 47725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.174851263397616e-05,
+      "loss": 2.6428,
+      "step": 47730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1739011202341e-05,
+      "loss": 2.6876,
+      "step": 47735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.172951053156037e-05,
+      "loss": 2.5724,
+      "step": 47740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.172001062203017e-05,
+      "loss": 2.6094,
+      "step": 47745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.171051147414618e-05,
+      "loss": 2.676,
+      "step": 47750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.170101308830421e-05,
+      "loss": 2.6534,
+      "step": 47755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1691515464900004e-05,
+      "loss": 2.6442,
+      "step": 47760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1682018604329285e-05,
+      "loss": 2.6692,
+      "step": 47765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1672522506987754e-05,
+      "loss": 2.5583,
+      "step": 47770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.166302717327106e-05,
+      "loss": 2.5099,
+      "step": 47775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.165353260357485e-05,
+      "loss": 2.6452,
+      "step": 47780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1644038798294707e-05,
+      "loss": 2.7415,
+      "step": 47785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.163454575782618e-05,
+      "loss": 2.7431,
+      "step": 47790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.162505348256485e-05,
+      "loss": 2.6718,
+      "step": 47795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.161556197290616e-05,
+      "loss": 2.6504,
+      "step": 47800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1606071229245635e-05,
+      "loss": 2.7029,
+      "step": 47805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1596581251978664e-05,
+      "loss": 2.6443,
+      "step": 47810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.158709204150069e-05,
+      "loss": 2.6677,
+      "step": 47815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.157760359820705e-05,
+      "loss": 2.6164,
+      "step": 47820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.156811592249311e-05,
+      "loss": 2.627,
+      "step": 47825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1558629014754175e-05,
+      "loss": 2.6135,
+      "step": 47830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.154914287538552e-05,
+      "loss": 2.6094,
+      "step": 47835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.153965750478241e-05,
+      "loss": 2.7634,
+      "step": 47840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.153017290334001e-05,
+      "loss": 2.4782,
+      "step": 47845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.152068907145356e-05,
+      "loss": 2.6218,
+      "step": 47850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.151120600951816e-05,
+      "loss": 2.6044,
+      "step": 47855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.150172371792896e-05,
+      "loss": 2.7574,
+      "step": 47860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.149224219708104e-05,
+      "loss": 2.6897,
+      "step": 47865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1482761447369426e-05,
+      "loss": 2.7208,
+      "step": 47870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.147328146918918e-05,
+      "loss": 2.6618,
+      "step": 47875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.146380226293525e-05,
+      "loss": 2.5952,
+      "step": 47880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.145432382900262e-05,
+      "loss": 2.8458,
+      "step": 47885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1444846167786215e-05,
+      "loss": 2.6604,
+      "step": 47890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1435369279680897e-05,
+      "loss": 2.7029,
+      "step": 47895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.142589316508158e-05,
+      "loss": 2.5491,
+      "step": 47900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1416417824383035e-05,
+      "loss": 2.6966,
+      "step": 47905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.14069432579801e-05,
+      "loss": 2.6646,
+      "step": 47910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.139746946626751e-05,
+      "loss": 2.5819,
+      "step": 47915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.138799644964e-05,
+      "loss": 2.6668,
+      "step": 47920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.137852420849229e-05,
+      "loss": 2.6243,
+      "step": 47925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.136905274321902e-05,
+      "loss": 2.7226,
+      "step": 47930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1359582054214846e-05,
+      "loss": 2.6594,
+      "step": 47935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.135011214187434e-05,
+      "loss": 2.6654,
+      "step": 47940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1340643006592105e-05,
+      "loss": 2.6953,
+      "step": 47945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1331174648762665e-05,
+      "loss": 2.6632,
+      "step": 47950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.132170706878051e-05,
+      "loss": 2.5581,
+      "step": 47955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.131224026704015e-05,
+      "loss": 2.6886,
+      "step": 47960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.130277424393598e-05,
+      "loss": 2.6084,
+      "step": 47965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.129330899986244e-05,
+      "loss": 2.641,
+      "step": 47970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.128384453521388e-05,
+      "loss": 2.4846,
+      "step": 47975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.127438085038465e-05,
+      "loss": 2.8053,
+      "step": 47980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.126491794576908e-05,
+      "loss": 2.5068,
+      "step": 47985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.125545582176142e-05,
+      "loss": 2.682,
+      "step": 47990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.124599447875593e-05,
+      "loss": 2.7886,
+      "step": 47995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.123653391714681e-05,
+      "loss": 2.667,
+      "step": 48000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 77054,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 6.925319084703744e+18,
+  "trial_name": null,
+  "trial_params": null
+}