diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,24019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.069933530379425,
+  "eval_steps": 500,
+  "global_step": 20000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999901916556e-05,
+      "loss": 1.7897,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996076662294e-05,
+      "loss": 1.8861,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999117249044e-05,
+      "loss": 1.4411,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998430665038e-05,
+      "loss": 1.4184,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999975479142666e-05,
+      "loss": 1.2592,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996468996798e-05,
+      "loss": 1.223,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995193912717e-05,
+      "loss": 1.2161,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993722662123e-05,
+      "loss": 1.2416,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999920552451324e-05,
+      "loss": 1.0517,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999901916618755e-05,
+      "loss": 1.2388,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988131912499e-05,
+      "loss": 1.2804,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985875997164e-05,
+      "loss": 1.0287,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983423916048e-05,
+      "loss": 1.1188,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980775669344e-05,
+      "loss": 1.1104,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999779312572584e-05,
+      "loss": 1.157,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974890680015e-05,
+      "loss": 1.1552,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971653937852e-05,
+      "loss": 1.1373,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999682210310237e-05,
+      "loss": 1.2307,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999645919598e-05,
+      "loss": 0.9955,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999960766724465e-05,
+      "loss": 0.9544,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999956745325318e-05,
+      "loss": 1.0432,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999952527762677e-05,
+      "loss": 1.1078,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999948114036871e-05,
+      "loss": 1.0228,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999435041482466e-05,
+      "loss": 1.0644,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999938698097166e-05,
+      "loss": 1.0653,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999933695884007e-05,
+      "loss": 1.0642,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99992849750916e-05,
+      "loss": 0.9661,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999923102973034e-05,
+      "loss": 1.0496,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999917512276053e-05,
+      "loss": 0.9529,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999911725418655e-05,
+      "loss": 0.999,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999905742401294e-05,
+      "loss": 1.023,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999899563224439e-05,
+      "loss": 1.0026,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999893187888577e-05,
+      "loss": 1.0805,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9998866163942055e-05,
+      "loss": 1.1233,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999879848741842e-05,
+      "loss": 1.1635,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999872884932016e-05,
+      "loss": 0.9927,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999865724965276e-05,
+      "loss": 1.096,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999858368842182e-05,
+      "loss": 1.0914,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999850816563312e-05,
+      "loss": 0.9758,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999843068129258e-05,
+      "loss": 0.9528,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9998351235406284e-05,
+      "loss": 1.033,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999826982798047e-05,
+      "loss": 1.2451,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999818645902152e-05,
+      "loss": 0.9208,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9998101128535984e-05,
+      "loss": 0.9928,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999801383653055e-05,
+      "loss": 0.8492,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999792458301207e-05,
+      "loss": 0.8356,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999783336798754e-05,
+      "loss": 0.8712,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999774019146413e-05,
+      "loss": 1.1794,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999764505344914e-05,
+      "loss": 0.8604,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999754795395004e-05,
+      "loss": 1.0337,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999744889297445e-05,
+      "loss": 0.9918,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999734787053014e-05,
+      "loss": 0.9943,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999724488662505e-05,
+      "loss": 1.0025,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999713994126724e-05,
+      "loss": 1.0377,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999703303446496e-05,
+      "loss": 0.7867,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999692416622659e-05,
+      "loss": 1.0016,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999681333656068e-05,
+      "loss": 0.8957,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999670054547592e-05,
+      "loss": 0.9476,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999658579298116e-05,
+      "loss": 0.8151,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99964690790854e-05,
+      "loss": 1.0061,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999635040379782e-05,
+      "loss": 0.9179,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999622976712771e-05,
+      "loss": 1.0512,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999610716908455e-05,
+      "loss": 0.9141,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999598260967795e-05,
+      "loss": 0.8681,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999585608891768e-05,
+      "loss": 0.8928,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999572760681368e-05,
+      "loss": 0.9659,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9995597163376025e-05,
+      "loss": 1.1282,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999546475861495e-05,
+      "loss": 1.0277,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9995330392540846e-05,
+      "loss": 0.8917,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999519406516426e-05,
+      "loss": 0.9693,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999505577649588e-05,
+      "loss": 0.9954,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9994915526546565e-05,
+      "loss": 0.7738,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999477331532732e-05,
+      "loss": 0.9392,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999462914284929e-05,
+      "loss": 0.9293,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9994483009123816e-05,
+      "loss": 1.0134,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999433491416233e-05,
+      "loss": 0.9113,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9994184857976484e-05,
+      "loss": 0.97,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999403284057803e-05,
+      "loss": 0.8572,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9993878861978914e-05,
+      "loss": 0.7934,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99937229221912e-05,
+      "loss": 1.1504,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999356502122714e-05,
+      "loss": 0.971,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9993405159099115e-05,
+      "loss": 0.7483,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999324333581967e-05,
+      "loss": 0.8933,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99930795514015e-05,
+      "loss": 0.8414,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9992913805857465e-05,
+      "loss": 0.991,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999274609920056e-05,
+      "loss": 1.0806,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999257643144396e-05,
+      "loss": 1.0021,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9992404802600963e-05,
+      "loss": 0.9346,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999223121268504e-05,
+      "loss": 0.8399,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999205566170981e-05,
+      "loss": 1.0498,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999187814968906e-05,
+      "loss": 0.8322,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999169867663671e-05,
+      "loss": 0.768,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999151724256684e-05,
+      "loss": 0.9191,
+      "step": 465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999133384749369e-05,
+      "loss": 0.8601,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999114849143165e-05,
+      "loss": 0.9834,
+      "step": 475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999096117439527e-05,
+      "loss": 1.0768,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999077189639924e-05,
+      "loss": 0.8655,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999058065745841e-05,
+      "loss": 1.01,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999038745758779e-05,
+      "loss": 1.0327,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999019229680254e-05,
+      "loss": 0.8839,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998999517511798e-05,
+      "loss": 1.0411,
+      "step": 505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998979609254957e-05,
+      "loss": 0.8918,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998959504911293e-05,
+      "loss": 0.8533,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9989392044823836e-05,
+      "loss": 0.9949,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998918707969822e-05,
+      "loss": 0.9551,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988980153752164e-05,
+      "loss": 1.0028,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998877126700191e-05,
+      "loss": 0.9293,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988560419463836e-05,
+      "loss": 0.9163,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988347611154504e-05,
+      "loss": 0.9049,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988132842090596e-05,
+      "loss": 0.9456,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998791611228897e-05,
+      "loss": 0.9466,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998769742176663e-05,
+      "loss": 0.9389,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998747677054074e-05,
+      "loss": 0.8853,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998725415862861e-05,
+      "loss": 0.797,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998702958604772e-05,
+      "loss": 1.0129,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998680305281568e-05,
+      "loss": 0.87,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998657455895026e-05,
+      "loss": 0.9552,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99863441044694e-05,
+      "loss": 0.7886,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9986111689391174e-05,
+      "loss": 0.9748,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998587731373383e-05,
+      "loss": 0.9811,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9985640977515744e-05,
+      "loss": 0.8928,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998540268075548e-05,
+      "loss": 0.9,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998516242347172e-05,
+      "loss": 0.8575,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998492020568332e-05,
+      "loss": 0.8665,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998467602740929e-05,
+      "loss": 0.9402,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998442988866879e-05,
+      "loss": 0.9656,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9984181789481134e-05,
+      "loss": 0.9597,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998393172986577e-05,
+      "loss": 0.8586,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998367970984236e-05,
+      "loss": 0.8484,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998342572943064e-05,
+      "loss": 0.955,
+      "step": 650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998316978865055e-05,
+      "loss": 0.9496,
+      "step": 655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998291188752219e-05,
+      "loss": 0.8029,
+      "step": 660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998265202606578e-05,
+      "loss": 1.0053,
+      "step": 665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9982390204301715e-05,
+      "loss": 1.0095,
+      "step": 670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9982126422250534e-05,
+      "loss": 0.9966,
+      "step": 675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9981860679932944e-05,
+      "loss": 0.8871,
+      "step": 680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99815929773698e-05,
+      "loss": 0.7858,
+      "step": 685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9981323314582085e-05,
+      "loss": 0.9204,
+      "step": 690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9981051691590984e-05,
+      "loss": 1.0274,
+      "step": 695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99807781084178e-05,
+      "loss": 1.0154,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9980502565083996e-05,
+      "loss": 0.7178,
+      "step": 705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99802250616112e-05,
+      "loss": 0.9133,
+      "step": 710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997994559802118e-05,
+      "loss": 1.034,
+      "step": 715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9979664174335874e-05,
+      "loss": 1.0744,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9979380790577364e-05,
+      "loss": 0.9425,
+      "step": 725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997909544676788e-05,
+      "loss": 0.9228,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997880814292981e-05,
+      "loss": 0.7898,
+      "step": 735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99785188790857e-05,
+      "loss": 0.9179,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9978227655258246e-05,
+      "loss": 0.9558,
+      "step": 745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997793447147031e-05,
+      "loss": 0.8552,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997763932774489e-05,
+      "loss": 0.8245,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997734222410514e-05,
+      "loss": 0.9163,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997704316057438e-05,
+      "loss": 1.0522,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997674213717607e-05,
+      "loss": 0.7773,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9976439153933843e-05,
+      "loss": 0.8591,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997613421087146e-05,
+      "loss": 0.9712,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997582730801285e-05,
+      "loss": 0.7712,
+      "step": 785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99755184453821e-05,
+      "loss": 0.797,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997520762300344e-05,
+      "loss": 1.0037,
+      "step": 795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997489484090127e-05,
+      "loss": 1.0809,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9974580099100124e-05,
+      "loss": 0.9685,
+      "step": 805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9974263397624695e-05,
+      "loss": 0.8071,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9973944736499837e-05,
+      "loss": 0.9863,
+      "step": 815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997362411575056e-05,
+      "loss": 0.8905,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9973301535402025e-05,
+      "loss": 0.8756,
+      "step": 825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997297699547953e-05,
+      "loss": 0.9958,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997265049600854e-05,
+      "loss": 1.1043,
+      "step": 835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99723220370147e-05,
+      "loss": 0.8283,
+      "step": 840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997199161852375e-05,
+      "loss": 0.958,
+      "step": 845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997165924056164e-05,
+      "loss": 1.1133,
+      "step": 850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997132490315444e-05,
+      "loss": 0.9223,
+      "step": 855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9970988606328397e-05,
+      "loss": 0.8501,
+      "step": 860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997065035010987e-05,
+      "loss": 0.8628,
+      "step": 865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997031013452543e-05,
+      "loss": 0.8997,
+      "step": 870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996996795960176e-05,
+      "loss": 0.7625,
+      "step": 875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996962382536572e-05,
+      "loss": 0.8228,
+      "step": 880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99692777318443e-05,
+      "loss": 0.7735,
+      "step": 885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9968929679064655e-05,
+      "loss": 0.8135,
+      "step": 890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9968579667054117e-05,
+      "loss": 0.8264,
+      "step": 895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996822769584013e-05,
+      "loss": 0.8968,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996787376545031e-05,
+      "loss": 0.8806,
+      "step": 905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9967517875912446e-05,
+      "loss": 0.799,
+      "step": 910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9967160027254446e-05,
+      "loss": 1.0222,
+      "step": 915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9966800219504405e-05,
+      "loss": 0.906,
+      "step": 920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9966438452690545e-05,
+      "loss": 0.8508,
+      "step": 925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9966074726841254e-05,
+      "loss": 0.9311,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996570904198508e-05,
+      "loss": 0.8446,
+      "step": 935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99653413981507e-05,
+      "loss": 0.8019,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996497179536699e-05,
+      "loss": 0.8191,
+      "step": 945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9964600233662915e-05,
+      "loss": 1.0145,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996422671306766e-05,
+      "loss": 0.8515,
+      "step": 955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996385123361053e-05,
+      "loss": 0.8751,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996347379532097e-05,
+      "loss": 0.8246,
+      "step": 965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996309439822862e-05,
+      "loss": 0.8811,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996271304236323e-05,
+      "loss": 0.8678,
+      "step": 975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996232972775474e-05,
+      "loss": 0.7357,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9961944454433204e-05,
+      "loss": 0.8088,
+      "step": 985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996155722242888e-05,
+      "loss": 0.8257,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996116803177214e-05,
+      "loss": 0.8482,
+      "step": 995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996077688249352e-05,
+      "loss": 0.928,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996038377462372e-05,
+      "loss": 0.8771,
+      "step": 1005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9959988708193585e-05,
+      "loss": 1.0969,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9959591683234106e-05,
+      "loss": 0.811,
+      "step": 1015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9959192699776445e-05,
+      "loss": 0.8073,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99587917578519e-05,
+      "loss": 0.8761,
+      "step": 1025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995838885749194e-05,
+      "loss": 0.8305,
+      "step": 1030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995798399872818e-05,
+      "loss": 0.9013,
+      "step": 1035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995757718159238e-05,
+      "loss": 0.8605,
+      "step": 1040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995716840611647e-05,
+      "loss": 0.7716,
+      "step": 1045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9956757672332515e-05,
+      "loss": 0.9188,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995634498027275e-05,
+      "loss": 0.9129,
+      "step": 1055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995593032996957e-05,
+      "loss": 0.8646,
+      "step": 1060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995551372145549e-05,
+      "loss": 0.847,
+      "step": 1065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995509515476321e-05,
+      "loss": 0.9522,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9954674629925576e-05,
+      "loss": 0.9094,
+      "step": 1075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995425214697558e-05,
+      "loss": 0.8598,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995382770594637e-05,
+      "loss": 0.7556,
+      "step": 1085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995340130687126e-05,
+      "loss": 0.891,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9952972949783706e-05,
+      "loss": 0.7336,
+      "step": 1095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995254263471732e-05,
+      "loss": 0.8502,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995211036170585e-05,
+      "loss": 0.8469,
+      "step": 1105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995167613078324e-05,
+      "loss": 0.8072,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995123994198355e-05,
+      "loss": 0.7748,
+      "step": 1115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9950801795341006e-05,
+      "loss": 0.9986,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995036169089e-05,
+      "loss": 0.9872,
+      "step": 1125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994991962866505e-05,
+      "loss": 0.9266,
+      "step": 1130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9949475608700846e-05,
+      "loss": 0.9211,
+      "step": 1135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994902963103224e-05,
+      "loss": 0.8535,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994858169569422e-05,
+      "loss": 0.692,
+      "step": 1145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994813180272192e-05,
+      "loss": 0.7803,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994767995215067e-05,
+      "loss": 0.7417,
+      "step": 1155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99472261440159e-05,
+      "loss": 0.7841,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9946770378353225e-05,
+      "loss": 0.8977,
+      "step": 1165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994631265519842e-05,
+      "loss": 0.7649,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994585297458739e-05,
+      "loss": 0.9532,
+      "step": 1175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99453913365562e-05,
+      "loss": 0.9149,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994492774114109e-05,
+      "loss": 0.7847,
+      "step": 1185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994446218837842e-05,
+      "loss": 0.8236,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994399467830473e-05,
+      "loss": 0.8189,
+      "step": 1195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99435252109567e-05,
+      "loss": 0.9335,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994305378637116e-05,
+      "loss": 0.9427,
+      "step": 1205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994258040458513e-05,
+      "loss": 0.8016,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9942105065635713e-05,
+      "loss": 1.0169,
+      "step": 1215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.994162776956024e-05,
+      "loss": 0.8399,
+      "step": 1220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.994114851639615e-05,
+      "loss": 0.9022,
+      "step": 1225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9940667306181056e-05,
+      "loss": 0.787,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.994018413895271e-05,
+      "loss": 0.9417,
+      "step": 1235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9939699014749015e-05,
+      "loss": 0.834,
+      "step": 1240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.993921193360806e-05,
+      "loss": 0.9071,
+      "step": 1245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.993872289556804e-05,
+      "loss": 0.9327,
+      "step": 1250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9938231900667345e-05,
+      "loss": 0.9787,
+      "step": 1255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.99377389489445e-05,
+      "loss": 0.8714,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.993724404043819e-05,
+      "loss": 0.995,
+      "step": 1265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.993674717518723e-05,
+      "loss": 0.7538,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9936248353230624e-05,
+      "loss": 0.9466,
+      "step": 1275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.993574757460751e-05,
+      "loss": 1.0344,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.993524483935718e-05,
+      "loss": 0.9573,
+      "step": 1285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.993474014751908e-05,
+      "loss": 0.9206,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9934233499132817e-05,
+      "loss": 0.9753,
+      "step": 1295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9933724894238146e-05,
+      "loss": 0.7576,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.993321433287496e-05,
+      "loss": 0.9328,
+      "step": 1305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9932701815083346e-05,
+      "loss": 0.8051,
+      "step": 1310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.99321873409035e-05,
+      "loss": 0.8396,
+      "step": 1315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9931670910375805e-05,
+      "loss": 0.8743,
+      "step": 1320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.993115252354077e-05,
+      "loss": 0.8973,
+      "step": 1325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9930632180439085e-05,
+      "loss": 0.8204,
+      "step": 1330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.993010988111157e-05,
+      "loss": 0.8987,
+      "step": 1335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992958562559921e-05,
+      "loss": 0.9644,
+      "step": 1340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992905941394315e-05,
+      "loss": 0.8282,
+      "step": 1345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9928531246184664e-05,
+      "loss": 0.7461,
+      "step": 1350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992800112236521e-05,
+      "loss": 0.9401,
+      "step": 1355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9927469042526374e-05,
+      "loss": 0.7926,
+      "step": 1360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992693500670992e-05,
+      "loss": 0.9393,
+      "step": 1365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9926399014957735e-05,
+      "loss": 0.7676,
+      "step": 1370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992586106731189e-05,
+      "loss": 0.882,
+      "step": 1375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992532116381459e-05,
+      "loss": 0.8941,
+      "step": 1380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9924779304508205e-05,
+      "loss": 0.9527,
+      "step": 1385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992423548943524e-05,
+      "loss": 0.9959,
+      "step": 1390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9923689718638386e-05,
+      "loss": 0.9233,
+      "step": 1395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992314199216046e-05,
+      "loss": 0.8311,
+      "step": 1400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.992259231004443e-05,
+      "loss": 0.849,
+      "step": 1405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9922040672333435e-05,
+      "loss": 0.8877,
+      "step": 1410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.992148707907077e-05,
+      "loss": 0.8116,
+      "step": 1415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9920931530299854e-05,
+      "loss": 0.8211,
+      "step": 1420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9920374026064294e-05,
+      "loss": 0.7091,
+      "step": 1425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991981456640783e-05,
+      "loss": 0.8518,
+      "step": 1430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9919253151374355e-05,
+      "loss": 0.7311,
+      "step": 1435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9918689781007946e-05,
+      "loss": 0.9369,
+      "step": 1440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9918124455352774e-05,
+      "loss": 0.8322,
+      "step": 1445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991755717445322e-05,
+      "loss": 0.8642,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.99169879383538e-05,
+      "loss": 0.8866,
+      "step": 1455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9916416747099166e-05,
+      "loss": 0.9548,
+      "step": 1460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991584360073415e-05,
+      "loss": 0.967,
+      "step": 1465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991526849930371e-05,
+      "loss": 0.8791,
+      "step": 1470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9914691442852976e-05,
+      "loss": 1.0645,
+      "step": 1475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991411243142725e-05,
+      "loss": 0.9111,
+      "step": 1480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991353146507194e-05,
+      "loss": 0.8932,
+      "step": 1485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991294854383264e-05,
+      "loss": 0.8218,
+      "step": 1490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991236366775509e-05,
+      "loss": 0.887,
+      "step": 1495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991177683688518e-05,
+      "loss": 0.8274,
+      "step": 1500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991118805126897e-05,
+      "loss": 0.9471,
+      "step": 1505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991059731095265e-05,
+      "loss": 0.7971,
+      "step": 1510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9910004615982565e-05,
+      "loss": 0.9093,
+      "step": 1515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990940996640524e-05,
+      "loss": 0.8429,
+      "step": 1520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990881336226732e-05,
+      "loss": 0.8453,
+      "step": 1525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990821480361563e-05,
+      "loss": 0.8797,
+      "step": 1530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990761429049713e-05,
+      "loss": 0.9274,
+      "step": 1535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990701182295894e-05,
+      "loss": 0.8274,
+      "step": 1540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990640740104834e-05,
+      "loss": 0.8599,
+      "step": 1545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990580102481275e-05,
+      "loss": 0.7916,
+      "step": 1550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990519269429975e-05,
+      "loss": 0.777,
+      "step": 1555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990458240955708e-05,
+      "loss": 0.8698,
+      "step": 1560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990397017063262e-05,
+      "loss": 0.7045,
+      "step": 1565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990335597757442e-05,
+      "loss": 0.897,
+      "step": 1570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990273983043067e-05,
+      "loss": 0.9374,
+      "step": 1575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9902121729249704e-05,
+      "loss": 0.9631,
+      "step": 1580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990150167408004e-05,
+      "loss": 0.9525,
+      "step": 1585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9900879664970335e-05,
+      "loss": 0.7914,
+      "step": 1590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.990025570196937e-05,
+      "loss": 0.8088,
+      "step": 1595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989962978512613e-05,
+      "loss": 0.9072,
+      "step": 1600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989900191448972e-05,
+      "loss": 0.8763,
+      "step": 1605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989837209010941e-05,
+      "loss": 0.8714,
+      "step": 1610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9897740312034616e-05,
+      "loss": 0.8351,
+      "step": 1615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9897106580314913e-05,
+      "loss": 0.9255,
+      "step": 1620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989647089500003e-05,
+      "loss": 0.8401,
+      "step": 1625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989583325613984e-05,
+      "loss": 0.7644,
+      "step": 1630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989519366378438e-05,
+      "loss": 0.8625,
+      "step": 1635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989455211798384e-05,
+      "loss": 1.0019,
+      "step": 1640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989390861878856e-05,
+      "loss": 0.8116,
+      "step": 1645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989326316624903e-05,
+      "loss": 0.9912,
+      "step": 1650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9892615760415905e-05,
+      "loss": 0.7684,
+      "step": 1655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9891966401339975e-05,
+      "loss": 0.8937,
+      "step": 1660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989131508907219e-05,
+      "loss": 0.946,
+      "step": 1665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989066182366367e-05,
+      "loss": 0.8702,
+      "step": 1670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9890006605165665e-05,
+      "loss": 0.8557,
+      "step": 1675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9889349433629584e-05,
+      "loss": 0.9027,
+      "step": 1680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988869030910701e-05,
+      "loss": 0.7927,
+      "step": 1685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988802923164965e-05,
+      "loss": 0.839,
+      "step": 1690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9887366201309374e-05,
+      "loss": 0.9143,
+      "step": 1695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9886701218138205e-05,
+      "loss": 0.934,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988603428218834e-05,
+      "loss": 0.8441,
+      "step": 1705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988536539351209e-05,
+      "loss": 0.926,
+      "step": 1710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988469455216196e-05,
+      "loss": 0.7813,
+      "step": 1715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988402175819058e-05,
+      "loss": 0.8033,
+      "step": 1720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988334701165074e-05,
+      "loss": 0.7388,
+      "step": 1725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988267031259538e-05,
+      "loss": 0.9363,
+      "step": 1730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988199166107761e-05,
+      "loss": 0.9499,
+      "step": 1735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988131105715068e-05,
+      "loss": 0.9348,
+      "step": 1740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9880628500867985e-05,
+      "loss": 0.8518,
+      "step": 1745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.987994399228309e-05,
+      "loss": 1.003,
+      "step": 1750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.987925753144971e-05,
+      "loss": 0.9232,
+      "step": 1755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.987856911842171e-05,
+      "loss": 0.9477,
+      "step": 1760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.987787875325309e-05,
+      "loss": 0.7504,
+      "step": 1765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9877186435998045e-05,
+      "loss": 0.9137,
+      "step": 1770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9876492166710884e-05,
+      "loss": 0.8655,
+      "step": 1775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987579594544608e-05,
+      "loss": 0.6944,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987509777225827e-05,
+      "loss": 0.8168,
+      "step": 1785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987439764720225e-05,
+      "loss": 0.9912,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9873695570332934e-05,
+      "loss": 0.9112,
+      "step": 1795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9872991541705424e-05,
+      "loss": 0.8157,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987228556137497e-05,
+      "loss": 0.869,
+      "step": 1805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987157762939695e-05,
+      "loss": 0.8406,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9870867745826923e-05,
+      "loss": 0.8385,
+      "step": 1815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987015591072058e-05,
+      "loss": 0.9317,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9869442124133805e-05,
+      "loss": 0.8695,
+      "step": 1825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9868726386122575e-05,
+      "loss": 0.9023,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986800869674307e-05,
+      "loss": 0.6973,
+      "step": 1835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9867289056051605e-05,
+      "loss": 0.9736,
+      "step": 1840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986656746410464e-05,
+      "loss": 0.806,
+      "step": 1845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986584392095879e-05,
+      "loss": 0.6967,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9865118426670845e-05,
+      "loss": 0.8629,
+      "step": 1855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986439098129772e-05,
+      "loss": 0.8126,
+      "step": 1860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9863661584896506e-05,
+      "loss": 0.8886,
+      "step": 1865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9862930237524425e-05,
+      "loss": 1.0021,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986219693923887e-05,
+      "loss": 0.7813,
+      "step": 1875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9861461690097377e-05,
+      "loss": 0.719,
+      "step": 1880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9860724490157645e-05,
+      "loss": 0.7863,
+      "step": 1885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9859985339477516e-05,
+      "loss": 0.8206,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.985924423811499e-05,
+      "loss": 0.8631,
+      "step": 1895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.985850118612821e-05,
+      "loss": 1.0037,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9857756183575494e-05,
+      "loss": 0.9073,
+      "step": 1905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.985700923051529e-05,
+      "loss": 0.8824,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9856260327006213e-05,
+      "loss": 0.8418,
+      "step": 1915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.985550947310703e-05,
+      "loss": 0.8256,
+      "step": 1920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.985475666887666e-05,
+      "loss": 0.8181,
+      "step": 1925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.985400191437416e-05,
+      "loss": 0.8676,
+      "step": 1930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.985324520965876e-05,
+      "loss": 0.9315,
+      "step": 1935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.985248655478984e-05,
+      "loss": 0.9193,
+      "step": 1940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.985172594982693e-05,
+      "loss": 0.6758,
+      "step": 1945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.985096339482971e-05,
+      "loss": 1.0395,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9850198889858005e-05,
+      "loss": 0.9421,
+      "step": 1955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.984943243497182e-05,
+      "loss": 0.8176,
+      "step": 1960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.984866403023129e-05,
+      "loss": 0.8059,
+      "step": 1965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9847893675696697e-05,
+      "loss": 0.7081,
+      "step": 1970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9847121371428504e-05,
+      "loss": 1.0082,
+      "step": 1975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.984634711748731e-05,
+      "loss": 0.8226,
+      "step": 1980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.984557091393387e-05,
+      "loss": 0.8726,
+      "step": 1985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9844792760829075e-05,
+      "loss": 0.8124,
+      "step": 1990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9844012658234e-05,
+      "loss": 0.9257,
+      "step": 1995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9843230606209845e-05,
+      "loss": 0.8869,
+      "step": 2000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.984244660481798e-05,
+      "loss": 0.8605,
+      "step": 2005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9841660654119924e-05,
+      "loss": 0.8098,
+      "step": 2010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9840872754177356e-05,
+      "loss": 0.9607,
+      "step": 2015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.984008290505209e-05,
+      "loss": 0.9345,
+      "step": 2020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9839291106806095e-05,
+      "loss": 0.7879,
+      "step": 2025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9838497359501524e-05,
+      "loss": 0.7783,
+      "step": 2030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9837701663200636e-05,
+      "loss": 0.7527,
+      "step": 2035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.983690401796588e-05,
+      "loss": 0.8556,
+      "step": 2040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.983610442385984e-05,
+      "loss": 0.8758,
+      "step": 2045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9835302880945266e-05,
+      "loss": 0.9996,
+      "step": 2050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9834499389285036e-05,
+      "loss": 0.9905,
+      "step": 2055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9833693948942217e-05,
+      "loss": 0.8837,
+      "step": 2060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.983288655998e-05,
+      "loss": 0.7579,
+      "step": 2065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.983207722246173e-05,
+      "loss": 0.9043,
+      "step": 2070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.983126593645092e-05,
+      "loss": 0.9022,
+      "step": 2075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9830452702011236e-05,
+      "loss": 1.0319,
+      "step": 2080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9829637519206486e-05,
+      "loss": 0.8624,
+      "step": 2085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9828820388100625e-05,
+      "loss": 0.8045,
+      "step": 2090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982800130875778e-05,
+      "loss": 0.9147,
+      "step": 2095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982718028124223e-05,
+      "loss": 0.9279,
+      "step": 2100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982635730561837e-05,
+      "loss": 0.8517,
+      "step": 2105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.98255323819508e-05,
+      "loss": 0.6737,
+      "step": 2110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9824705510304247e-05,
+      "loss": 0.7965,
+      "step": 2115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982387669074359e-05,
+      "loss": 0.7378,
+      "step": 2120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9823045923333855e-05,
+      "loss": 0.8486,
+      "step": 2125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982221320814024e-05,
+      "loss": 0.8096,
+      "step": 2130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982137854522809e-05,
+      "loss": 0.7806,
+      "step": 2135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982054193466289e-05,
+      "loss": 0.8132,
+      "step": 2140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981970337651029e-05,
+      "loss": 0.7834,
+      "step": 2145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.981886287083607e-05,
+      "loss": 0.8362,
+      "step": 2150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.981802041770621e-05,
+      "loss": 0.9176,
+      "step": 2155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.981717601718681e-05,
+      "loss": 0.8825,
+      "step": 2160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.981632966934411e-05,
+      "loss": 0.9248,
+      "step": 2165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.981548137424453e-05,
+      "loss": 0.7836,
+      "step": 2170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9814631131954635e-05,
+      "loss": 0.8113,
+      "step": 2175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.981377894254114e-05,
+      "loss": 0.8362,
+      "step": 2180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.981292480607091e-05,
+      "loss": 0.8749,
+      "step": 2185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.981206872261097e-05,
+      "loss": 0.9427,
+      "step": 2190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.98112106922285e-05,
+      "loss": 0.8892,
+      "step": 2195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.981035071499081e-05,
+      "loss": 0.8796,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980948879096539e-05,
+      "loss": 0.9027,
+      "step": 2205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980862492021987e-05,
+      "loss": 0.8985,
+      "step": 2210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9807759102822036e-05,
+      "loss": 0.8132,
+      "step": 2215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980689133883983e-05,
+      "loss": 0.7887,
+      "step": 2220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980602162834135e-05,
+      "loss": 0.8031,
+      "step": 2225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9805149971394815e-05,
+      "loss": 0.6112,
+      "step": 2230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9804276368068634e-05,
+      "loss": 0.7819,
+      "step": 2235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980340081843137e-05,
+      "loss": 0.8431,
+      "step": 2240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.98025233225517e-05,
+      "loss": 0.8264,
+      "step": 2245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980164388049849e-05,
+      "loss": 0.8417,
+      "step": 2250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980076249234075e-05,
+      "loss": 0.8623,
+      "step": 2255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979987915814763e-05,
+      "loss": 0.7658,
+      "step": 2260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979899387798846e-05,
+      "loss": 0.8775,
+      "step": 2265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979810665193269e-05,
+      "loss": 0.8722,
+      "step": 2270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979721748004993e-05,
+      "loss": 0.7641,
+      "step": 2275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979632636240997e-05,
+      "loss": 0.7558,
+      "step": 2280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979543329908273e-05,
+      "loss": 0.8515,
+      "step": 2285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979453829013827e-05,
+      "loss": 0.8285,
+      "step": 2290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979364133564684e-05,
+      "loss": 0.9257,
+      "step": 2295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.97927424356788e-05,
+      "loss": 0.7933,
+      "step": 2300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.97918415903047e-05,
+      "loss": 0.7068,
+      "step": 2305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979093879959521e-05,
+      "loss": 0.7912,
+      "step": 2310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9790034063621196e-05,
+      "loss": 0.7526,
+      "step": 2315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.978912738245362e-05,
+      "loss": 0.8986,
+      "step": 2320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.978821875616364e-05,
+      "loss": 0.8767,
+      "step": 2325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9787308184822554e-05,
+      "loss": 0.8165,
+      "step": 2330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9786395668501815e-05,
+      "loss": 0.7368,
+      "step": 2335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978548120727302e-05,
+      "loss": 0.9021,
+      "step": 2340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978456480120792e-05,
+      "loss": 0.8389,
+      "step": 2345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978364645037843e-05,
+      "loss": 0.7349,
+      "step": 2350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.97827261548566e-05,
+      "loss": 0.7653,
+      "step": 2355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9781803914714654e-05,
+      "loss": 0.7587,
+      "step": 2360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978087973002495e-05,
+      "loss": 0.8199,
+      "step": 2365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9779953600860005e-05,
+      "loss": 0.7978,
+      "step": 2370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.97790255272925e-05,
+      "loss": 1.0208,
+      "step": 2375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.977809550939525e-05,
+      "loss": 0.7798,
+      "step": 2380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.977716354724122e-05,
+      "loss": 0.8695,
+      "step": 2385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.977622964090356e-05,
+      "loss": 0.8421,
+      "step": 2390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9775293790455536e-05,
+      "loss": 0.7212,
+      "step": 2395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.977435599597058e-05,
+      "loss": 0.8967,
+      "step": 2400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9773416257522286e-05,
+      "loss": 0.895,
+      "step": 2405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.977247457518439e-05,
+      "loss": 0.9811,
+      "step": 2410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9771530949030776e-05,
+      "loss": 0.8655,
+      "step": 2415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.97705853791355e-05,
+      "loss": 0.9284,
+      "step": 2420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.976963786557274e-05,
+      "loss": 0.675,
+      "step": 2425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.976868840841686e-05,
+      "loss": 0.8499,
+      "step": 2430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.976773700774236e-05,
+      "loss": 1.0544,
+      "step": 2435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9766783663623886e-05,
+      "loss": 0.9486,
+      "step": 2440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.976582837613624e-05,
+      "loss": 0.663,
+      "step": 2445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.97648711453544e-05,
+      "loss": 0.7747,
+      "step": 2450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9763911971353447e-05,
+      "loss": 0.7663,
+      "step": 2455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9762950854208666e-05,
+      "loss": 0.6607,
+      "step": 2460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.976198779399548e-05,
+      "loss": 0.813,
+      "step": 2465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.976102279078944e-05,
+      "loss": 0.7706,
+      "step": 2470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.976005584466626e-05,
+      "loss": 0.8038,
+      "step": 2475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.975908695570183e-05,
+      "loss": 0.8027,
+      "step": 2480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9758116123972173e-05,
+      "loss": 0.8031,
+      "step": 2485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.975714334955346e-05,
+      "loss": 0.9296,
+      "step": 2490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.975616863252203e-05,
+      "loss": 0.9807,
+      "step": 2495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9755191972954366e-05,
+      "loss": 0.9434,
+      "step": 2500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9754213370927086e-05,
+      "loss": 0.8829,
+      "step": 2505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.975323282651701e-05,
+      "loss": 1.0117,
+      "step": 2510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.975225033980104e-05,
+      "loss": 1.1015,
+      "step": 2515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9751265910856294e-05,
+      "loss": 0.9377,
+      "step": 2520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.975027953976e-05,
+      "loss": 0.7232,
+      "step": 2525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9749291226589584e-05,
+      "loss": 0.7432,
+      "step": 2530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974830097142257e-05,
+      "loss": 0.9603,
+      "step": 2535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974730877433666e-05,
+      "loss": 0.6104,
+      "step": 2540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974631463540973e-05,
+      "loss": 0.7624,
+      "step": 2545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974531855471976e-05,
+      "loss": 0.893,
+      "step": 2550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974432053234493e-05,
+      "loss": 0.7997,
+      "step": 2555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9743320568363536e-05,
+      "loss": 0.8702,
+      "step": 2560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974231866285406e-05,
+      "loss": 0.9106,
+      "step": 2565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9741314815895104e-05,
+      "loss": 0.7824,
+      "step": 2570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9740309027565434e-05,
+      "loss": 0.792,
+      "step": 2575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973930129794398e-05,
+      "loss": 0.8312,
+      "step": 2580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9738291627109814e-05,
+      "loss": 0.9209,
+      "step": 2585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9737280015142165e-05,
+      "loss": 0.7518,
+      "step": 2590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9736266462120394e-05,
+      "loss": 0.845,
+      "step": 2595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9735250968124054e-05,
+      "loss": 0.9205,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973423353323281e-05,
+      "loss": 0.7602,
+      "step": 2605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973321415752651e-05,
+      "loss": 0.7718,
+      "step": 2610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973219284108513e-05,
+      "loss": 0.9397,
+      "step": 2615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9731169583988815e-05,
+      "loss": 0.5993,
+      "step": 2620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973014438631786e-05,
+      "loss": 0.9764,
+      "step": 2625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97291172481527e-05,
+      "loss": 0.8947,
+      "step": 2630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.972808816957394e-05,
+      "loss": 0.7173,
+      "step": 2635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9727057150662315e-05,
+      "loss": 0.7808,
+      "step": 2640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9726024191498744e-05,
+      "loss": 0.8548,
+      "step": 2645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.972498929216427e-05,
+      "loss": 0.771,
+      "step": 2650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9723952452740095e-05,
+      "loss": 0.7493,
+      "step": 2655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.972291367330759e-05,
+      "loss": 0.8956,
+      "step": 2660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9721872953948244e-05,
+      "loss": 1.0156,
+      "step": 2665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.972083029474374e-05,
+      "loss": 0.9684,
+      "step": 2670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971978569577587e-05,
+      "loss": 0.7829,
+      "step": 2675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9718739157126627e-05,
+      "loss": 0.7946,
+      "step": 2680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97176906788781e-05,
+      "loss": 0.8013,
+      "step": 2685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971664026111259e-05,
+      "loss": 0.7803,
+      "step": 2690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97155879039125e-05,
+      "loss": 0.8442,
+      "step": 2695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9714533607360404e-05,
+      "loss": 0.9028,
+      "step": 2700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971347737153904e-05,
+      "loss": 0.945,
+      "step": 2705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971241919653128e-05,
+      "loss": 0.8276,
+      "step": 2710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9711359082420156e-05,
+      "loss": 0.841,
+      "step": 2715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.971029702928886e-05,
+      "loss": 0.9866,
+      "step": 2720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9709233037220713e-05,
+      "loss": 0.7829,
+      "step": 2725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.970816710629922e-05,
+      "loss": 0.8162,
+      "step": 2730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9707099236608e-05,
+      "loss": 0.8764,
+      "step": 2735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.970602942823087e-05,
+      "loss": 0.7875,
+      "step": 2740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.970495768125176e-05,
+      "loss": 0.863,
+      "step": 2745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9703883995754766e-05,
+      "loss": 0.9755,
+      "step": 2750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9702808371824136e-05,
+      "loss": 0.7829,
+      "step": 2755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9701730809544286e-05,
+      "loss": 0.7883,
+      "step": 2760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.970065130899974e-05,
+      "loss": 0.9066,
+      "step": 2765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9699569870275236e-05,
+      "loss": 0.8159,
+      "step": 2770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.969848649345561e-05,
+      "loss": 0.8148,
+      "step": 2775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.969740117862587e-05,
+      "loss": 0.7785,
+      "step": 2780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.969631392587119e-05,
+      "loss": 0.8492,
+      "step": 2785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9695224735276874e-05,
+      "loss": 0.8909,
+      "step": 2790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9694133606928386e-05,
+      "loss": 0.8065,
+      "step": 2795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9693040540911353e-05,
+      "loss": 1.0437,
+      "step": 2800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9691945537311536e-05,
+      "loss": 0.7663,
+      "step": 2805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.969084859621486e-05,
+      "loss": 0.761,
+      "step": 2810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968974971770739e-05,
+      "loss": 0.8197,
+      "step": 2815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9688648901875365e-05,
+      "loss": 0.8564,
+      "step": 2820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968754614880515e-05,
+      "loss": 0.7359,
+      "step": 2825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968644145858328e-05,
+      "loss": 0.8015,
+      "step": 2830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968533483129644e-05,
+      "loss": 0.7678,
+      "step": 2835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9684226267031466e-05,
+      "loss": 0.7663,
+      "step": 2840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968311576587533e-05,
+      "loss": 0.8433,
+      "step": 2845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968200332791518e-05,
+      "loss": 0.8641,
+      "step": 2850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.96808889532383e-05,
+      "loss": 0.7179,
+      "step": 2855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9679772641932134e-05,
+      "loss": 0.7958,
+      "step": 2860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967865439408428e-05,
+      "loss": 0.7732,
+      "step": 2865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9677534209782475e-05,
+      "loss": 0.7677,
+      "step": 2870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967641208911461e-05,
+      "loss": 0.954,
+      "step": 2875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967528803216876e-05,
+      "loss": 0.9279,
+      "step": 2880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.96741620390331e-05,
+      "loss": 0.8701,
+      "step": 2885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9673034109796e-05,
+      "loss": 0.813,
+      "step": 2890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967190424454595e-05,
+      "loss": 0.9161,
+      "step": 2895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.967077244337161e-05,
+      "loss": 0.8249,
+      "step": 2900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.966963870636181e-05,
+      "loss": 0.8396,
+      "step": 2905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9668503033605485e-05,
+      "loss": 0.792,
+      "step": 2910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.966736542519176e-05,
+      "loss": 0.706,
+      "step": 2915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.966622588120989e-05,
+      "loss": 0.7794,
+      "step": 2920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.966508440174931e-05,
+      "loss": 0.8807,
+      "step": 2925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.966394098689956e-05,
+      "loss": 0.8132,
+      "step": 2930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.966279563675038e-05,
+      "loss": 0.7731,
+      "step": 2935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.966164835139164e-05,
+      "loss": 0.8436,
+      "step": 2940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.966049913091336e-05,
+      "loss": 0.7004,
+      "step": 2945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9659347975405726e-05,
+      "loss": 0.8035,
+      "step": 2950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9658194884959046e-05,
+      "loss": 0.8535,
+      "step": 2955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.965703985966381e-05,
+      "loss": 0.7824,
+      "step": 2960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.965588289961066e-05,
+      "loss": 0.8664,
+      "step": 2965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.965472400489035e-05,
+      "loss": 0.7619,
+      "step": 2970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.965356317559384e-05,
+      "loss": 0.8276,
+      "step": 2975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.965240041181222e-05,
+      "loss": 0.8057,
+      "step": 2980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.96512357136367e-05,
+      "loss": 0.909,
+      "step": 2985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.96500690811587e-05,
+      "loss": 0.8557,
+      "step": 2990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9648900514469745e-05,
+      "loss": 0.765,
+      "step": 2995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.964773001366153e-05,
+      "loss": 1.0673,
+      "step": 3000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.96465575788259e-05,
+      "loss": 0.8582,
+      "step": 3005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9645383210054865e-05,
+      "loss": 0.9613,
+      "step": 3010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9644206907440563e-05,
+      "loss": 0.8421,
+      "step": 3015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9643028671075285e-05,
+      "loss": 0.9535,
+      "step": 3020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.96418485010515e-05,
+      "loss": 0.9872,
+      "step": 3025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.964066639746181e-05,
+      "loss": 0.9017,
+      "step": 3030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.963948236039897e-05,
+      "loss": 0.8539,
+      "step": 3035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.963829638995588e-05,
+      "loss": 0.8314,
+      "step": 3040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9637108486225604e-05,
+      "loss": 0.8791,
+      "step": 3045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.963591864930135e-05,
+      "loss": 0.7487,
+      "step": 3050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.963472687927649e-05,
+      "loss": 0.8403,
+      "step": 3055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9633533176244526e-05,
+      "loss": 0.7483,
+      "step": 3060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.963233754029913e-05,
+      "loss": 0.8811,
+      "step": 3065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9631139971534126e-05,
+      "loss": 0.8106,
+      "step": 3070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.962994047004348e-05,
+      "loss": 0.8174,
+      "step": 3075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.96287390359213e-05,
+      "loss": 0.7392,
+      "step": 3080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.962753566926187e-05,
+      "loss": 0.7308,
+      "step": 3085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.962633037015961e-05,
+      "loss": 0.7862,
+      "step": 3090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.962512313870911e-05,
+      "loss": 0.7249,
+      "step": 3095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.962391397500508e-05,
+      "loss": 0.7166,
+      "step": 3100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.96227028791424e-05,
+      "loss": 0.861,
+      "step": 3105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.962148985121612e-05,
+      "loss": 0.8061,
+      "step": 3110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.962027489132139e-05,
+      "loss": 0.7095,
+      "step": 3115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.961905799955357e-05,
+      "loss": 0.839,
+      "step": 3120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.961783917600815e-05,
+      "loss": 0.8933,
+      "step": 3125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9616618420780745e-05,
+      "loss": 0.8229,
+      "step": 3130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.961539573396715e-05,
+      "loss": 0.8342,
+      "step": 3135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.961417111566332e-05,
+      "loss": 0.7522,
+      "step": 3140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.961294456596533e-05,
+      "loss": 0.7861,
+      "step": 3145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.961171608496943e-05,
+      "loss": 0.7584,
+      "step": 3150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.961048567277201e-05,
+      "loss": 0.8426,
+      "step": 3155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.960925332946963e-05,
+      "loss": 0.8247,
+      "step": 3160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.960801905515897e-05,
+      "loss": 0.7422,
+      "step": 3165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.960678284993689e-05,
+      "loss": 0.7993,
+      "step": 3170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.96055447139004e-05,
+      "loss": 0.746,
+      "step": 3175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.960430464714663e-05,
+      "loss": 0.8515,
+      "step": 3180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9603062649772895e-05,
+      "loss": 0.7546,
+      "step": 3185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.960181872187666e-05,
+      "loss": 0.7612,
+      "step": 3190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9600572863555515e-05,
+      "loss": 0.8212,
+      "step": 3195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.959932507490723e-05,
+      "loss": 0.8165,
+      "step": 3200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9598075356029713e-05,
+      "loss": 0.8134,
+      "step": 3205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9596823707021026e-05,
+      "loss": 0.8584,
+      "step": 3210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9595570127979375e-05,
+      "loss": 0.8665,
+      "step": 3215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.959431461900313e-05,
+      "loss": 0.8252,
+      "step": 3220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.95930571801908e-05,
+      "loss": 0.7456,
+      "step": 3225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.959179781164107e-05,
+      "loss": 0.7836,
+      "step": 3230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.959053651345275e-05,
+      "loss": 0.9068,
+      "step": 3235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.958927328572479e-05,
+      "loss": 0.9326,
+      "step": 3240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.95882613143405e-05,
+      "loss": 0.8543,
+      "step": 3245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.958699461369112e-05,
+      "loss": 0.845,
+      "step": 3250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.958572598378003e-05,
+      "loss": 0.765,
+      "step": 3255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.958445542470679e-05,
+      "loss": 0.8575,
+      "step": 3260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.958318293657108e-05,
+      "loss": 0.8214,
+      "step": 3265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9581908519472766e-05,
+      "loss": 0.9318,
+      "step": 3270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.958063217351183e-05,
+      "loss": 0.8527,
+      "step": 3275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9579353898788435e-05,
+      "loss": 0.9015,
+      "step": 3280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.957807369540289e-05,
+      "loss": 0.9015,
+      "step": 3285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9576791563455625e-05,
+      "loss": 0.789,
+      "step": 3290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.957550750304727e-05,
+      "loss": 0.8802,
+      "step": 3295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.957422151427856e-05,
+      "loss": 0.8692,
+      "step": 3300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9572933597250424e-05,
+      "loss": 0.9522,
+      "step": 3305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9571643752063894e-05,
+      "loss": 0.8021,
+      "step": 3310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.95703519788202e-05,
+      "loss": 0.8531,
+      "step": 3315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.95690582776207e-05,
+      "loss": 1.0268,
+      "step": 3320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.95677626485669e-05,
+      "loss": 0.6887,
+      "step": 3325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.956646509176047e-05,
+      "loss": 0.6657,
+      "step": 3330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9565165607303225e-05,
+      "loss": 0.6664,
+      "step": 3335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.956386419529713e-05,
+      "loss": 0.8534,
+      "step": 3340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9562560855844295e-05,
+      "loss": 0.7976,
+      "step": 3345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9561255589047e-05,
+      "loss": 0.7612,
+      "step": 3350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.955994839500766e-05,
+      "loss": 0.8446,
+      "step": 3355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.955863927382885e-05,
+      "loss": 0.7642,
+      "step": 3360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.955732822561329e-05,
+      "loss": 0.7925,
+      "step": 3365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.955601525046385e-05,
+      "loss": 0.9055,
+      "step": 3370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.955470034848355e-05,
+      "loss": 0.6796,
+      "step": 3375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.955338351977559e-05,
+      "loss": 0.8158,
+      "step": 3380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9552064764443275e-05,
+      "loss": 0.8094,
+      "step": 3385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.955074408259008e-05,
+      "loss": 0.8934,
+      "step": 3390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9549421474319654e-05,
+      "loss": 0.8261,
+      "step": 3395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.954809693973578e-05,
+      "loss": 0.9384,
+      "step": 3400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.954677047894236e-05,
+      "loss": 0.9363,
+      "step": 3405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.95454420920435e-05,
+      "loss": 0.9239,
+      "step": 3410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9544111779143435e-05,
+      "loss": 0.9831,
+      "step": 3415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.954277954034654e-05,
+      "loss": 0.7436,
+      "step": 3420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.954144537575736e-05,
+      "loss": 0.8792,
+      "step": 3425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.954010928548057e-05,
+      "loss": 0.8038,
+      "step": 3430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9538771269621024e-05,
+      "loss": 0.9046,
+      "step": 3435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.953743132828371e-05,
+      "loss": 0.8244,
+      "step": 3440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.953608946157376e-05,
+      "loss": 0.9734,
+      "step": 3445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.953474566959647e-05,
+      "loss": 0.7814,
+      "step": 3450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9533399952457276e-05,
+      "loss": 0.8054,
+      "step": 3455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.953205231026179e-05,
+      "loss": 0.7926,
+      "step": 3460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.953070274311574e-05,
+      "loss": 0.8162,
+      "step": 3465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9529351251125035e-05,
+      "loss": 0.7372,
+      "step": 3470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9527997834395706e-05,
+      "loss": 0.8506,
+      "step": 3475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.952664249303397e-05,
+      "loss": 1.0776,
+      "step": 3480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.952528522714616e-05,
+      "loss": 0.8315,
+      "step": 3485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9523926036838784e-05,
+      "loss": 0.9055,
+      "step": 3490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.952256492221849e-05,
+      "loss": 0.7596,
+      "step": 3495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.952120188339209e-05,
+      "loss": 0.7789,
+      "step": 3500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9519836920466535e-05,
+      "loss": 0.8056,
+      "step": 3505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.951847003354892e-05,
+      "loss": 0.7515,
+      "step": 3510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.95171012227465e-05,
+      "loss": 0.8561,
+      "step": 3515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.951573048816668e-05,
+      "loss": 0.8129,
+      "step": 3520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.951435782991704e-05,
+      "loss": 0.716,
+      "step": 3525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.951298324810525e-05,
+      "loss": 0.8735,
+      "step": 3530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.95116067428392e-05,
+      "loss": 0.8052,
+      "step": 3535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.951022831422689e-05,
+      "loss": 0.8611,
+      "step": 3540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9508847962376484e-05,
+      "loss": 0.8507,
+      "step": 3545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9507465687396284e-05,
+      "loss": 0.8489,
+      "step": 3550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9506081489394764e-05,
+      "loss": 0.8963,
+      "step": 3555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.950469536848053e-05,
+      "loss": 0.8285,
+      "step": 3560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.950330732476235e-05,
+      "loss": 0.6962,
+      "step": 3565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9501917358349134e-05,
+      "loss": 0.8291,
+      "step": 3570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9500525469349955e-05,
+      "loss": 0.9397,
+      "step": 3575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.949913165787403e-05,
+      "loss": 0.8595,
+      "step": 3580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.949773592403072e-05,
+      "loss": 0.8645,
+      "step": 3585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.949633826792955e-05,
+      "loss": 0.6876,
+      "step": 3590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.949493868968019e-05,
+      "loss": 0.6638,
+      "step": 3595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9493537189392445e-05,
+      "loss": 0.9154,
+      "step": 3600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.949213376717631e-05,
+      "loss": 0.6988,
+      "step": 3605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9490728423141896e-05,
+      "loss": 0.716,
+      "step": 3610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9489321157399475e-05,
+      "loss": 0.7978,
+      "step": 3615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.948791197005947e-05,
+      "loss": 0.866,
+      "step": 3620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.948650086123245e-05,
+      "loss": 0.9086,
+      "step": 3625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.948508783102916e-05,
+      "loss": 0.894,
+      "step": 3630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9483672879560454e-05,
+      "loss": 0.8542,
+      "step": 3635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.948225600693737e-05,
+      "loss": 0.8554,
+      "step": 3640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.948083721327108e-05,
+      "loss": 0.8713,
+      "step": 3645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.947941649867292e-05,
+      "loss": 0.897,
+      "step": 3650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.947799386325436e-05,
+      "loss": 0.8087,
+      "step": 3655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.947656930712703e-05,
+      "loss": 0.8883,
+      "step": 3660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.947514283040272e-05,
+      "loss": 0.8465,
+      "step": 3665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.947371443319335e-05,
+      "loss": 0.8383,
+      "step": 3670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9472284115611005e-05,
+      "loss": 0.8494,
+      "step": 3675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.947085187776792e-05,
+      "loss": 0.7423,
+      "step": 3680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9469417719776477e-05,
+      "loss": 0.7621,
+      "step": 3685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9467981641749216e-05,
+      "loss": 0.7575,
+      "step": 3690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9466543643798805e-05,
+      "loss": 0.8,
+      "step": 3695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9465103726038096e-05,
+      "loss": 0.7279,
+      "step": 3700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.946366188858006e-05,
+      "loss": 0.9188,
+      "step": 3705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9462218131537845e-05,
+      "loss": 0.901,
+      "step": 3710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.946077245502474e-05,
+      "loss": 0.8387,
+      "step": 3715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9459324859154167e-05,
+      "loss": 0.7089,
+      "step": 3720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.945787534403973e-05,
+      "loss": 0.8077,
+      "step": 3725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.945642390979516e-05,
+      "loss": 0.8124,
+      "step": 3730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.945497055653434e-05,
+      "loss": 0.8187,
+      "step": 3735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.945351528437132e-05,
+      "loss": 0.8432,
+      "step": 3740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.945205809342029e-05,
+      "loss": 1.0281,
+      "step": 3745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.945059898379559e-05,
+      "loss": 0.946,
+      "step": 3750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.944913795561171e-05,
+      "loss": 0.9212,
+      "step": 3755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9447675008983295e-05,
+      "loss": 0.9404,
+      "step": 3760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9446210144025134e-05,
+      "loss": 0.8842,
+      "step": 3765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.944474336085217e-05,
+      "loss": 0.7175,
+      "step": 3770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.94432746595795e-05,
+      "loss": 0.8221,
+      "step": 3775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.944180404032236e-05,
+      "loss": 0.6941,
+      "step": 3780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9440331503196165e-05,
+      "loss": 0.9288,
+      "step": 3785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9438857048316434e-05,
+      "loss": 0.7445,
+      "step": 3790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.943738067579888e-05,
+      "loss": 0.8878,
+      "step": 3795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.943590238575935e-05,
+      "loss": 0.8777,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.943442217831382e-05,
+      "loss": 0.9116,
+      "step": 3805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.943294005357846e-05,
+      "loss": 0.9225,
+      "step": 3810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.943145601166956e-05,
+      "loss": 0.8456,
+      "step": 3815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.942997005270357e-05,
+      "loss": 0.7643,
+      "step": 3820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.942848217679709e-05,
+      "loss": 0.7678,
+      "step": 3825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.942699238406685e-05,
+      "loss": 0.796,
+      "step": 3830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9425500674629775e-05,
+      "loss": 0.9026,
+      "step": 3835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9424007048602905e-05,
+      "loss": 0.8451,
+      "step": 3840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.942251150610343e-05,
+      "loss": 0.8219,
+      "step": 3845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9421014047248716e-05,
+      "loss": 0.758,
+      "step": 3850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9419514672156255e-05,
+      "loss": 0.8383,
+      "step": 3855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.941801338094369e-05,
+      "loss": 0.858,
+      "step": 3860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.941651017372885e-05,
+      "loss": 0.8233,
+      "step": 3865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9415005050629654e-05,
+      "loss": 0.9621,
+      "step": 3870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.941349801176422e-05,
+      "loss": 0.9742,
+      "step": 3875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9411989057250805e-05,
+      "loss": 0.8342,
+      "step": 3880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.941047818720781e-05,
+      "loss": 0.7426,
+      "step": 3885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9408965401753776e-05,
+      "loss": 0.857,
+      "step": 3890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9407450701007416e-05,
+      "loss": 0.8559,
+      "step": 3895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9405934085087584e-05,
+      "loss": 0.7515,
+      "step": 3900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.940441555411328e-05,
+      "loss": 0.8281,
+      "step": 3905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.940289510820367e-05,
+      "loss": 0.9546,
+      "step": 3910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9401372747478035e-05,
+      "loss": 0.8277,
+      "step": 3915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.939984847205585e-05,
+      "loss": 0.9051,
+      "step": 3920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.939832228205672e-05,
+      "loss": 0.9353,
+      "step": 3925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9396794177600384e-05,
+      "loss": 0.928,
+      "step": 3930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9395264158806764e-05,
+      "loss": 0.7046,
+      "step": 3935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9393732225795906e-05,
+      "loss": 0.7513,
+      "step": 3940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.939219837868801e-05,
+      "loss": 0.7943,
+      "step": 3945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.939066261760346e-05,
+      "loss": 0.7632,
+      "step": 3950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9389124942662727e-05,
+      "loss": 0.7753,
+      "step": 3955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.938758535398649e-05,
+      "loss": 0.8438,
+      "step": 3960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.938604385169555e-05,
+      "loss": 0.8776,
+      "step": 3965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.938450043591085e-05,
+      "loss": 0.6746,
+      "step": 3970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.938295510675352e-05,
+      "loss": 0.7182,
+      "step": 3975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.938140786434481e-05,
+      "loss": 0.9462,
+      "step": 3980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.937985870880612e-05,
+      "loss": 0.8194,
+      "step": 3985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9378307640259005e-05,
+      "loss": 1.0525,
+      "step": 3990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.937675465882517e-05,
+      "loss": 0.8388,
+      "step": 3995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.93751997646265e-05,
+      "loss": 0.8251,
+      "step": 4000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.937364295778497e-05,
+      "loss": 0.7671,
+      "step": 4005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.937208423842276e-05,
+      "loss": 1.005,
+      "step": 4010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9370523606662155e-05,
+      "loss": 0.9077,
+      "step": 4015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9368961062625626e-05,
+      "loss": 0.943,
+      "step": 4020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9367396606435787e-05,
+      "loss": 0.7335,
+      "step": 4025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9365830238215395e-05,
+      "loss": 0.8139,
+      "step": 4030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9364261958087346e-05,
+      "loss": 0.8767,
+      "step": 4035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9362691766174704e-05,
+      "loss": 0.8758,
+      "step": 4040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9361119662600674e-05,
+      "loss": 0.8697,
+      "step": 4045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9359545647488624e-05,
+      "loss": 0.8138,
+      "step": 4050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.935796972096205e-05,
+      "loss": 0.8757,
+      "step": 4055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.935639188314461e-05,
+      "loss": 0.7957,
+      "step": 4060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.935481213416012e-05,
+      "loss": 0.8402,
+      "step": 4065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9353230474132536e-05,
+      "loss": 0.8074,
+      "step": 4070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.935164690318596e-05,
+      "loss": 0.6974,
+      "step": 4075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.935006142144465e-05,
+      "loss": 0.8779,
+      "step": 4080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.934847402903303e-05,
+      "loss": 0.702,
+      "step": 4085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.934688472607563e-05,
+      "loss": 0.8912,
+      "step": 4090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9345293512697175e-05,
+      "loss": 0.8372,
+      "step": 4095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.934370038902253e-05,
+      "loss": 0.7317,
+      "step": 4100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.934210535517668e-05,
+      "loss": 0.901,
+      "step": 4105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9340508411284804e-05,
+      "loss": 0.7522,
+      "step": 4110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.933890955747219e-05,
+      "loss": 0.7643,
+      "step": 4115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.93373087938643e-05,
+      "loss": 0.866,
+      "step": 4120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9335706120586754e-05,
+      "loss": 0.8287,
+      "step": 4125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9334101537765296e-05,
+      "loss": 0.769,
+      "step": 4130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.933249504552583e-05,
+      "loss": 0.9201,
+      "step": 4135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.933088664399442e-05,
+      "loss": 0.6929,
+      "step": 4140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.932927633329727e-05,
+      "loss": 0.7547,
+      "step": 4145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.932766411356074e-05,
+      "loss": 0.7361,
+      "step": 4150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9326049984911326e-05,
+      "loss": 0.7689,
+      "step": 4155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.932443394747569e-05,
+      "loss": 0.7908,
+      "step": 4160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9322816001380635e-05,
+      "loss": 0.9189,
+      "step": 4165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9321196146753114e-05,
+      "loss": 0.8301,
+      "step": 4170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.931957438372024e-05,
+      "loss": 0.8996,
+      "step": 4175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.931795071240926e-05,
+      "loss": 0.764,
+      "step": 4180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.931632513294758e-05,
+      "loss": 0.803,
+      "step": 4185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.931469764546275e-05,
+      "loss": 0.9746,
+      "step": 4190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9313068250082486e-05,
+      "loss": 0.8741,
+      "step": 4195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9311436946934624e-05,
+      "loss": 0.9009,
+      "step": 4200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.930980373614718e-05,
+      "loss": 0.7202,
+      "step": 4205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9308168617848304e-05,
+      "loss": 0.8097,
+      "step": 4210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9306531592166294e-05,
+      "loss": 0.6738,
+      "step": 4215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.930489265922961e-05,
+      "loss": 0.7737,
+      "step": 4220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.930325181916684e-05,
+      "loss": 0.8228,
+      "step": 4225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9301609072106756e-05,
+      "loss": 0.8239,
+      "step": 4230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.929996441817825e-05,
+      "loss": 0.9161,
+      "step": 4235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9298317857510354e-05,
+      "loss": 0.9567,
+      "step": 4240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9296669390232294e-05,
+      "loss": 0.8271,
+      "step": 4245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.929501901647341e-05,
+      "loss": 0.6943,
+      "step": 4250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9293366736363205e-05,
+      "loss": 0.6149,
+      "step": 4255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.929171255003132e-05,
+      "loss": 0.7123,
+      "step": 4260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.929005645760756e-05,
+      "loss": 0.8766,
+      "step": 4265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9288398459221875e-05,
+      "loss": 0.9577,
+      "step": 4270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.928673855500435e-05,
+      "loss": 0.7385,
+      "step": 4275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9285076745085256e-05,
+      "loss": 0.8711,
+      "step": 4280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9283413029594964e-05,
+      "loss": 0.8776,
+      "step": 4285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.928174740866404e-05,
+      "loss": 0.8267,
+      "step": 4290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.928007988242317e-05,
+      "loss": 0.7509,
+      "step": 4295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.92784104510032e-05,
+      "loss": 1.0717,
+      "step": 4300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.927673911453512e-05,
+      "loss": 0.7941,
+      "step": 4305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9275065873150086e-05,
+      "loss": 0.8065,
+      "step": 4310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9273390726979384e-05,
+      "loss": 0.7789,
+      "step": 4315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9271713676154465e-05,
+      "loss": 0.8482,
+      "step": 4320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.927003472080691e-05,
+      "loss": 0.9035,
+      "step": 4325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9268353861068475e-05,
+      "loss": 0.8393,
+      "step": 4330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.926667109707104e-05,
+      "loss": 1.0194,
+      "step": 4335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9264986428946644e-05,
+      "loss": 0.8048,
+      "step": 4340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.926329985682749e-05,
+      "loss": 0.8594,
+      "step": 4345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9261611380845915e-05,
+      "loss": 0.8202,
+      "step": 4350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9259921001134404e-05,
+      "loss": 0.8064,
+      "step": 4355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.925822871782559e-05,
+      "loss": 0.804,
+      "step": 4360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9256534531052265e-05,
+      "loss": 0.9993,
+      "step": 4365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9254838440947374e-05,
+      "loss": 0.904,
+      "step": 4370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9253140447644e-05,
+      "loss": 0.8427,
+      "step": 4375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.925144055127537e-05,
+      "loss": 0.8224,
+      "step": 4380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9249738751974886e-05,
+      "loss": 0.9162,
+      "step": 4385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9248035049876074e-05,
+      "loss": 0.8402,
+      "step": 4390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9246329445112604e-05,
+      "loss": 0.7632,
+      "step": 4395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9244621937818335e-05,
+      "loss": 0.73,
+      "step": 4400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.924291252812723e-05,
+      "loss": 0.747,
+      "step": 4405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9241201216173435e-05,
+      "loss": 0.7069,
+      "step": 4410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.923948800209122e-05,
+      "loss": 0.811,
+      "step": 4415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.923777288601502e-05,
+      "loss": 0.8599,
+      "step": 4420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.923605586807941e-05,
+      "loss": 0.8569,
+      "step": 4425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.923433694841913e-05,
+      "loss": 0.8376,
+      "step": 4430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9232616127169045e-05,
+      "loss": 0.6783,
+      "step": 4435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.923089340446419e-05,
+      "loss": 0.7608,
+      "step": 4440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.922916878043974e-05,
+      "loss": 0.9222,
+      "step": 4445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.922744225523102e-05,
+      "loss": 0.8178,
+      "step": 4450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.922571382897351e-05,
+      "loss": 0.7488,
+      "step": 4455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.922398350180282e-05,
+      "loss": 0.9123,
+      "step": 4460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9222251273854734e-05,
+      "loss": 0.7817,
+      "step": 4465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9220517145265175e-05,
+      "loss": 0.8018,
+      "step": 4470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9218781116170215e-05,
+      "loss": 0.6371,
+      "step": 4475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.921704318670607e-05,
+      "loss": 0.8798,
+      "step": 4480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.921530335700911e-05,
+      "loss": 0.789,
+      "step": 4485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9213561627215855e-05,
+      "loss": 0.8795,
+      "step": 4490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.921181799746297e-05,
+      "loss": 0.7122,
+      "step": 4495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.921007246788728e-05,
+      "loss": 0.7823,
+      "step": 4500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.920832503862575e-05,
+      "loss": 0.8281,
+      "step": 4505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.920657570981548e-05,
+      "loss": 0.8589,
+      "step": 4510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.920482448159375e-05,
+      "loss": 0.916,
+      "step": 4515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.920307135409797e-05,
+      "loss": 0.762,
+      "step": 4520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.92013163274657e-05,
+      "loss": 0.7822,
+      "step": 4525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.919955940183465e-05,
+      "loss": 0.8692,
+      "step": 4530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9197800577342677e-05,
+      "loss": 0.886,
+      "step": 4535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9196039854127805e-05,
+      "loss": 0.8793,
+      "step": 4540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9194277232328176e-05,
+      "loss": 0.7658,
+      "step": 4545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.919251271208211e-05,
+      "loss": 0.7981,
+      "step": 4550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9190746293528056e-05,
+      "loss": 0.8411,
+      "step": 4555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.918897797680462e-05,
+      "loss": 0.9487,
+      "step": 4560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.918720776205055e-05,
+      "loss": 0.7555,
+      "step": 4565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9185435649404766e-05,
+      "loss": 0.7755,
+      "step": 4570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.91836616390063e-05,
+      "loss": 0.914,
+      "step": 4575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9181885730994356e-05,
+      "loss": 0.7696,
+      "step": 4580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9180107925508306e-05,
+      "loss": 0.8708,
+      "step": 4585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9178328222687625e-05,
+      "loss": 0.7438,
+      "step": 4590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.917654662267197e-05,
+      "loss": 0.8309,
+      "step": 4595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.917476312560113e-05,
+      "loss": 0.7166,
+      "step": 4600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.917297773161506e-05,
+      "loss": 0.8975,
+      "step": 4605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.917119044085385e-05,
+      "loss": 0.7804,
+      "step": 4610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9169401253457737e-05,
+      "loss": 0.97,
+      "step": 4615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9167610169567127e-05,
+      "loss": 0.9211,
+      "step": 4620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9165817189322546e-05,
+      "loss": 0.8738,
+      "step": 4625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.916402231286469e-05,
+      "loss": 0.8577,
+      "step": 4630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9162225540334396e-05,
+      "loss": 0.8111,
+      "step": 4635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9160426871872645e-05,
+      "loss": 0.7975,
+      "step": 4640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9158626307620586e-05,
+      "loss": 0.7764,
+      "step": 4645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.91568238477195e-05,
+      "loss": 0.9106,
+      "step": 4650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.915501949231081e-05,
+      "loss": 0.8469,
+      "step": 4655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.91532132415361e-05,
+      "loss": 0.9166,
+      "step": 4660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9151405095537116e-05,
+      "loss": 0.7444,
+      "step": 4665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.914959505445572e-05,
+      "loss": 0.8629,
+      "step": 4670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9147783118433947e-05,
+      "loss": 0.8151,
+      "step": 4675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9145969287613976e-05,
+      "loss": 0.597,
+      "step": 4680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9144153562138126e-05,
+      "loss": 0.7783,
+      "step": 4685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9142335942148873e-05,
+      "loss": 0.7502,
+      "step": 4690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9140516427788845e-05,
+      "loss": 0.7419,
+      "step": 4695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.913869501920081e-05,
+      "loss": 0.8133,
+      "step": 4700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.913687171652769e-05,
+      "loss": 0.7746,
+      "step": 4705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9135046519912545e-05,
+      "loss": 0.7946,
+      "step": 4710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.91332194294986e-05,
+      "loss": 0.8155,
+      "step": 4715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9131390445429224e-05,
+      "loss": 0.8734,
+      "step": 4720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9129559567847926e-05,
+      "loss": 0.732,
+      "step": 4725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9127726796898367e-05,
+      "loss": 0.8291,
+      "step": 4730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.912589213272436e-05,
+      "loss": 0.6404,
+      "step": 4735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.912405557546987e-05,
+      "loss": 0.8591,
+      "step": 4740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9122217125279e-05,
+      "loss": 0.6889,
+      "step": 4745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.912037678229602e-05,
+      "loss": 0.9284,
+      "step": 4750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9118534546665314e-05,
+      "loss": 0.7802,
+      "step": 4755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.911669041853145e-05,
+      "loss": 0.9108,
+      "step": 4760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9114844398039125e-05,
+      "loss": 0.7587,
+      "step": 4765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.91129964853332e-05,
+      "loss": 0.9941,
+      "step": 4770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.911114668055867e-05,
+      "loss": 0.884,
+      "step": 4775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9109294983860675e-05,
+      "loss": 0.8066,
+      "step": 4780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9107441395384526e-05,
+      "loss": 0.8286,
+      "step": 4785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.910558591527565e-05,
+      "loss": 0.7533,
+      "step": 4790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.910372854367966e-05,
+      "loss": 0.9376,
+      "step": 4795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.910186928074229e-05,
+      "loss": 0.8105,
+      "step": 4800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.910000812660942e-05,
+      "loss": 0.679,
+      "step": 4805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.90981450814271e-05,
+      "loss": 0.7108,
+      "step": 4810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9096280145341514e-05,
+      "loss": 0.8678,
+      "step": 4815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9094413318499e-05,
+      "loss": 0.9949,
+      "step": 4820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.909254460104604e-05,
+      "loss": 0.8043,
+      "step": 4825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9090673993129265e-05,
+      "loss": 0.7961,
+      "step": 4830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9088801494895465e-05,
+      "loss": 0.7947,
+      "step": 4835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.908692710649156e-05,
+      "loss": 0.7128,
+      "step": 4840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.908505082806462e-05,
+      "loss": 0.6772,
+      "step": 4845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.908317265976188e-05,
+      "loss": 0.5842,
+      "step": 4850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9081292601730714e-05,
+      "loss": 0.8447,
+      "step": 4855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.907941065411864e-05,
+      "loss": 0.8872,
+      "step": 4860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.907752681707333e-05,
+      "loss": 0.6919,
+      "step": 4865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9075641090742606e-05,
+      "loss": 0.7229,
+      "step": 4870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.907375347527443e-05,
+      "loss": 0.9401,
+      "step": 4875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9071863970816925e-05,
+      "loss": 0.8628,
+      "step": 4880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.906997257751834e-05,
+      "loss": 0.8033,
+      "step": 4885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.90680792955271e-05,
+      "loss": 0.844,
+      "step": 4890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.906618412499176e-05,
+      "loss": 0.9655,
+      "step": 4895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.906428706606102e-05,
+      "loss": 0.9286,
+      "step": 4900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.906238811888375e-05,
+      "loss": 0.8596,
+      "step": 4905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9060487283608945e-05,
+      "loss": 0.7534,
+      "step": 4910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.905858456038577e-05,
+      "loss": 0.7926,
+      "step": 4915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.90566799493635e-05,
+      "loss": 0.7473,
+      "step": 4920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.905477345069161e-05,
+      "loss": 0.8419,
+      "step": 4925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.905286506451968e-05,
+      "loss": 0.7639,
+      "step": 4930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.905095479099747e-05,
+      "loss": 0.8235,
+      "step": 4935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.904904263027486e-05,
+      "loss": 0.7668,
+      "step": 4940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.904712858250189e-05,
+      "loss": 0.7284,
+      "step": 4945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.904521264782875e-05,
+      "loss": 0.7893,
+      "step": 4950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.904329482640579e-05,
+      "loss": 0.8401,
+      "step": 4955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.904137511838348e-05,
+      "loss": 0.96,
+      "step": 4960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.903945352391247e-05,
+      "loss": 0.8928,
+      "step": 4965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.903753004314352e-05,
+      "loss": 0.8379,
+      "step": 4970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9035604676227574e-05,
+      "loss": 0.8085,
+      "step": 4975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.903367742331571e-05,
+      "loss": 0.7874,
+      "step": 4980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9031748284559145e-05,
+      "loss": 0.6945,
+      "step": 4985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.902981726010926e-05,
+      "loss": 0.9188,
+      "step": 4990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.902788435011757e-05,
+      "loss": 0.8353,
+      "step": 4995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9025949554735736e-05,
+      "loss": 0.8942,
+      "step": 5000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.90240128741156e-05,
+      "loss": 0.6747,
+      "step": 5005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.90220743084091e-05,
+      "loss": 0.6895,
+      "step": 5010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.902013385776838e-05,
+      "loss": 0.9258,
+      "step": 5015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.901819152234567e-05,
+      "loss": 0.678,
+      "step": 5020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9016247302293396e-05,
+      "loss": 0.8835,
+      "step": 5025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.90143011977641e-05,
+      "loss": 0.934,
+      "step": 5030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.901235320891051e-05,
+      "loss": 0.7437,
+      "step": 5035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.901040333588545e-05,
+      "loss": 0.9084,
+      "step": 5040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.900845157884195e-05,
+      "loss": 1.0283,
+      "step": 5045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.900649793793313e-05,
+      "loss": 1.0257,
+      "step": 5050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9004542413312305e-05,
+      "loss": 0.8054,
+      "step": 5055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9002585005132914e-05,
+      "loss": 0.7625,
+      "step": 5060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9000625713548545e-05,
+      "loss": 0.7329,
+      "step": 5065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.899866453871294e-05,
+      "loss": 0.7228,
+      "step": 5070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.899670148077998e-05,
+      "loss": 0.8376,
+      "step": 5075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.899473653990371e-05,
+      "loss": 0.9449,
+      "step": 5080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.899276971623831e-05,
+      "loss": 0.9834,
+      "step": 5085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8990801009938103e-05,
+      "loss": 0.8887,
+      "step": 5090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.898883042115758e-05,
+      "loss": 0.8239,
+      "step": 5095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.898685795005135e-05,
+      "loss": 0.7858,
+      "step": 5100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.89848835967742e-05,
+      "loss": 0.9186,
+      "step": 5105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.898290736148105e-05,
+      "loss": 0.7272,
+      "step": 5110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.898092924432695e-05,
+      "loss": 0.606,
+      "step": 5115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.897894924546714e-05,
+      "loss": 0.7839,
+      "step": 5120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.897696736505698e-05,
+      "loss": 0.7314,
+      "step": 5125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8974983603251964e-05,
+      "loss": 0.9404,
+      "step": 5130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8972997960207776e-05,
+      "loss": 0.8616,
+      "step": 5135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.897101043608021e-05,
+      "loss": 0.7068,
+      "step": 5140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.896902103102522e-05,
+      "loss": 0.79,
+      "step": 5145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.896702974519891e-05,
+      "loss": 0.8714,
+      "step": 5150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8965036578757536e-05,
+      "loss": 0.7654,
+      "step": 5155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.896304153185748e-05,
+      "loss": 0.8567,
+      "step": 5160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8961044604655304e-05,
+      "loss": 0.8752,
+      "step": 5165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.895904579730769e-05,
+      "loss": 0.7139,
+      "step": 5170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8957045109971476e-05,
+      "loss": 0.936,
+      "step": 5175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.895504254280366e-05,
+      "loss": 0.6209,
+      "step": 5180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8953038095961366e-05,
+      "loss": 0.9938,
+      "step": 5185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.895103176960188e-05,
+      "loss": 0.6716,
+      "step": 5190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.894902356388263e-05,
+      "loss": 0.9064,
+      "step": 5195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.894701347896121e-05,
+      "loss": 0.7788,
+      "step": 5200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.894500151499532e-05,
+      "loss": 0.7092,
+      "step": 5205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.894298767214285e-05,
+      "loss": 0.8147,
+      "step": 5210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.894097195056181e-05,
+      "loss": 0.9334,
+      "step": 5215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.893895435041037e-05,
+      "loss": 0.7708,
+      "step": 5220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8936934871846854e-05,
+      "loss": 0.8733,
+      "step": 5225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8934913515029703e-05,
+      "loss": 0.8391,
+      "step": 5230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.893289028011755e-05,
+      "loss": 0.7536,
+      "step": 5235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8930865167269126e-05,
+      "loss": 0.8303,
+      "step": 5240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8928838176643357e-05,
+      "loss": 0.8649,
+      "step": 5245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.892680930839929e-05,
+      "loss": 0.8889,
+      "step": 5250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.892477856269611e-05,
+      "loss": 0.9678,
+      "step": 5255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.892274593969318e-05,
+      "loss": 0.7796,
+      "step": 5260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8920711439549985e-05,
+      "loss": 0.6292,
+      "step": 5265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.891867506242617e-05,
+      "loss": 0.9088,
+      "step": 5270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8916636808481514e-05,
+      "loss": 0.8349,
+      "step": 5275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8914596677875965e-05,
+      "loss": 0.6519,
+      "step": 5280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8912554670769596e-05,
+      "loss": 0.8597,
+      "step": 5285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.891051078732263e-05,
+      "loss": 0.7662,
+      "step": 5290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.890846502769547e-05,
+      "loss": 0.789,
+      "step": 5295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8906417392048606e-05,
+      "loss": 0.8971,
+      "step": 5300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.890436788054274e-05,
+      "loss": 0.8498,
+      "step": 5305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.890231649333867e-05,
+      "loss": 0.854,
+      "step": 5310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8900263230597377e-05,
+      "loss": 0.9026,
+      "step": 5315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.889820809247996e-05,
+      "loss": 0.7707,
+      "step": 5320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.889615107914768e-05,
+      "loss": 0.8193,
+      "step": 5325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.889409219076195e-05,
+      "loss": 0.8653,
+      "step": 5330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.889203142748433e-05,
+      "loss": 0.848,
+      "step": 5335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.888996878947652e-05,
+      "loss": 0.8134,
+      "step": 5340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.888790427690035e-05,
+      "loss": 0.6155,
+      "step": 5345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.888583788991783e-05,
+      "loss": 0.8795,
+      "step": 5350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8883769628691106e-05,
+      "loss": 0.8331,
+      "step": 5355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.888169949338246e-05,
+      "loss": 0.7025,
+      "step": 5360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8879627484154335e-05,
+      "loss": 0.8499,
+      "step": 5365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8877553601169314e-05,
+      "loss": 0.7206,
+      "step": 5370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8875477844590125e-05,
+      "loss": 1.0382,
+      "step": 5375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8873400214579634e-05,
+      "loss": 0.7875,
+      "step": 5380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8871320711300893e-05,
+      "loss": 0.7156,
+      "step": 5385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8869239334917054e-05,
+      "loss": 0.9238,
+      "step": 5390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8867156085591444e-05,
+      "loss": 0.6847,
+      "step": 5395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.886507096348752e-05,
+      "loss": 0.6943,
+      "step": 5400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8862983968768905e-05,
+      "loss": 0.8144,
+      "step": 5405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.886089510159936e-05,
+      "loss": 0.8644,
+      "step": 5410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8858804362142775e-05,
+      "loss": 0.8313,
+      "step": 5415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.885671175056322e-05,
+      "loss": 0.6607,
+      "step": 5420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8854617267024885e-05,
+      "loss": 0.8752,
+      "step": 5425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.885252091169214e-05,
+      "loss": 0.7901,
+      "step": 5430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8850422684729436e-05,
+      "loss": 0.9387,
+      "step": 5435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.884832258630145e-05,
+      "loss": 0.7988,
+      "step": 5440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.884622061657297e-05,
+      "loss": 0.7975,
+      "step": 5445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.884411677570891e-05,
+      "loss": 0.9318,
+      "step": 5450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.884201106387436e-05,
+      "loss": 0.8113,
+      "step": 5455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.883990348123456e-05,
+      "loss": 0.8999,
+      "step": 5460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8837794027954864e-05,
+      "loss": 0.8144,
+      "step": 5465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.883568270420081e-05,
+      "loss": 0.6924,
+      "step": 5470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.883356951013807e-05,
+      "loss": 0.7643,
+      "step": 5475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8831454445932444e-05,
+      "loss": 0.8673,
+      "step": 5480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.88293375117499e-05,
+      "loss": 0.7358,
+      "step": 5485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.882721870775655e-05,
+      "loss": 0.8188,
+      "step": 5490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.882509803411865e-05,
+      "loss": 0.7592,
+      "step": 5495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8822975491002606e-05,
+      "loss": 0.8642,
+      "step": 5500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8820851078574955e-05,
+      "loss": 0.8981,
+      "step": 5505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8818724797002404e-05,
+      "loss": 0.7225,
+      "step": 5510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.881659664645178e-05,
+      "loss": 0.7061,
+      "step": 5515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8814466627090094e-05,
+      "loss": 0.7144,
+      "step": 5520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.881233473908447e-05,
+      "loss": 0.774,
+      "step": 5525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8810200982602186e-05,
+      "loss": 0.9191,
+      "step": 5530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.880806535781069e-05,
+      "loss": 0.7728,
+      "step": 5535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8805927864877526e-05,
+      "loss": 0.7725,
+      "step": 5540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.880378850397044e-05,
+      "loss": 0.8261,
+      "step": 5545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.88016472752573e-05,
+      "loss": 0.7638,
+      "step": 5550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.879950417890611e-05,
+      "loss": 0.8227,
+      "step": 5555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.879735921508504e-05,
+      "loss": 0.9262,
+      "step": 5560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8795212383962396e-05,
+      "loss": 0.8089,
+      "step": 5565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.879306368570663e-05,
+      "loss": 0.8948,
+      "step": 5570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.879091312048635e-05,
+      "loss": 0.7509,
+      "step": 5575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.878876068847029e-05,
+      "loss": 0.8049,
+      "step": 5580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8786606389827366e-05,
+      "loss": 0.694,
+      "step": 5585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8784450224726606e-05,
+      "loss": 0.7351,
+      "step": 5590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8782292193337195e-05,
+      "loss": 0.8148,
+      "step": 5595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.878013229582847e-05,
+      "loss": 0.7501,
+      "step": 5600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8777970532369915e-05,
+      "loss": 0.7474,
+      "step": 5605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.877580690313115e-05,
+      "loss": 0.9247,
+      "step": 5610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.877364140828196e-05,
+      "loss": 0.8913,
+      "step": 5615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.877147404799224e-05,
+      "loss": 0.8387,
+      "step": 5620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.876930482243208e-05,
+      "loss": 0.7817,
+      "step": 5625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.876713373177168e-05,
+      "loss": 0.7441,
+      "step": 5630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.87649607761814e-05,
+      "loss": 0.806,
+      "step": 5635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.876278595583176e-05,
+      "loss": 0.6691,
+      "step": 5640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.876060927089338e-05,
+      "loss": 0.918,
+      "step": 5645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.875843072153709e-05,
+      "loss": 0.9452,
+      "step": 5650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.87562503079338e-05,
+      "loss": 0.7853,
+      "step": 5655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.875406803025463e-05,
+      "loss": 0.7425,
+      "step": 5660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.875188388867081e-05,
+      "loss": 0.8371,
+      "step": 5665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.87496978833537e-05,
+      "loss": 0.7431,
+      "step": 5670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.874751001447486e-05,
+      "loss": 0.8835,
+      "step": 5675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.874532028220595e-05,
+      "loss": 0.8855,
+      "step": 5680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8743128686718797e-05,
+      "loss": 0.8672,
+      "step": 5685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.874093522818536e-05,
+      "loss": 0.9386,
+      "step": 5690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.873873990677775e-05,
+      "loss": 0.9365,
+      "step": 5695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8736542722668234e-05,
+      "loss": 0.8297,
+      "step": 5700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8734343676029216e-05,
+      "loss": 0.7469,
+      "step": 5705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.873214276703325e-05,
+      "loss": 0.9627,
+      "step": 5710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8729939995853034e-05,
+      "loss": 0.7226,
+      "step": 5715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.872773536266142e-05,
+      "loss": 0.8238,
+      "step": 5720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8725528867631375e-05,
+      "loss": 0.8192,
+      "step": 5725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8723320510936064e-05,
+      "loss": 0.8981,
+      "step": 5730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.872111029274875e-05,
+      "loss": 0.7524,
+      "step": 5735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.871889821324286e-05,
+      "loss": 0.6757,
+      "step": 5740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8716684272591984e-05,
+      "loss": 0.9226,
+      "step": 5745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8714468470969836e-05,
+      "loss": 0.9441,
+      "step": 5750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.871225080855028e-05,
+      "loss": 0.8892,
+      "step": 5755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8710031285507326e-05,
+      "loss": 0.8726,
+      "step": 5760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.870780990201514e-05,
+      "loss": 0.6471,
+      "step": 5765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.870558665824803e-05,
+      "loss": 0.9857,
+      "step": 5770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.870336155438044e-05,
+      "loss": 0.8579,
+      "step": 5775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.870113459058696e-05,
+      "loss": 0.8928,
+      "step": 5780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.869890576704234e-05,
+      "loss": 0.6858,
+      "step": 5785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.869667508392148e-05,
+      "loss": 0.7689,
+      "step": 5790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.86944425413994e-05,
+      "loss": 0.9229,
+      "step": 5795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8692208139651274e-05,
+      "loss": 0.7077,
+      "step": 5800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.868997187885244e-05,
+      "loss": 0.9604,
+      "step": 5805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.868773375917837e-05,
+      "loss": 0.8189,
+      "step": 5810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.868549378080468e-05,
+      "loss": 0.6778,
+      "step": 5815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.868325194390714e-05,
+      "loss": 0.8863,
+      "step": 5820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8681008248661645e-05,
+      "loss": 0.7959,
+      "step": 5825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8678762695244265e-05,
+      "loss": 0.7375,
+      "step": 5830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8676515283831195e-05,
+      "loss": 0.7799,
+      "step": 5835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.867426601459877e-05,
+      "loss": 0.8646,
+      "step": 5840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8672014887723505e-05,
+      "loss": 0.7989,
+      "step": 5845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.866976190338203e-05,
+      "loss": 0.7567,
+      "step": 5850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.866750706175113e-05,
+      "loss": 0.8682,
+      "step": 5855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8665250363007726e-05,
+      "loss": 0.9152,
+      "step": 5860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.866299180732891e-05,
+      "loss": 0.8303,
+      "step": 5865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.866073139489188e-05,
+      "loss": 0.8081,
+      "step": 5870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8658469125874026e-05,
+      "loss": 0.7795,
+      "step": 5875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8656205000452855e-05,
+      "loss": 0.9127,
+      "step": 5880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8653939018806014e-05,
+      "loss": 0.9329,
+      "step": 5885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8651671181111325e-05,
+      "loss": 0.7977,
+      "step": 5890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.864940148754673e-05,
+      "loss": 0.8458,
+      "step": 5895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.864712993829033e-05,
+      "loss": 0.6276,
+      "step": 5900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.864485653352034e-05,
+      "loss": 0.8415,
+      "step": 5905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8642581273415183e-05,
+      "loss": 0.6388,
+      "step": 5910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.864030415815337e-05,
+      "loss": 0.9421,
+      "step": 5915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.863802518791358e-05,
+      "loss": 0.9388,
+      "step": 5920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.863574436287465e-05,
+      "loss": 0.6694,
+      "step": 5925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.863346168321553e-05,
+      "loss": 0.9516,
+      "step": 5930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8631177149115346e-05,
+      "loss": 0.9281,
+      "step": 5935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.862889076075336e-05,
+      "loss": 0.8082,
+      "step": 5940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.862660251830897e-05,
+      "loss": 0.8666,
+      "step": 5945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.862431242196173e-05,
+      "loss": 0.8533,
+      "step": 5950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8622020471891336e-05,
+      "loss": 0.9231,
+      "step": 5955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.861972666827763e-05,
+      "loss": 0.6603,
+      "step": 5960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8617431011300596e-05,
+      "loss": 0.7848,
+      "step": 5965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8615133501140374e-05,
+      "loss": 0.8393,
+      "step": 5970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.861283413797724e-05,
+      "loss": 0.9132,
+      "step": 5975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8610532921991615e-05,
+      "loss": 0.8903,
+      "step": 5980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.860822985336407e-05,
+      "loss": 0.8436,
+      "step": 5985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8605924932275315e-05,
+      "loss": 0.8425,
+      "step": 5990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.860361815890622e-05,
+      "loss": 0.8307,
+      "step": 5995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.860130953343778e-05,
+      "loss": 0.8491,
+      "step": 6000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.859899905605114e-05,
+      "loss": 0.7904,
+      "step": 6005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8596686726927616e-05,
+      "loss": 0.8774,
+      "step": 6010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8594372546248634e-05,
+      "loss": 0.7607,
+      "step": 6015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.859205651419578e-05,
+      "loss": 0.8349,
+      "step": 6020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.858973863095079e-05,
+      "loss": 0.7619,
+      "step": 6025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.858741889669554e-05,
+      "loss": 0.7539,
+      "step": 6030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8585097311612046e-05,
+      "loss": 0.8129,
+      "step": 6035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8582773875882494e-05,
+      "loss": 0.8613,
+      "step": 6040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.858044858968918e-05,
+      "loss": 0.6603,
+      "step": 6045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.857812145321455e-05,
+      "loss": 1.1471,
+      "step": 6050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.857579246664123e-05,
+      "loss": 0.9696,
+      "step": 6055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8573461630151975e-05,
+      "loss": 0.8304,
+      "step": 6060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8571128943929644e-05,
+      "loss": 0.9174,
+      "step": 6065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8568794408157305e-05,
+      "loss": 0.7374,
+      "step": 6070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.856645802301812e-05,
+      "loss": 0.7429,
+      "step": 6075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.856411978869544e-05,
+      "loss": 0.654,
+      "step": 6080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.856177970537272e-05,
+      "loss": 0.7194,
+      "step": 6085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.85594377732336e-05,
+      "loss": 0.8306,
+      "step": 6090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.855709399246181e-05,
+      "loss": 0.7886,
+      "step": 6095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8554748363241286e-05,
+      "loss": 0.8086,
+      "step": 6100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8552400885756074e-05,
+      "loss": 0.5731,
+      "step": 6105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8550051560190377e-05,
+      "loss": 0.8254,
+      "step": 6110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8547700386728534e-05,
+      "loss": 0.7931,
+      "step": 6115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.854534736555504e-05,
+      "loss": 0.8459,
+      "step": 6120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.854299249685451e-05,
+      "loss": 0.825,
+      "step": 6125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.854063578081175e-05,
+      "loss": 0.7507,
+      "step": 6130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.853827721761167e-05,
+      "loss": 0.7269,
+      "step": 6135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.853591680743933e-05,
+      "loss": 0.8861,
+      "step": 6140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.853355455047996e-05,
+      "loss": 0.9874,
+      "step": 6145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8531190446918916e-05,
+      "loss": 0.7483,
+      "step": 6150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8528824496941695e-05,
+      "loss": 0.8492,
+      "step": 6155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.852645670073395e-05,
+      "loss": 0.7936,
+      "step": 6160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.852408705848146e-05,
+      "loss": 0.8202,
+      "step": 6165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.852171557037019e-05,
+      "loss": 0.7985,
+      "step": 6170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.85193422365862e-05,
+      "loss": 0.8009,
+      "step": 6175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.851696705731573e-05,
+      "loss": 0.8705,
+      "step": 6180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.851459003274515e-05,
+      "loss": 0.8157,
+      "step": 6185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8512211163060974e-05,
+      "loss": 0.8059,
+      "step": 6190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.850983044844987e-05,
+      "loss": 0.9143,
+      "step": 6195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8507447889098634e-05,
+      "loss": 0.9241,
+      "step": 6200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.850506348519424e-05,
+      "loss": 0.7165,
+      "step": 6205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.850267723692376e-05,
+      "loss": 0.8503,
+      "step": 6210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8500289144474436e-05,
+      "loss": 0.7503,
+      "step": 6215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.849789920803367e-05,
+      "loss": 0.959,
+      "step": 6220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.849550742778899e-05,
+      "loss": 0.8166,
+      "step": 6225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.849311380392806e-05,
+      "loss": 0.7547,
+      "step": 6230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8490718336638705e-05,
+      "loss": 0.7028,
+      "step": 6235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.84883210261089e-05,
+      "loss": 0.9626,
+      "step": 6240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.848592187252674e-05,
+      "loss": 0.7794,
+      "step": 6245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8483520876080484e-05,
+      "loss": 0.9036,
+      "step": 6250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.848111803695853e-05,
+      "loss": 0.8925,
+      "step": 6255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8478713355349426e-05,
+      "loss": 0.8219,
+      "step": 6260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.847630683144185e-05,
+      "loss": 0.7414,
+      "step": 6265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.847389846542464e-05,
+      "loss": 0.8704,
+      "step": 6270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8471488257486766e-05,
+      "loss": 0.7168,
+      "step": 6275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.846907620781736e-05,
+      "loss": 0.854,
+      "step": 6280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.846666231660568e-05,
+      "loss": 0.8836,
+      "step": 6285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8464246584041145e-05,
+      "loss": 0.9883,
+      "step": 6290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.84618290103133e-05,
+      "loss": 0.752,
+      "step": 6295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.845940959561185e-05,
+      "loss": 0.7937,
+      "step": 6300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.845698834012663e-05,
+      "loss": 1.0104,
+      "step": 6305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8454565244047634e-05,
+      "loss": 0.7503,
+      "step": 6310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8452140307565e-05,
+      "loss": 0.7292,
+      "step": 6315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8449713530869e-05,
+      "loss": 0.9591,
+      "step": 6320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.844728491415005e-05,
+      "loss": 0.7253,
+      "step": 6325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.844485445759872e-05,
+      "loss": 0.6955,
+      "step": 6330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.844242216140573e-05,
+      "loss": 0.8349,
+      "step": 6335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8439988025761915e-05,
+      "loss": 0.7472,
+      "step": 6340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.843755205085829e-05,
+      "loss": 0.8337,
+      "step": 6345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.843511423688599e-05,
+      "loss": 0.768,
+      "step": 6350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.843267458403631e-05,
+      "loss": 0.7714,
+      "step": 6355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8430233092500666e-05,
+      "loss": 0.7844,
+      "step": 6360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8427789762470654e-05,
+      "loss": 0.9785,
+      "step": 6365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.842534459413798e-05,
+      "loss": 0.8883,
+      "step": 6370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8422897587694503e-05,
+      "loss": 0.7626,
+      "step": 6375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.842044874333225e-05,
+      "loss": 0.726,
+      "step": 6380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8417998061243366e-05,
+      "loss": 0.6785,
+      "step": 6385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.841554554162014e-05,
+      "loss": 0.8227,
+      "step": 6390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.841309118465503e-05,
+      "loss": 0.8717,
+      "step": 6395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.84106349905406e-05,
+      "loss": 0.551,
+      "step": 6400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.84081769594696e-05,
+      "loss": 0.8788,
+      "step": 6405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8405717091634894e-05,
+      "loss": 0.64,
+      "step": 6410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8403255387229496e-05,
+      "loss": 0.8521,
+      "step": 6415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.840079184644658e-05,
+      "loss": 0.8083,
+      "step": 6420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.839832646947944e-05,
+      "loss": 0.707,
+      "step": 6425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.839585925652153e-05,
+      "loss": 0.7592,
+      "step": 6430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.839339020776645e-05,
+      "loss": 0.8081,
+      "step": 6435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.839091932340793e-05,
+      "loss": 0.7157,
+      "step": 6440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.838844660363985e-05,
+      "loss": 0.9523,
+      "step": 6445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.838597204865625e-05,
+      "loss": 0.8926,
+      "step": 6450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.838349565865129e-05,
+      "loss": 0.7616,
+      "step": 6455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8381017433819284e-05,
+      "loss": 0.7135,
+      "step": 6460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.837853737435471e-05,
+      "loss": 0.79,
+      "step": 6465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8376055480452134e-05,
+      "loss": 0.7665,
+      "step": 6470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8373571752306327e-05,
+      "loss": 0.9513,
+      "step": 6475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.837108619011217e-05,
+      "loss": 0.7946,
+      "step": 6480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.83685987940647e-05,
+      "loss": 0.7588,
+      "step": 6485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8366109564359095e-05,
+      "loss": 0.8488,
+      "step": 6490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.836361850119069e-05,
+      "loss": 0.7705,
+      "step": 6495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.836112560475492e-05,
+      "loss": 0.8837,
+      "step": 6500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.835863087524742e-05,
+      "loss": 0.7212,
+      "step": 6505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.835613431286394e-05,
+      "loss": 0.8228,
+      "step": 6510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.835363591780037e-05,
+      "loss": 0.8566,
+      "step": 6515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.835113569025275e-05,
+      "loss": 0.7953,
+      "step": 6520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8348633630417265e-05,
+      "loss": 0.8232,
+      "step": 6525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8346129738490256e-05,
+      "loss": 0.8565,
+      "step": 6530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.834362401466818e-05,
+      "loss": 0.8472,
+      "step": 6535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.834111645914766e-05,
+      "loss": 0.8275,
+      "step": 6540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.833860707212545e-05,
+      "loss": 0.7908,
+      "step": 6545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.833609585379847e-05,
+      "loss": 0.7838,
+      "step": 6550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8333582804363756e-05,
+      "loss": 0.7953,
+      "step": 6555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.833106792401848e-05,
+      "loss": 0.7829,
+      "step": 6560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.832855121296001e-05,
+      "loss": 0.8694,
+      "step": 6565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8326032671385804e-05,
+      "loss": 0.7641,
+      "step": 6570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8323512299493495e-05,
+      "loss": 0.8141,
+      "step": 6575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8320990097480834e-05,
+      "loss": 0.7804,
+      "step": 6580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.831846606554574e-05,
+      "loss": 0.8158,
+      "step": 6585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.831594020388627e-05,
+      "loss": 0.9951,
+      "step": 6590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.83134125127006e-05,
+      "loss": 0.8265,
+      "step": 6595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.83108829921871e-05,
+      "loss": 0.7968,
+      "step": 6600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.830835164254424e-05,
+      "loss": 0.7983,
+      "step": 6605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.830581846397063e-05,
+      "loss": 0.7525,
+      "step": 6610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8303283456665056e-05,
+      "loss": 0.8614,
+      "step": 6615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.830074662082643e-05,
+      "loss": 0.7994,
+      "step": 6620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8298207956653814e-05,
+      "loss": 0.7816,
+      "step": 6625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.82956674643464e-05,
+      "loss": 0.9112,
+      "step": 6630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8293125144103535e-05,
+      "loss": 0.7195,
+      "step": 6635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.829058099612471e-05,
+      "loss": 0.8317,
+      "step": 6640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.828803502060956e-05,
+      "loss": 0.7947,
+      "step": 6645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.828548721775784e-05,
+      "loss": 0.7425,
+      "step": 6650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.828293758776949e-05,
+      "loss": 0.8062,
+      "step": 6655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8280386130844555e-05,
+      "loss": 0.8531,
+      "step": 6660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.827783284718326e-05,
+      "loss": 0.87,
+      "step": 6665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8275277736985924e-05,
+      "loss": 0.773,
+      "step": 6670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8272720800453065e-05,
+      "loss": 0.8636,
+      "step": 6675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.82701620377853e-05,
+      "loss": 0.8619,
+      "step": 6680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8267601449183416e-05,
+      "loss": 0.7709,
+      "step": 6685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8265039034848324e-05,
+      "loss": 0.7648,
+      "step": 6690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.82624747949811e-05,
+      "loss": 0.8286,
+      "step": 6695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825990872978296e-05,
+      "loss": 0.8407,
+      "step": 6700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825734083945523e-05,
+      "loss": 0.847,
+      "step": 6705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8254771124199416e-05,
+      "loss": 0.845,
+      "step": 6710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825219958421715e-05,
+      "loss": 0.8051,
+      "step": 6715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8249626219710234e-05,
+      "loss": 0.757,
+      "step": 6720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8247051030880565e-05,
+      "loss": 0.6884,
+      "step": 6725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8244474017930215e-05,
+      "loss": 0.9335,
+      "step": 6730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.82418951810614e-05,
+      "loss": 0.7192,
+      "step": 6735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.823931452047647e-05,
+      "loss": 0.8617,
+      "step": 6740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.823673203637794e-05,
+      "loss": 0.8135,
+      "step": 6745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8234147728968414e-05,
+      "loss": 0.853,
+      "step": 6750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8231561598450706e-05,
+      "loss": 0.7494,
+      "step": 6755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8228973645027713e-05,
+      "loss": 0.7688,
+      "step": 6760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.822638386890253e-05,
+      "loss": 0.7991,
+      "step": 6765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.822379227027834e-05,
+      "loss": 0.826,
+      "step": 6770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8221198849358527e-05,
+      "loss": 0.8499,
+      "step": 6775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.821860360634657e-05,
+      "loss": 0.7452,
+      "step": 6780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8216006541446117e-05,
+      "loss": 0.7455,
+      "step": 6785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8213407654860945e-05,
+      "loss": 0.7513,
+      "step": 6790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.821080694679498e-05,
+      "loss": 0.8145,
+      "step": 6795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.820820441745231e-05,
+      "loss": 0.8254,
+      "step": 6800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.820560006703711e-05,
+      "loss": 0.7937,
+      "step": 6805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.820299389575378e-05,
+      "loss": 0.7425,
+      "step": 6810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.820038590380678e-05,
+      "loss": 0.8957,
+      "step": 6815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8197776091400766e-05,
+      "loss": 0.8821,
+      "step": 6820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.819516445874052e-05,
+      "loss": 0.7784,
+      "step": 6825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.819255100603097e-05,
+      "loss": 0.618,
+      "step": 6830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.818993573347719e-05,
+      "loss": 0.8112,
+      "step": 6835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8187318641284376e-05,
+      "loss": 0.9088,
+      "step": 6840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8184699729657895e-05,
+      "loss": 0.8004,
+      "step": 6845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.818207899880325e-05,
+      "loss": 0.8575,
+      "step": 6850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817945644892606e-05,
+      "loss": 0.7702,
+      "step": 6855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817683208023213e-05,
+      "loss": 0.9045,
+      "step": 6860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817420589292738e-05,
+      "loss": 0.8171,
+      "step": 6865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8171577887217866e-05,
+      "loss": 0.9869,
+      "step": 6870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8168948063309814e-05,
+      "loss": 0.8393,
+      "step": 6875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8166316421409574e-05,
+      "loss": 0.8222,
+      "step": 6880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8163682961723646e-05,
+      "loss": 0.8546,
+      "step": 6885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.816104768445865e-05,
+      "loss": 0.8804,
+      "step": 6890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.815841058982139e-05,
+      "loss": 0.9524,
+      "step": 6895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.815577167801878e-05,
+      "loss": 0.7327,
+      "step": 6900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8153130949257884e-05,
+      "loss": 0.9141,
+      "step": 6905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8150488403745925e-05,
+      "loss": 0.8751,
+      "step": 6910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8147844041690244e-05,
+      "loss": 0.7618,
+      "step": 6915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.814519786329833e-05,
+      "loss": 0.6011,
+      "step": 6920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.814254986877784e-05,
+      "loss": 0.7932,
+      "step": 6925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.813990005833653e-05,
+      "loss": 0.8013,
+      "step": 6930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8137248432182334e-05,
+      "loss": 0.9176,
+      "step": 6935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.813459499052332e-05,
+      "loss": 0.8428,
+      "step": 6940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8131939733567686e-05,
+      "loss": 0.8801,
+      "step": 6945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.812928266152379e-05,
+      "loss": 0.9083,
+      "step": 6950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.812662377460012e-05,
+      "loss": 0.6987,
+      "step": 6955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.812396307300531e-05,
+      "loss": 0.8763,
+      "step": 6960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.812130055694813e-05,
+      "loss": 0.7517,
+      "step": 6965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.811863622663752e-05,
+      "loss": 0.9541,
+      "step": 6970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.811597008228251e-05,
+      "loss": 0.6921,
+      "step": 6975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.811330212409233e-05,
+      "loss": 0.6888,
+      "step": 6980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.811063235227632e-05,
+      "loss": 0.8537,
+      "step": 6985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.810796076704396e-05,
+      "loss": 0.7204,
+      "step": 6990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.810528736860488e-05,
+      "loss": 0.6506,
+      "step": 6995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.810261215716887e-05,
+      "loss": 0.6871,
+      "step": 7000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8099935132945825e-05,
+      "loss": 0.7543,
+      "step": 7005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8097256296145816e-05,
+      "loss": 0.8341,
+      "step": 7010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8094575646979036e-05,
+      "loss": 0.8501,
+      "step": 7015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8091893185655826e-05,
+      "loss": 0.8359,
+      "step": 7020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.808920891238667e-05,
+      "loss": 0.8277,
+      "step": 7025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8086522827382204e-05,
+      "loss": 0.7674,
+      "step": 7030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.808383493085319e-05,
+      "loss": 0.7716,
+      "step": 7035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.808114522301054e-05,
+      "loss": 0.7889,
+      "step": 7040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8078453704065294e-05,
+      "loss": 0.788,
+      "step": 7045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8075760374228665e-05,
+      "loss": 0.7184,
+      "step": 7050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.807306523371198e-05,
+      "loss": 0.742,
+      "step": 7055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.807036828272672e-05,
+      "loss": 0.8931,
+      "step": 7060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.80676695214845e-05,
+      "loss": 0.835,
+      "step": 7065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8064968950197095e-05,
+      "loss": 0.8265,
+      "step": 7070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8062266569076405e-05,
+      "loss": 0.8526,
+      "step": 7075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8059562378334475e-05,
+      "loss": 0.7497,
+      "step": 7080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.805685637818349e-05,
+      "loss": 0.7826,
+      "step": 7085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8054148568835786e-05,
+      "loss": 0.842,
+      "step": 7090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8051438950503845e-05,
+      "loss": 0.8569,
+      "step": 7095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.804872752340027e-05,
+      "loss": 0.887,
+      "step": 7100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.804601428773782e-05,
+      "loss": 0.7942,
+      "step": 7105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.80432992437294e-05,
+      "loss": 0.6507,
+      "step": 7110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.804058239158804e-05,
+      "loss": 0.9072,
+      "step": 7115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.803786373152692e-05,
+      "loss": 0.6998,
+      "step": 7120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8035143263759386e-05,
+      "loss": 0.712,
+      "step": 7125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.803242098849889e-05,
+      "loss": 0.8511,
+      "step": 7130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.802969690595903e-05,
+      "loss": 0.8195,
+      "step": 7135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8026971016353584e-05,
+      "loss": 0.6801,
+      "step": 7140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.802424331989641e-05,
+      "loss": 0.8112,
+      "step": 7145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8021513816801565e-05,
+      "loss": 0.844,
+      "step": 7150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.801878250728322e-05,
+      "loss": 0.8558,
+      "step": 7155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.801604939155569e-05,
+      "loss": 0.7182,
+      "step": 7160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.801331446983342e-05,
+      "loss": 0.814,
+      "step": 7165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.801057774233104e-05,
+      "loss": 0.877,
+      "step": 7170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8007839209263276e-05,
+      "loss": 0.8308,
+      "step": 7175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8005098870845e-05,
+      "loss": 0.7684,
+      "step": 7180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.800235672729125e-05,
+      "loss": 0.8682,
+      "step": 7185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7999612778817196e-05,
+      "loss": 0.9712,
+      "step": 7190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.799686702563814e-05,
+      "loss": 0.788,
+      "step": 7195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.799411946796954e-05,
+      "loss": 0.8992,
+      "step": 7200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.799137010602698e-05,
+      "loss": 0.9255,
+      "step": 7205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.79886189400262e-05,
+      "loss": 0.6577,
+      "step": 7210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.798586597018306e-05,
+      "loss": 0.6142,
+      "step": 7215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7983111196713605e-05,
+      "loss": 0.7263,
+      "step": 7220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.798035461983397e-05,
+      "loss": 0.8851,
+      "step": 7225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.797759623976045e-05,
+      "loss": 0.8105,
+      "step": 7230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.797483605670951e-05,
+      "loss": 0.8346,
+      "step": 7235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.797207407089772e-05,
+      "loss": 0.7503,
+      "step": 7240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7969310282541794e-05,
+      "loss": 0.8372,
+      "step": 7245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7966544691858615e-05,
+      "loss": 0.8527,
+      "step": 7250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.796377729906518e-05,
+      "loss": 0.8144,
+      "step": 7255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.796100810437864e-05,
+      "loss": 0.8474,
+      "step": 7260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.795823710801628e-05,
+      "loss": 0.6952,
+      "step": 7265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.795546431019555e-05,
+      "loss": 0.9214,
+      "step": 7270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7952689711133994e-05,
+      "loss": 0.7239,
+      "step": 7275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7949913311049345e-05,
+      "loss": 0.9013,
+      "step": 7280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.794713511015945e-05,
+      "loss": 0.9118,
+      "step": 7285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.794435510868231e-05,
+      "loss": 0.7586,
+      "step": 7290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.794157330683606e-05,
+      "loss": 0.6804,
+      "step": 7295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7938789704838984e-05,
+      "loss": 0.8301,
+      "step": 7300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7936004302909486e-05,
+      "loss": 0.9335,
+      "step": 7305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.793321710126615e-05,
+      "loss": 0.8204,
+      "step": 7310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7930428100127664e-05,
+      "loss": 0.8582,
+      "step": 7315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.792763729971288e-05,
+      "loss": 0.7994,
+      "step": 7320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.792484470024078e-05,
+      "loss": 0.662,
+      "step": 7325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.792205030193049e-05,
+      "loss": 0.7738,
+      "step": 7330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.791925410500127e-05,
+      "loss": 0.8079,
+      "step": 7335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.791645610967254e-05,
+      "loss": 0.71,
+      "step": 7340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.791365631616384e-05,
+      "loss": 0.6819,
+      "step": 7345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7910854724694864e-05,
+      "loss": 0.7996,
+      "step": 7350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.790805133548545e-05,
+      "loss": 0.8071,
+      "step": 7355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.790524614875557e-05,
+      "loss": 0.8915,
+      "step": 7360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7902439164725334e-05,
+      "loss": 0.9415,
+      "step": 7365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.789963038361499e-05,
+      "loss": 0.854,
+      "step": 7370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.789681980564494e-05,
+      "loss": 0.716,
+      "step": 7375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7894007431035726e-05,
+      "loss": 0.7333,
+      "step": 7380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7891193260008026e-05,
+      "loss": 0.7083,
+      "step": 7385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.788837729278265e-05,
+      "loss": 0.7627,
+      "step": 7390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.788555952958056e-05,
+      "loss": 0.9727,
+      "step": 7395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.788273997062286e-05,
+      "loss": 0.9049,
+      "step": 7400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7879918616130795e-05,
+      "loss": 0.9316,
+      "step": 7405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7877095466325745e-05,
+      "loss": 0.8453,
+      "step": 7410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.787427052142923e-05,
+      "loss": 0.8353,
+      "step": 7415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.787144378166291e-05,
+      "loss": 0.8048,
+      "step": 7420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7868615247248604e-05,
+      "loss": 0.8047,
+      "step": 7425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.786578491840825e-05,
+      "loss": 0.7862,
+      "step": 7430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7862952795363925e-05,
+      "loss": 0.7956,
+      "step": 7435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.786011887833788e-05,
+      "loss": 0.7657,
+      "step": 7440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7857283167552455e-05,
+      "loss": 0.8208,
+      "step": 7445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.785444566323018e-05,
+      "loss": 0.9138,
+      "step": 7450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.78516063655937e-05,
+      "loss": 0.7994,
+      "step": 7455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.78487652748658e-05,
+      "loss": 0.7435,
+      "step": 7460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7845922391269423e-05,
+      "loss": 0.7738,
+      "step": 7465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7843077715027626e-05,
+      "loss": 0.7367,
+      "step": 7470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7840231246363634e-05,
+      "loss": 0.831,
+      "step": 7475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.783738298550079e-05,
+      "loss": 0.7333,
+      "step": 7480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.783453293266259e-05,
+      "loss": 0.8165,
+      "step": 7485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7831681088072676e-05,
+      "loss": 0.7677,
+      "step": 7490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.782882745195482e-05,
+      "loss": 0.72,
+      "step": 7495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.782597202453293e-05,
+      "loss": 0.6487,
+      "step": 7500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.782311480603107e-05,
+      "loss": 0.8733,
+      "step": 7505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.782025579667343e-05,
+      "loss": 0.7825,
+      "step": 7510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7817394996684354e-05,
+      "loss": 0.7005,
+      "step": 7515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.781453240628831e-05,
+      "loss": 0.7942,
+      "step": 7520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.781166802570994e-05,
+      "loss": 0.8163,
+      "step": 7525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7808801855173976e-05,
+      "loss": 0.9127,
+      "step": 7530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7805933894905324e-05,
+      "loss": 0.8118,
+      "step": 7535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.780306414512903e-05,
+      "loss": 0.7944,
+      "step": 7540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.780019260607027e-05,
+      "loss": 0.7226,
+      "step": 7545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7797319277954366e-05,
+      "loss": 0.8094,
+      "step": 7550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.779444416100677e-05,
+      "loss": 0.7522,
+      "step": 7555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7791567255453104e-05,
+      "loss": 0.8632,
+      "step": 7560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7788688561519085e-05,
+      "loss": 0.6781,
+      "step": 7565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.778580807943061e-05,
+      "loss": 0.8259,
+      "step": 7570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7782925809413695e-05,
+      "loss": 0.6863,
+      "step": 7575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.77800417516945e-05,
+      "loss": 0.8084,
+      "step": 7580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.777715590649935e-05,
+      "loss": 0.7288,
+      "step": 7585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.777426827405465e-05,
+      "loss": 0.7919,
+      "step": 7590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.777137885458701e-05,
+      "loss": 0.7569,
+      "step": 7595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7768487648323144e-05,
+      "loss": 0.7948,
+      "step": 7600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.776559465548992e-05,
+      "loss": 0.8371,
+      "step": 7605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.776269987631434e-05,
+      "loss": 0.7119,
+      "step": 7610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.775980331102356e-05,
+      "loss": 0.7255,
+      "step": 7615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7756904959844835e-05,
+      "loss": 0.7239,
+      "step": 7620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.775400482300561e-05,
+      "loss": 0.8782,
+      "step": 7625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.775110290073344e-05,
+      "loss": 0.9146,
+      "step": 7630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.774819919325605e-05,
+      "loss": 0.7668,
+      "step": 7635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.774529370080125e-05,
+      "loss": 0.9005,
+      "step": 7640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7742386423597064e-05,
+      "loss": 0.8589,
+      "step": 7645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.773947736187158e-05,
+      "loss": 0.7268,
+      "step": 7650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.773656651585309e-05,
+      "loss": 0.806,
+      "step": 7655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.773365388576998e-05,
+      "loss": 0.8754,
+      "step": 7660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.77307394718508e-05,
+      "loss": 0.6792,
+      "step": 7665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.772782327432425e-05,
+      "loss": 0.7556,
+      "step": 7670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.772490529341913e-05,
+      "loss": 0.8999,
+      "step": 7675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.772198552936442e-05,
+      "loss": 0.7984,
+      "step": 7680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.771906398238922e-05,
+      "loss": 0.6748,
+      "step": 7685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7716140652722774e-05,
+      "loss": 0.9188,
+      "step": 7690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.771321554059446e-05,
+      "loss": 0.7875,
+      "step": 7695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.771028864623382e-05,
+      "loss": 0.7906,
+      "step": 7700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.77073599698705e-05,
+      "loss": 0.8845,
+      "step": 7705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7704429511734314e-05,
+      "loss": 0.793,
+      "step": 7710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.770149727205521e-05,
+      "loss": 0.7802,
+      "step": 7715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.769856325106325e-05,
+      "loss": 0.9725,
+      "step": 7720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.769562744898868e-05,
+      "loss": 0.8408,
+      "step": 7725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.769268986606185e-05,
+      "loss": 0.8148,
+      "step": 7730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7689750502513255e-05,
+      "loss": 0.9142,
+      "step": 7735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.768680935857356e-05,
+      "loss": 0.8864,
+      "step": 7740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7683866434473536e-05,
+      "loss": 0.7944,
+      "step": 7745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.76809217304441e-05,
+      "loss": 0.8669,
+      "step": 7750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.767797524671632e-05,
+      "loss": 0.8018,
+      "step": 7755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7675026983521395e-05,
+      "loss": 0.7345,
+      "step": 7760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.767207694109066e-05,
+      "loss": 0.6435,
+      "step": 7765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7669125119655604e-05,
+      "loss": 0.7386,
+      "step": 7770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.766617151944784e-05,
+      "loss": 0.7342,
+      "step": 7775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7663216140699144e-05,
+      "loss": 0.8328,
+      "step": 7780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7660258983641384e-05,
+      "loss": 0.8689,
+      "step": 7785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7657300048506624e-05,
+      "loss": 0.9127,
+      "step": 7790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.765433933552703e-05,
+      "loss": 0.8793,
+      "step": 7795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.765137684493493e-05,
+      "loss": 0.8439,
+      "step": 7800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7648412576962776e-05,
+      "loss": 0.6499,
+      "step": 7805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.764544653184316e-05,
+      "loss": 0.7619,
+      "step": 7810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.764247870980882e-05,
+      "loss": 0.9041,
+      "step": 7815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.763950911109263e-05,
+      "loss": 0.7256,
+      "step": 7820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7636537735927613e-05,
+      "loss": 0.8597,
+      "step": 7825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7633564584546916e-05,
+      "loss": 0.8381,
+      "step": 7830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7630589657183835e-05,
+      "loss": 0.8253,
+      "step": 7835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.76276129540718e-05,
+      "loss": 0.723,
+      "step": 7840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.762463447544438e-05,
+      "loss": 0.8066,
+      "step": 7845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7621654221535293e-05,
+      "loss": 0.809,
+      "step": 7850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.76186721925784e-05,
+      "loss": 0.7983,
+      "step": 7855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7615688388807665e-05,
+      "loss": 0.7596,
+      "step": 7860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.761270281045724e-05,
+      "loss": 0.9377,
+      "step": 7865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7609715457761386e-05,
+      "loss": 0.7259,
+      "step": 7870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.760672633095451e-05,
+      "loss": 0.8541,
+      "step": 7875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.760373543027116e-05,
+      "loss": 0.9423,
+      "step": 7880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.760074275594602e-05,
+      "loss": 0.7212,
+      "step": 7885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.759774830821392e-05,
+      "loss": 0.8465,
+      "step": 7890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.759475208730982e-05,
+      "loss": 0.987,
+      "step": 7895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7591754093468834e-05,
+      "loss": 0.7307,
+      "step": 7900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7588754326926184e-05,
+      "loss": 0.8346,
+      "step": 7905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.758575278791727e-05,
+      "loss": 0.7288,
+      "step": 7910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7582749476677605e-05,
+      "loss": 0.7893,
+      "step": 7915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7579744393442855e-05,
+      "loss": 0.8403,
+      "step": 7920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.757673753844881e-05,
+      "loss": 0.7984,
+      "step": 7925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7573728911931424e-05,
+      "loss": 0.8746,
+      "step": 7930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7570718514126764e-05,
+      "loss": 0.8429,
+      "step": 7935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.756770634527104e-05,
+      "loss": 0.8346,
+      "step": 7940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7564692405600616e-05,
+      "loss": 0.8847,
+      "step": 7945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7561676695351976e-05,
+      "loss": 0.6616,
+      "step": 7950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.755865921476177e-05,
+      "loss": 0.804,
+      "step": 7955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7555639964066754e-05,
+      "loss": 0.7363,
+      "step": 7960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.755261894350385e-05,
+      "loss": 0.8548,
+      "step": 7965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.75495961533101e-05,
+      "loss": 0.8168,
+      "step": 7970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.75465715937227e-05,
+      "loss": 0.6345,
+      "step": 7975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.754354526497897e-05,
+      "loss": 0.9102,
+      "step": 7980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.754051716731638e-05,
+      "loss": 0.7096,
+      "step": 7985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7537487300972525e-05,
+      "loss": 0.7356,
+      "step": 7990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7534455666185166e-05,
+      "loss": 0.688,
+      "step": 7995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.753142226319218e-05,
+      "loss": 0.689,
+      "step": 8000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.752838709223158e-05,
+      "loss": 0.7881,
+      "step": 8005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7525350153541534e-05,
+      "loss": 0.9296,
+      "step": 8010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7522311447360343e-05,
+      "loss": 0.8627,
+      "step": 8015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7519270973926433e-05,
+      "loss": 0.793,
+      "step": 8020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.751622873347838e-05,
+      "loss": 0.7584,
+      "step": 8025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.751318472625492e-05,
+      "loss": 0.8536,
+      "step": 8030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.751013895249489e-05,
+      "loss": 0.8413,
+      "step": 8035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7507091412437276e-05,
+      "loss": 0.8454,
+      "step": 8040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.750404210632122e-05,
+      "loss": 0.7701,
+      "step": 8045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.750099103438599e-05,
+      "loss": 0.8004,
+      "step": 8050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7497938196871e-05,
+      "loss": 0.9889,
+      "step": 8055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.749488359401577e-05,
+      "loss": 0.7515,
+      "step": 8060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7491827226060014e-05,
+      "loss": 0.9324,
+      "step": 8065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7488769093243535e-05,
+      "loss": 0.8818,
+      "step": 8070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.748570919580631e-05,
+      "loss": 0.7374,
+      "step": 8075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7482647533988435e-05,
+      "loss": 0.9266,
+      "step": 8080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7479584108030136e-05,
+      "loss": 0.7565,
+      "step": 8085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.747651891817181e-05,
+      "loss": 0.8086,
+      "step": 8090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.747345196465396e-05,
+      "loss": 0.9391,
+      "step": 8095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7470383247717244e-05,
+      "loss": 0.6459,
+      "step": 8100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.746731276760245e-05,
+      "loss": 0.7412,
+      "step": 8105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.746424052455052e-05,
+      "loss": 0.7047,
+      "step": 8110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.74611665188025e-05,
+      "loss": 0.7226,
+      "step": 8115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.745809075059963e-05,
+      "loss": 0.8196,
+      "step": 8120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.745501322018322e-05,
+      "loss": 0.5435,
+      "step": 8125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7451933927794777e-05,
+      "loss": 0.8914,
+      "step": 8130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.744885287367592e-05,
+      "loss": 0.8608,
+      "step": 8135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.744577005806841e-05,
+      "loss": 0.859,
+      "step": 8140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.744268548121414e-05,
+      "loss": 0.8198,
+      "step": 8145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.743959914335515e-05,
+      "loss": 0.882,
+      "step": 8150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.743651104473361e-05,
+      "loss": 0.735,
+      "step": 8155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.743342118559184e-05,
+      "loss": 0.6242,
+      "step": 8160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7430329566172284e-05,
+      "loss": 0.7974,
+      "step": 8165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.742723618671754e-05,
+      "loss": 0.7568,
+      "step": 8170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7424141047470326e-05,
+      "loss": 0.8233,
+      "step": 8175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.742104414867352e-05,
+      "loss": 0.7301,
+      "step": 8180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.741794549057012e-05,
+      "loss": 0.7417,
+      "step": 8185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.741484507340326e-05,
+      "loss": 0.8588,
+      "step": 8190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.741174289741622e-05,
+      "loss": 0.633,
+      "step": 8195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.740863896285243e-05,
+      "loss": 0.7923,
+      "step": 8200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.740553326995544e-05,
+      "loss": 0.8328,
+      "step": 8205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.740242581896894e-05,
+      "loss": 0.6267,
+      "step": 8210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.739931661013676e-05,
+      "loss": 0.8507,
+      "step": 8215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.739620564370288e-05,
+      "loss": 0.8477,
+      "step": 8220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7393092919911394e-05,
+      "loss": 0.8198,
+      "step": 8225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.738997843900656e-05,
+      "loss": 0.7434,
+      "step": 8230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.738686220123276e-05,
+      "loss": 0.8423,
+      "step": 8235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.738374420683451e-05,
+      "loss": 0.7117,
+      "step": 8240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.738062445605646e-05,
+      "loss": 0.7681,
+      "step": 8245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7377502949143423e-05,
+      "loss": 0.7729,
+      "step": 8250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.737437968634033e-05,
+      "loss": 0.8329,
+      "step": 8255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.737125466789224e-05,
+      "loss": 0.8426,
+      "step": 8260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.736812789404438e-05,
+      "loss": 0.6936,
+      "step": 8265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.736499936504209e-05,
+      "loss": 0.763,
+      "step": 8270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.736186908113086e-05,
+      "loss": 0.9509,
+      "step": 8275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.73587370425563e-05,
+      "loss": 0.823,
+      "step": 8280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.735560324956419e-05,
+      "loss": 0.9261,
+      "step": 8285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.735246770240042e-05,
+      "loss": 0.7226,
+      "step": 8290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.734933040131102e-05,
+      "loss": 0.8563,
+      "step": 8295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.734619134654217e-05,
+      "loss": 0.8009,
+      "step": 8300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7343050538340186e-05,
+      "loss": 0.8793,
+      "step": 8305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.73399079769515e-05,
+      "loss": 0.8894,
+      "step": 8310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.733676366262272e-05,
+      "loss": 0.8194,
+      "step": 8315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.733361759560055e-05,
+      "loss": 0.8656,
+      "step": 8320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7330469776131884e-05,
+      "loss": 0.85,
+      "step": 8325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7327320204463675e-05,
+      "loss": 0.8959,
+      "step": 8330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.73241688808431e-05,
+      "loss": 0.8641,
+      "step": 8335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.732101580551741e-05,
+      "loss": 0.7719,
+      "step": 8340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.731786097873403e-05,
+      "loss": 0.8282,
+      "step": 8345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.73147044007405e-05,
+      "loss": 0.8455,
+      "step": 8350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7311546071784506e-05,
+      "loss": 0.7587,
+      "step": 8355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.730838599211388e-05,
+      "loss": 0.7136,
+      "step": 8360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.730522416197657e-05,
+      "loss": 0.79,
+      "step": 8365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.730206058162069e-05,
+      "loss": 0.8731,
+      "step": 8370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7298895251294464e-05,
+      "loss": 0.817,
+      "step": 8375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.729572817124627e-05,
+      "loss": 0.8334,
+      "step": 8380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.729255934172462e-05,
+      "loss": 0.7309,
+      "step": 8385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.728938876297816e-05,
+      "loss": 0.7559,
+      "step": 8390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7286216435255685e-05,
+      "loss": 0.7146,
+      "step": 8395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.728304235880609e-05,
+      "loss": 0.8805,
+      "step": 8400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.727986653387846e-05,
+      "loss": 0.9933,
+      "step": 8405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.727668896072199e-05,
+      "loss": 0.6786,
+      "step": 8410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7273509639586e-05,
+      "loss": 0.8467,
+      "step": 8415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.727032857071997e-05,
+      "loss": 0.8853,
+      "step": 8420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.72671457543735e-05,
+      "loss": 0.8622,
+      "step": 8425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7263961190796353e-05,
+      "loss": 0.6195,
+      "step": 8430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7260774880238396e-05,
+      "loss": 0.8503,
+      "step": 8435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7257586822949654e-05,
+      "loss": 0.7259,
+      "step": 8440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.725439701918028e-05,
+      "loss": 0.737,
+      "step": 8445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.725120546918057e-05,
+      "loss": 0.8469,
+      "step": 8450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.724801217320095e-05,
+      "loss": 0.7303,
+      "step": 8455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7244817131492004e-05,
+      "loss": 0.7396,
+      "step": 8460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.724162034430442e-05,
+      "loss": 0.946,
+      "step": 8465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.723842181188903e-05,
+      "loss": 0.7625,
+      "step": 8470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.723522153449684e-05,
+      "loss": 0.8024,
+      "step": 8475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7232019512378954e-05,
+      "loss": 0.739,
+      "step": 8480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7228815745786616e-05,
+      "loss": 0.6964,
+      "step": 8485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.722561023497123e-05,
+      "loss": 0.8487,
+      "step": 8490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.72224029801843e-05,
+      "loss": 0.8217,
+      "step": 8495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.721919398167751e-05,
+      "loss": 0.8176,
+      "step": 8500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.721598323970265e-05,
+      "loss": 0.7314,
+      "step": 8505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.721277075451166e-05,
+      "loss": 0.9069,
+      "step": 8510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7209556526356616e-05,
+      "loss": 0.8434,
+      "step": 8515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7206340555489714e-05,
+      "loss": 0.6144,
+      "step": 8520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7203122842163316e-05,
+      "loss": 0.8412,
+      "step": 8525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.719990338662989e-05,
+      "loss": 0.6959,
+      "step": 8530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7196682189142083e-05,
+      "loss": 0.6124,
+      "step": 8535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.719345924995263e-05,
+      "loss": 0.8663,
+      "step": 8540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7190234569314426e-05,
+      "loss": 0.8179,
+      "step": 8545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.718700814748051e-05,
+      "loss": 0.7653,
+      "step": 8550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7183779984704034e-05,
+      "loss": 0.8646,
+      "step": 8555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.718055008123832e-05,
+      "loss": 0.7132,
+      "step": 8560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.71773184373368e-05,
+      "loss": 0.8337,
+      "step": 8565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.717408505325305e-05,
+      "loss": 0.6969,
+      "step": 8570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.717084992924078e-05,
+      "loss": 0.7786,
+      "step": 8575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.716761306555384e-05,
+      "loss": 0.863,
+      "step": 8580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7164374462446224e-05,
+      "loss": 0.8722,
+      "step": 8585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7161134120172045e-05,
+      "loss": 0.8318,
+      "step": 8590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7157892038985575e-05,
+      "loss": 0.7558,
+      "step": 8595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7154648219141195e-05,
+      "loss": 0.7415,
+      "step": 8600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.715140266089345e-05,
+      "loss": 0.645,
+      "step": 8605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7148155364496994e-05,
+      "loss": 0.8102,
+      "step": 8610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.714490633020664e-05,
+      "loss": 0.8066,
+      "step": 8615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7141655558277335e-05,
+      "loss": 0.8735,
+      "step": 8620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.713840304896415e-05,
+      "loss": 0.7703,
+      "step": 8625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7135148802522294e-05,
+      "loss": 0.7764,
+      "step": 8630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.713189281920712e-05,
+      "loss": 0.771,
+      "step": 8635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.712863509927413e-05,
+      "loss": 0.6983,
+      "step": 8640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.712537564297892e-05,
+      "loss": 0.6602,
+      "step": 8645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.712211445057727e-05,
+      "loss": 0.855,
+      "step": 8650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.711885152232507e-05,
+      "loss": 0.8558,
+      "step": 8655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.711558685847834e-05,
+      "loss": 0.838,
+      "step": 8660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.711232045929327e-05,
+      "loss": 0.7582,
+      "step": 8665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.710905232502614e-05,
+      "loss": 0.7527,
+      "step": 8670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.71057824559334e-05,
+      "loss": 0.7696,
+      "step": 8675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.710251085227163e-05,
+      "loss": 0.7761,
+      "step": 8680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.709923751429755e-05,
+      "loss": 0.8355,
+      "step": 8685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7095962442267974e-05,
+      "loss": 0.7672,
+      "step": 8690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7092685636439926e-05,
+      "loss": 0.8688,
+      "step": 8695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.708940709707051e-05,
+      "loss": 0.6361,
+      "step": 8700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7086126824416965e-05,
+      "loss": 0.7768,
+      "step": 8705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.708284481873672e-05,
+      "loss": 0.7763,
+      "step": 8710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7079561080287274e-05,
+      "loss": 0.7752,
+      "step": 8715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.707627560932629e-05,
+      "loss": 0.8171,
+      "step": 8720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7072988406111595e-05,
+      "loss": 0.8556,
+      "step": 8725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.70696994709011e-05,
+      "loss": 0.6163,
+      "step": 8730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.706640880395289e-05,
+      "loss": 0.8174,
+      "step": 8735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.706311640552517e-05,
+      "loss": 0.7525,
+      "step": 8740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7059822275876286e-05,
+      "loss": 0.7923,
+      "step": 8745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.705652641526471e-05,
+      "loss": 0.7434,
+      "step": 8750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.705322882394906e-05,
+      "loss": 0.802,
+      "step": 8755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.704992950218809e-05,
+      "loss": 0.8143,
+      "step": 8760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7046628450240696e-05,
+      "loss": 0.8047,
+      "step": 8765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.704332566836588e-05,
+      "loss": 0.6798,
+      "step": 8770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.704002115682281e-05,
+      "loss": 0.6715,
+      "step": 8775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7036714915870795e-05,
+      "loss": 0.6971,
+      "step": 8780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7033406945769246e-05,
+      "loss": 0.8731,
+      "step": 8785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.703009724677773e-05,
+      "loss": 0.6715,
+      "step": 8790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.702678581915596e-05,
+      "loss": 0.8188,
+      "step": 8795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.702347266316376e-05,
+      "loss": 0.6578,
+      "step": 8800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.702015777906111e-05,
+      "loss": 0.7778,
+      "step": 8805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.701684116710813e-05,
+      "loss": 0.7524,
+      "step": 8810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.701352282756504e-05,
+      "loss": 0.9254,
+      "step": 8815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.701020276069222e-05,
+      "loss": 0.8136,
+      "step": 8820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.700688096675021e-05,
+      "loss": 0.8761,
+      "step": 8825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.700355744599965e-05,
+      "loss": 0.7901,
+      "step": 8830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.700023219870131e-05,
+      "loss": 0.8181,
+      "step": 8835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.699690522511612e-05,
+      "loss": 0.7215,
+      "step": 8840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.699357652550515e-05,
+      "loss": 0.6972,
+      "step": 8845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.6990246100129576e-05,
+      "loss": 0.6351,
+      "step": 8850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.698691394925073e-05,
+      "loss": 0.7755,
+      "step": 8855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.698358007313008e-05,
+      "loss": 0.7657,
+      "step": 8860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.698024447202922e-05,
+      "loss": 0.7957,
+      "step": 8865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.697690714620988e-05,
+      "loss": 0.8218,
+      "step": 8870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.697356809593394e-05,
+      "loss": 0.8059,
+      "step": 8875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6970227321463396e-05,
+      "loss": 0.8861,
+      "step": 8880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.696688482306039e-05,
+      "loss": 0.6951,
+      "step": 8885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.69635406009872e-05,
+      "loss": 0.7081,
+      "step": 8890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.696019465550623e-05,
+      "loss": 0.7778,
+      "step": 8895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6956846986880035e-05,
+      "loss": 0.854,
+      "step": 8900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6953497595371275e-05,
+      "loss": 0.7726,
+      "step": 8905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6950146481242794e-05,
+      "loss": 0.7738,
+      "step": 8910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6946793644757526e-05,
+      "loss": 0.7351,
+      "step": 8915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.694343908617855e-05,
+      "loss": 0.7159,
+      "step": 8920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.694008280576911e-05,
+      "loss": 0.8371,
+      "step": 8925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.693672480379255e-05,
+      "loss": 0.7943,
+      "step": 8930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.693336508051236e-05,
+      "loss": 0.7223,
+      "step": 8935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.693000363619218e-05,
+      "loss": 0.6927,
+      "step": 8940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.692664047109574e-05,
+      "loss": 0.824,
+      "step": 8945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.692327558548697e-05,
+      "loss": 0.8323,
+      "step": 8950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6919908979629877e-05,
+      "loss": 0.7352,
+      "step": 8955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.691654065378865e-05,
+      "loss": 0.7701,
+      "step": 8960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6913170608227574e-05,
+      "loss": 0.8181,
+      "step": 8965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.69097988432111e-05,
+      "loss": 0.774,
+      "step": 8970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.690642535900378e-05,
+      "loss": 0.7599,
+      "step": 8975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.690305015587033e-05,
+      "loss": 0.7864,
+      "step": 8980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6899673234075604e-05,
+      "loss": 0.9758,
+      "step": 8985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.689629459388456e-05,
+      "loss": 0.7617,
+      "step": 8990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6892914235562313e-05,
+      "loss": 0.7715,
+      "step": 8995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6889532159374114e-05,
+      "loss": 0.8653,
+      "step": 9000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.688614836558534e-05,
+      "loss": 0.7397,
+      "step": 9005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.68827628544615e-05,
+      "loss": 0.8779,
+      "step": 9010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6879375626268265e-05,
+      "loss": 0.6842,
+      "step": 9015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6875986681271396e-05,
+      "loss": 0.7844,
+      "step": 9020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.687259601973683e-05,
+      "loss": 0.8776,
+      "step": 9025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6869203641930604e-05,
+      "loss": 0.8472,
+      "step": 9030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.686580954811892e-05,
+      "loss": 0.7991,
+      "step": 9035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.68624137385681e-05,
+      "loss": 0.8891,
+      "step": 9040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.68590162135446e-05,
+      "loss": 0.8141,
+      "step": 9045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6855616973315005e-05,
+      "loss": 0.7931,
+      "step": 9050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6852216018146064e-05,
+      "loss": 0.8202,
+      "step": 9055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.684881334830462e-05,
+      "loss": 0.6925,
+      "step": 9060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.684540896405767e-05,
+      "loss": 0.7158,
+      "step": 9065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.684200286567235e-05,
+      "loss": 0.7355,
+      "step": 9070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6838595053415926e-05,
+      "loss": 0.7648,
+      "step": 9075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.68351855275558e-05,
+      "loss": 0.821,
+      "step": 9080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.683177428835951e-05,
+      "loss": 0.8214,
+      "step": 9085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6828361336094705e-05,
+      "loss": 0.6237,
+      "step": 9090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.682494667102921e-05,
+      "loss": 0.9444,
+      "step": 9095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.682153029343095e-05,
+      "loss": 0.7413,
+      "step": 9100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6818112203568e-05,
+      "loss": 0.8892,
+      "step": 9105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.681469240170857e-05,
+      "loss": 0.8794,
+      "step": 9110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6811270888121006e-05,
+      "loss": 0.756,
+      "step": 9115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.680784766307377e-05,
+      "loss": 0.7525,
+      "step": 9120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.680442272683547e-05,
+      "loss": 0.9392,
+      "step": 9125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.680099607967487e-05,
+      "loss": 0.7322,
+      "step": 9130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.679756772186083e-05,
+      "loss": 0.8301,
+      "step": 9135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.679413765366236e-05,
+      "loss": 0.724,
+      "step": 9140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6790705875348614e-05,
+      "loss": 0.8354,
+      "step": 9145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.678727238718888e-05,
+      "loss": 0.8303,
+      "step": 9150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6783837189452565e-05,
+      "loss": 0.9175,
+      "step": 9155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.678040028240921e-05,
+      "loss": 0.8232,
+      "step": 9160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.67769616663285e-05,
+      "loss": 0.8086,
+      "step": 9165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.677352134148026e-05,
+      "loss": 0.9872,
+      "step": 9170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.677007930813445e-05,
+      "loss": 0.6335,
+      "step": 9175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.676663556656114e-05,
+      "loss": 0.8807,
+      "step": 9180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6763190117030543e-05,
+      "loss": 0.7734,
+      "step": 9185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6759742959813027e-05,
+      "loss": 0.8999,
+      "step": 9190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.675629409517907e-05,
+      "loss": 0.8552,
+      "step": 9195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6752843523399305e-05,
+      "loss": 0.6333,
+      "step": 9200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.674939124474447e-05,
+      "loss": 0.7171,
+      "step": 9205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.674593725948547e-05,
+      "loss": 0.8502,
+      "step": 9210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6742481567893324e-05,
+      "loss": 0.7937,
+      "step": 9215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6739024170239175e-05,
+      "loss": 0.7489,
+      "step": 9220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.673556506679434e-05,
+      "loss": 0.8656,
+      "step": 9225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6732104257830226e-05,
+      "loss": 0.9049,
+      "step": 9230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.672864174361839e-05,
+      "loss": 0.8415,
+      "step": 9235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6725177524430524e-05,
+      "loss": 0.6675,
+      "step": 9240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6721711600538466e-05,
+      "loss": 0.8559,
+      "step": 9245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.671824397221416e-05,
+      "loss": 0.6655,
+      "step": 9250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.671477463972972e-05,
+      "loss": 0.8897,
+      "step": 9255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.671130360335735e-05,
+      "loss": 0.6493,
+      "step": 9260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.670783086336943e-05,
+      "loss": 0.8354,
+      "step": 9265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.670435642003844e-05,
+      "loss": 0.8187,
+      "step": 9270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.670088027363703e-05,
+      "loss": 0.8215,
+      "step": 9275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6697402424437934e-05,
+      "loss": 0.7948,
+      "step": 9280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6693922872714055e-05,
+      "loss": 0.8136,
+      "step": 9285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.669044161873845e-05,
+      "loss": 0.7081,
+      "step": 9290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.668695866278424e-05,
+      "loss": 0.7798,
+      "step": 9295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.668347400512474e-05,
+      "loss": 0.8168,
+      "step": 9300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.667998764603339e-05,
+      "loss": 0.843,
+      "step": 9305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.667649958578374e-05,
+      "loss": 0.7505,
+      "step": 9310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6673009824649495e-05,
+      "loss": 0.7307,
+      "step": 9315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.666951836290448e-05,
+      "loss": 0.867,
+      "step": 9320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.666602520082265e-05,
+      "loss": 0.8448,
+      "step": 9325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6662530338678126e-05,
+      "loss": 0.8344,
+      "step": 9330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.665903377674511e-05,
+      "loss": 0.9449,
+      "step": 9335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6655535515297985e-05,
+      "loss": 0.6916,
+      "step": 9340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6652035554611243e-05,
+      "loss": 0.866,
+      "step": 9345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.664853389495952e-05,
+      "loss": 0.9566,
+      "step": 9350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6645030536617565e-05,
+      "loss": 0.7587,
+      "step": 9355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.664152547986029e-05,
+      "loss": 0.8226,
+      "step": 9360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.663801872496273e-05,
+      "loss": 0.8038,
+      "step": 9365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6634510272200024e-05,
+      "loss": 0.8376,
+      "step": 9370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.663100012184749e-05,
+      "loss": 0.7771,
+      "step": 9375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6627488274180554e-05,
+      "loss": 0.769,
+      "step": 9380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.662397472947477e-05,
+      "loss": 0.846,
+      "step": 9385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.662045948800585e-05,
+      "loss": 0.7619,
+      "step": 9390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.661694255004961e-05,
+      "loss": 0.7534,
+      "step": 9395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6613423915882014e-05,
+      "loss": 0.805,
+      "step": 9400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.660990358577917e-05,
+      "loss": 0.9179,
+      "step": 9405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.66063815600173e-05,
+      "loss": 0.711,
+      "step": 9410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.660285783887276e-05,
+      "loss": 0.9107,
+      "step": 9415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.659933242262204e-05,
+      "loss": 0.666,
+      "step": 9420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6595805311541793e-05,
+      "loss": 0.7529,
+      "step": 9425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.659227650590876e-05,
+      "loss": 0.8221,
+      "step": 9430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.658874600599984e-05,
+      "loss": 0.7809,
+      "step": 9435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.658521381209206e-05,
+      "loss": 0.749,
+      "step": 9440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.658167992446257e-05,
+      "loss": 0.636,
+      "step": 9445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6578144343388676e-05,
+      "loss": 0.7896,
+      "step": 9450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6574607069147804e-05,
+      "loss": 0.884,
+      "step": 9455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.65710681020175e-05,
+      "loss": 0.9028,
+      "step": 9460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.656752744227547e-05,
+      "loss": 0.7473,
+      "step": 9465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.656398509019954e-05,
+      "loss": 0.6627,
+      "step": 9470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6560441046067636e-05,
+      "loss": 0.7747,
+      "step": 9475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6556895310157886e-05,
+      "loss": 0.7177,
+      "step": 9480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.655334788274849e-05,
+      "loss": 0.5813,
+      "step": 9485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6549798764117814e-05,
+      "loss": 0.7672,
+      "step": 9490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.654624795454433e-05,
+      "loss": 0.8411,
+      "step": 9495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.654269545430668e-05,
+      "loss": 0.802,
+      "step": 9500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.653914126368361e-05,
+      "loss": 0.8177,
+      "step": 9505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.653558538295399e-05,
+      "loss": 0.9998,
+      "step": 9510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6532027812396864e-05,
+      "loss": 0.765,
+      "step": 9515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.652846855229136e-05,
+      "loss": 0.8213,
+      "step": 9520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6524907602916776e-05,
+      "loss": 0.7947,
+      "step": 9525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.652134496455251e-05,
+      "loss": 0.8174,
+      "step": 9530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.651778063747815e-05,
+      "loss": 0.7767,
+      "step": 9535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.651421462197333e-05,
+      "loss": 0.8102,
+      "step": 9540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.65106469183179e-05,
+      "loss": 0.7869,
+      "step": 9545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.650707752679178e-05,
+      "loss": 0.8698,
+      "step": 9550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.650350644767507e-05,
+      "loss": 0.7939,
+      "step": 9555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.649993368124797e-05,
+      "loss": 0.8774,
+      "step": 9560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.649635922779081e-05,
+      "loss": 0.9838,
+      "step": 9565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.649278308758409e-05,
+      "loss": 0.8833,
+      "step": 9570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.648920526090841e-05,
+      "loss": 0.763,
+      "step": 9575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6485625748044506e-05,
+      "loss": 0.7466,
+      "step": 9580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.648204454927325e-05,
+      "loss": 0.759,
+      "step": 9585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.647846166487566e-05,
+      "loss": 0.6762,
+      "step": 9590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6474877095132854e-05,
+      "loss": 0.7819,
+      "step": 9595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.647129084032612e-05,
+      "loss": 0.7402,
+      "step": 9600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.646770290073684e-05,
+      "loss": 0.8402,
+      "step": 9605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.646411327664657e-05,
+      "loss": 0.913,
+      "step": 9610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.646052196833696e-05,
+      "loss": 0.8166,
+      "step": 9615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.645692897608982e-05,
+      "loss": 0.8481,
+      "step": 9620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.645333430018707e-05,
+      "loss": 0.7684,
+      "step": 9625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6449737940910766e-05,
+      "loss": 0.6542,
+      "step": 9630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6446139898543124e-05,
+      "loss": 0.9037,
+      "step": 9635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.644254017336645e-05,
+      "loss": 0.7863,
+      "step": 9640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.643893876566323e-05,
+      "loss": 0.7283,
+      "step": 9645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.643533567571603e-05,
+      "loss": 0.8001,
+      "step": 9650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6431730903807574e-05,
+      "loss": 0.9337,
+      "step": 9655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.642812445022073e-05,
+      "loss": 0.7278,
+      "step": 9660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6424516315238476e-05,
+      "loss": 0.771,
+      "step": 9665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.642090649914393e-05,
+      "loss": 0.7458,
+      "step": 9670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.641729500222034e-05,
+      "loss": 0.8958,
+      "step": 9675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.64136818247511e-05,
+      "loss": 0.7685,
+      "step": 9680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6410066967019716e-05,
+      "loss": 0.9039,
+      "step": 9685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.640645042930985e-05,
+      "loss": 0.7324,
+      "step": 9690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6402832211905244e-05,
+      "loss": 0.9173,
+      "step": 9695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.639921231508983e-05,
+      "loss": 0.8128,
+      "step": 9700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.639559073914767e-05,
+      "loss": 0.7656,
+      "step": 9705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.63919674843629e-05,
+      "loss": 0.6259,
+      "step": 9710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.638834255101985e-05,
+      "loss": 0.7749,
+      "step": 9715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.638471593940293e-05,
+      "loss": 0.8371,
+      "step": 9720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.638108764979675e-05,
+      "loss": 0.8643,
+      "step": 9725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.637745768248597e-05,
+      "loss": 0.8237,
+      "step": 9730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6373826037755454e-05,
+      "loss": 0.8107,
+      "step": 9735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.637019271589014e-05,
+      "loss": 0.8004,
+      "step": 9740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.636655771717513e-05,
+      "loss": 0.7253,
+      "step": 9745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.636292104189566e-05,
+      "loss": 0.8304,
+      "step": 9750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.635928269033708e-05,
+      "loss": 0.8822,
+      "step": 9755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.635564266278488e-05,
+      "loss": 0.8726,
+      "step": 9760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.635200095952468e-05,
+      "loss": 0.8136,
+      "step": 9765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6348357580842246e-05,
+      "loss": 0.8866,
+      "step": 9770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6344712527023435e-05,
+      "loss": 0.804,
+      "step": 9775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.634106579835429e-05,
+      "loss": 0.671,
+      "step": 9780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.633741739512095e-05,
+      "loss": 0.8072,
+      "step": 9785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.633376731760968e-05,
+      "loss": 0.6597,
+      "step": 9790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.63301155661069e-05,
+      "loss": 0.7634,
+      "step": 9795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6326462140899154e-05,
+      "loss": 0.7513,
+      "step": 9800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.632280704227311e-05,
+      "loss": 0.8291,
+      "step": 9805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6319150270515576e-05,
+      "loss": 0.7837,
+      "step": 9810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6315491825913484e-05,
+      "loss": 0.8024,
+      "step": 9815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.63118317087539e-05,
+      "loss": 0.8076,
+      "step": 9820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.630816991932402e-05,
+      "loss": 0.6843,
+      "step": 9825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.630450645791118e-05,
+      "loss": 0.7388,
+      "step": 9830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.630084132480282e-05,
+      "loss": 0.7083,
+      "step": 9835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.629717452028656e-05,
+      "loss": 0.8184,
+      "step": 9840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6293506044650105e-05,
+      "loss": 0.8329,
+      "step": 9845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.628983589818131e-05,
+      "loss": 0.7568,
+      "step": 9850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.628616408116816e-05,
+      "loss": 0.9127,
+      "step": 9855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.628249059389877e-05,
+      "loss": 0.82,
+      "step": 9860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.627881543666139e-05,
+      "loss": 0.7945,
+      "step": 9865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6275138609744396e-05,
+      "loss": 0.7832,
+      "step": 9870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6271460113436295e-05,
+      "loss": 0.8366,
+      "step": 9875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.626777994802572e-05,
+      "loss": 0.6796,
+      "step": 9880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.626409811380147e-05,
+      "loss": 0.8064,
+      "step": 9885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.626041461105241e-05,
+      "loss": 0.6502,
+      "step": 9890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.625672944006759e-05,
+      "loss": 0.7781,
+      "step": 9895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.625304260113617e-05,
+      "loss": 0.6633,
+      "step": 9900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6249354094547456e-05,
+      "loss": 0.9537,
+      "step": 9905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6245663920590856e-05,
+      "loss": 0.7638,
+      "step": 9910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.624197207955594e-05,
+      "loss": 0.6981,
+      "step": 9915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6238278571732386e-05,
+      "loss": 0.7026,
+      "step": 9920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6234583397410017e-05,
+      "loss": 0.6439,
+      "step": 9925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6230886556878775e-05,
+      "loss": 0.8046,
+      "step": 9930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6227188050428734e-05,
+      "loss": 0.6273,
+      "step": 9935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.622348787835012e-05,
+      "loss": 0.7367,
+      "step": 9940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.621978604093326e-05,
+      "loss": 0.7979,
+      "step": 9945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.621608253846864e-05,
+      "loss": 0.7118,
+      "step": 9950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6212377371246845e-05,
+      "loss": 0.8271,
+      "step": 9955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.620867053955862e-05,
+      "loss": 0.8968,
+      "step": 9960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.620496204369482e-05,
+      "loss": 0.779,
+      "step": 9965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.620125188394644e-05,
+      "loss": 0.8184,
+      "step": 9970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.619754006060461e-05,
+      "loss": 0.8621,
+      "step": 9975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.619382657396057e-05,
+      "loss": 0.8337,
+      "step": 9980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.619011142430574e-05,
+      "loss": 0.8226,
+      "step": 9985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.618639461193159e-05,
+      "loss": 0.7623,
+      "step": 9990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.61826761371298e-05,
+      "loss": 0.7875,
+      "step": 9995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6178956000192127e-05,
+      "loss": 0.7813,
+      "step": 10000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.617523420141049e-05,
+      "loss": 0.7784,
+      "step": 10005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6171510741076916e-05,
+      "loss": 0.7646,
+      "step": 10010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6167785619483585e-05,
+      "loss": 0.7866,
+      "step": 10015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.616405883692279e-05,
+      "loss": 0.7771,
+      "step": 10020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.616033039368695e-05,
+      "loss": 0.7228,
+      "step": 10025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.615660029006864e-05,
+      "loss": 0.8959,
+      "step": 10030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.615286852636054e-05,
+      "loss": 0.9012,
+      "step": 10035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6149135102855466e-05,
+      "loss": 0.7588,
+      "step": 10040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614540001984638e-05,
+      "loss": 0.592,
+      "step": 10045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614166327762635e-05,
+      "loss": 0.7225,
+      "step": 10050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6137924876488596e-05,
+      "loss": 0.7672,
+      "step": 10055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6134184816726446e-05,
+      "loss": 0.8455,
+      "step": 10060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.613044309863338e-05,
+      "loss": 0.8741,
+      "step": 10065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6126699722502994e-05,
+      "loss": 0.894,
+      "step": 10070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.612295468862903e-05,
+      "loss": 0.7773,
+      "step": 10075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.611920799730532e-05,
+      "loss": 0.7497,
+      "step": 10080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.611545964882589e-05,
+      "loss": 0.7893,
+      "step": 10085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6111709643484844e-05,
+      "loss": 0.821,
+      "step": 10090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.610795798157642e-05,
+      "loss": 0.9373,
+      "step": 10095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.610420466339502e-05,
+      "loss": 0.8326,
+      "step": 10100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6100449689235145e-05,
+      "loss": 0.7538,
+      "step": 10105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6096693059391446e-05,
+      "loss": 0.656,
+      "step": 10110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609293477415868e-05,
+      "loss": 0.7189,
+      "step": 10115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6089174833831744e-05,
+      "loss": 0.9002,
+      "step": 10120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.608541323870568e-05,
+      "loss": 0.6934,
+      "step": 10125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6081649989075646e-05,
+      "loss": 0.7381,
+      "step": 10130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.607788508523693e-05,
+      "loss": 0.7228,
+      "step": 10135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.607411852748495e-05,
+      "loss": 0.7072,
+      "step": 10140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6070350316115266e-05,
+      "loss": 0.5865,
+      "step": 10145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.606658045142355e-05,
+      "loss": 0.8025,
+      "step": 10150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6062808933705604e-05,
+      "loss": 0.8481,
+      "step": 10155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.605903576325737e-05,
+      "loss": 0.7504,
+      "step": 10160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6055260940374924e-05,
+      "loss": 0.8906,
+      "step": 10165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.605148446535446e-05,
+      "loss": 0.7604,
+      "step": 10170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.60477063384923e-05,
+      "loss": 0.8846,
+      "step": 10175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6043926560084916e-05,
+      "loss": 0.7143,
+      "step": 10180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.604014513042888e-05,
+      "loss": 0.7782,
+      "step": 10185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6036362049820914e-05,
+      "loss": 0.7974,
+      "step": 10190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.603257731855787e-05,
+      "loss": 0.8214,
+      "step": 10195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.602879093693671e-05,
+      "loss": 0.7688,
+      "step": 10200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.602500290525455e-05,
+      "loss": 0.7307,
+      "step": 10205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6021213223808624e-05,
+      "loss": 0.7982,
+      "step": 10210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6017421892896287e-05,
+      "loss": 0.9078,
+      "step": 10215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.601362891281504e-05,
+      "loss": 0.9685,
+      "step": 10220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.600983428386251e-05,
+      "loss": 0.7428,
+      "step": 10225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.600603800633645e-05,
+      "loss": 0.8188,
+      "step": 10230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.600224008053472e-05,
+      "loss": 0.6804,
+      "step": 10235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5998440506755356e-05,
+      "loss": 0.8504,
+      "step": 10240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.599463928529649e-05,
+      "loss": 0.9281,
+      "step": 10245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5990836416456385e-05,
+      "loss": 0.7945,
+      "step": 10250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.598703190053344e-05,
+      "loss": 0.6668,
+      "step": 10255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.59832257378262e-05,
+      "loss": 0.8056,
+      "step": 10260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5979417928633305e-05,
+      "loss": 0.7964,
+      "step": 10265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.597560847325355e-05,
+      "loss": 1.035,
+      "step": 10270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.597179737198584e-05,
+      "loss": 0.8906,
+      "step": 10275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5967984625129235e-05,
+      "loss": 0.7478,
+      "step": 10280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.59641702329829e-05,
+      "loss": 0.8341,
+      "step": 10285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5960354195846136e-05,
+      "loss": 0.7146,
+      "step": 10290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.595653651401838e-05,
+      "loss": 0.7813,
+      "step": 10295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.595271718779919e-05,
+      "loss": 0.7597,
+      "step": 10300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.594889621748825e-05,
+      "loss": 0.8173,
+      "step": 10305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.594507360338539e-05,
+      "loss": 0.6584,
+      "step": 10310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.594124934579056e-05,
+      "loss": 0.7892,
+      "step": 10315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.593742344500384e-05,
+      "loss": 0.7498,
+      "step": 10320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.593359590132541e-05,
+      "loss": 0.7586,
+      "step": 10325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.592976671505563e-05,
+      "loss": 0.7721,
+      "step": 10330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5925935886494955e-05,
+      "loss": 0.7091,
+      "step": 10335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5922103415943986e-05,
+      "loss": 0.8232,
+      "step": 10340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5918269303703425e-05,
+      "loss": 0.7667,
+      "step": 10345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5914433550074145e-05,
+      "loss": 0.8969,
+      "step": 10350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.591059615535711e-05,
+      "loss": 0.8925,
+      "step": 10355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5906757119853435e-05,
+      "loss": 0.7632,
+      "step": 10360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5902916443864354e-05,
+      "loss": 0.8219,
+      "step": 10365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.589907412769123e-05,
+      "loss": 0.9253,
+      "step": 10370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.589523017163557e-05,
+      "loss": 0.7863,
+      "step": 10375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5891384575998974e-05,
+      "loss": 0.8144,
+      "step": 10380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.588753734108321e-05,
+      "loss": 0.6786,
+      "step": 10385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.588368846719016e-05,
+      "loss": 0.7238,
+      "step": 10390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.587983795462183e-05,
+      "loss": 0.7105,
+      "step": 10395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.587598580368034e-05,
+      "loss": 0.8256,
+      "step": 10400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.587213201466798e-05,
+      "loss": 0.738,
+      "step": 10405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.586827658788714e-05,
+      "loss": 0.6176,
+      "step": 10410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.586441952364032e-05,
+      "loss": 0.759,
+      "step": 10415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5860560822230206e-05,
+      "loss": 0.795,
+      "step": 10420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.585670048395955e-05,
+      "loss": 0.7823,
+      "step": 10425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.585283850913128e-05,
+      "loss": 0.8893,
+      "step": 10430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.584897489804841e-05,
+      "loss": 0.7336,
+      "step": 10435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.584510965101413e-05,
+      "loss": 0.7736,
+      "step": 10440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5841242768331713e-05,
+      "loss": 0.8292,
+      "step": 10445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.583737425030459e-05,
+      "loss": 0.8093,
+      "step": 10450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.583350409723631e-05,
+      "loss": 0.8556,
+      "step": 10455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.582963230943056e-05,
+      "loss": 0.7383,
+      "step": 10460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.582575888719113e-05,
+      "loss": 0.8949,
+      "step": 10465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5821883830821966e-05,
+      "loss": 0.8084,
+      "step": 10470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.581800714062713e-05,
+      "loss": 0.7045,
+      "step": 10475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5814128816910805e-05,
+      "loss": 0.7557,
+      "step": 10480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.581024885997732e-05,
+      "loss": 0.859,
+      "step": 10485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5806367270131125e-05,
+      "loss": 0.7317,
+      "step": 10490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5802484047676784e-05,
+      "loss": 0.8322,
+      "step": 10495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5798599192919014e-05,
+      "loss": 0.6434,
+      "step": 10500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.579471270616264e-05,
+      "loss": 0.8798,
+      "step": 10505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.579082458771261e-05,
+      "loss": 0.731,
+      "step": 10510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.578693483787404e-05,
+      "loss": 0.7318,
+      "step": 10515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5783043456952126e-05,
+      "loss": 0.8649,
+      "step": 10520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.577915044525221e-05,
+      "loss": 0.7678,
+      "step": 10525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5775255803079776e-05,
+      "loss": 0.6588,
+      "step": 10530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.577135953074042e-05,
+      "loss": 0.764,
+      "step": 10535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.576746162853986e-05,
+      "loss": 0.9417,
+      "step": 10540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.576356209678396e-05,
+      "loss": 0.7035,
+      "step": 10545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5759660935778716e-05,
+      "loss": 0.7982,
+      "step": 10550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.575575814583022e-05,
+      "loss": 0.6423,
+      "step": 10555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.575185372724472e-05,
+      "loss": 0.7089,
+      "step": 10560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5747947680328574e-05,
+      "loss": 0.6855,
+      "step": 10565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.574404000538829e-05,
+      "loss": 0.784,
+      "step": 10570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.574013070273049e-05,
+      "loss": 0.7017,
+      "step": 10575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5736219772661906e-05,
+      "loss": 0.7412,
+      "step": 10580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.573230721548944e-05,
+      "loss": 0.8399,
+      "step": 10585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.572839303152008e-05,
+      "loss": 0.75,
+      "step": 10590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5724477221060965e-05,
+      "loss": 0.8845,
+      "step": 10595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.572055978441937e-05,
+      "loss": 0.7167,
+      "step": 10600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.571664072190266e-05,
+      "loss": 0.8206,
+      "step": 10605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.571272003381836e-05,
+      "loss": 0.7914,
+      "step": 10610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.570879772047412e-05,
+      "loss": 0.9396,
+      "step": 10615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5704873782177704e-05,
+      "loss": 0.8118,
+      "step": 10620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5700948219237015e-05,
+      "loss": 0.7251,
+      "step": 10625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.569702103196008e-05,
+      "loss": 0.7845,
+      "step": 10630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.569309222065505e-05,
+      "loss": 0.68,
+      "step": 10635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.56891617856302e-05,
+      "loss": 0.7675,
+      "step": 10640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.568522972719395e-05,
+      "loss": 0.663,
+      "step": 10645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.568129604565483e-05,
+      "loss": 0.7192,
+      "step": 10650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5677360741321495e-05,
+      "loss": 0.7118,
+      "step": 10655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5673423814502754e-05,
+      "loss": 0.7883,
+      "step": 10660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.566948526550751e-05,
+      "loss": 0.7901,
+      "step": 10665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.566554509464482e-05,
+      "loss": 0.7413,
+      "step": 10670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.566160330222384e-05,
+      "loss": 0.8076,
+      "step": 10675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.565765988855389e-05,
+      "loss": 0.7808,
+      "step": 10680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.565371485394438e-05,
+      "loss": 0.7337,
+      "step": 10685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5649768198704867e-05,
+      "loss": 0.768,
+      "step": 10690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.564581992314504e-05,
+      "loss": 0.6409,
+      "step": 10695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.564187002757471e-05,
+      "loss": 0.9355,
+      "step": 10700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.563791851230379e-05,
+      "loss": 0.9525,
+      "step": 10705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.563396537764238e-05,
+      "loss": 0.7921,
+      "step": 10710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.563001062390062e-05,
+      "loss": 0.6906,
+      "step": 10715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.562605425138887e-05,
+      "loss": 0.8456,
+      "step": 10720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5622096260417556e-05,
+      "loss": 0.7792,
+      "step": 10725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5618136651297255e-05,
+      "loss": 0.6354,
+      "step": 10730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.561417542433866e-05,
+      "loss": 0.7007,
+      "step": 10735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.561021257985259e-05,
+      "loss": 0.8787,
+      "step": 10740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5606248118150005e-05,
+      "loss": 0.7196,
+      "step": 10745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5602282039541984e-05,
+      "loss": 0.938,
+      "step": 10750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.559831434433973e-05,
+      "loss": 0.8162,
+      "step": 10755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5594345032854566e-05,
+      "loss": 0.8315,
+      "step": 10760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.559037410539797e-05,
+      "loss": 0.9728,
+      "step": 10765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.558640156228151e-05,
+      "loss": 0.9368,
+      "step": 10770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5582427403816906e-05,
+      "loss": 0.7606,
+      "step": 10775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.557845163031601e-05,
+      "loss": 0.7466,
+      "step": 10780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.557447424209076e-05,
+      "loss": 0.9268,
+      "step": 10785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.557049523945327e-05,
+      "loss": 0.8981,
+      "step": 10790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.556651462271575e-05,
+      "loss": 0.98,
+      "step": 10795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5562532392190556e-05,
+      "loss": 0.8729,
+      "step": 10800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.555854854819015e-05,
+      "loss": 0.6925,
+      "step": 10805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.555456309102714e-05,
+      "loss": 0.8089,
+      "step": 10810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.555057602101424e-05,
+      "loss": 0.6679,
+      "step": 10815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5546587338464316e-05,
+      "loss": 0.7568,
+      "step": 10820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.554259704369034e-05,
+      "loss": 0.8799,
+      "step": 10825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5538605137005416e-05,
+      "loss": 0.827,
+      "step": 10830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.553461161872278e-05,
+      "loss": 0.7771,
+      "step": 10835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5530616489155785e-05,
+      "loss": 0.6974,
+      "step": 10840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.552661974861793e-05,
+      "loss": 0.8431,
+      "step": 10845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5522621397422805e-05,
+      "loss": 0.818,
+      "step": 10850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.551862143588416e-05,
+      "loss": 0.8314,
+      "step": 10855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.551461986431587e-05,
+      "loss": 0.7711,
+      "step": 10860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.551061668303189e-05,
+      "loss": 0.7723,
+      "step": 10865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.550661189234638e-05,
+      "loss": 0.897,
+      "step": 10870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.550260549257356e-05,
+      "loss": 0.7719,
+      "step": 10875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.549859748402779e-05,
+      "loss": 0.6446,
+      "step": 10880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.549458786702358e-05,
+      "loss": 0.91,
+      "step": 10885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.549057664187556e-05,
+      "loss": 0.7295,
+      "step": 10890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5486563808898465e-05,
+      "loss": 0.7739,
+      "step": 10895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.548254936840716e-05,
+      "loss": 0.8616,
+      "step": 10900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5478533320716664e-05,
+      "loss": 0.6855,
+      "step": 10905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.547451566614209e-05,
+      "loss": 0.7904,
+      "step": 10910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5470496404998694e-05,
+      "loss": 0.7656,
+      "step": 10915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.546647553760186e-05,
+      "loss": 0.6199,
+      "step": 10920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5462453064267085e-05,
+      "loss": 0.7672,
+      "step": 10925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.545842898531001e-05,
+      "loss": 0.9704,
+      "step": 10930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5454403301046376e-05,
+      "loss": 0.8178,
+      "step": 10935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5450376011792076e-05,
+      "loss": 0.7388,
+      "step": 10940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.544634711786312e-05,
+      "loss": 0.8091,
+      "step": 10945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.544231661957563e-05,
+      "loss": 0.8149,
+      "step": 10950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.543828451724588e-05,
+      "loss": 0.6827,
+      "step": 10955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.543425081119024e-05,
+      "loss": 0.7941,
+      "step": 10960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.543021550172524e-05,
+      "loss": 0.8306,
+      "step": 10965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.542617858916751e-05,
+      "loss": 0.9001,
+      "step": 10970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.542214007383381e-05,
+      "loss": 0.6837,
+      "step": 10975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.541809995604104e-05,
+      "loss": 0.758,
+      "step": 10980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.541405823610619e-05,
+      "loss": 0.8964,
+      "step": 10985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.541001491434643e-05,
+      "loss": 0.8149,
+      "step": 10990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.540596999107901e-05,
+      "loss": 0.5443,
+      "step": 10995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.540192346662133e-05,
+      "loss": 0.7648,
+      "step": 11000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5397875341290906e-05,
+      "loss": 0.7131,
+      "step": 11005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.539382561540537e-05,
+      "loss": 0.714,
+      "step": 11010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5389774289282506e-05,
+      "loss": 0.6926,
+      "step": 11015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5385721363240205e-05,
+      "loss": 0.7779,
+      "step": 11020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.538166683759648e-05,
+      "loss": 0.7415,
+      "step": 11025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.537761071266948e-05,
+      "loss": 0.7751,
+      "step": 11030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.537355298877747e-05,
+      "loss": 0.8838,
+      "step": 11035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.536949366623887e-05,
+      "loss": 0.723,
+      "step": 11040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5365432745372173e-05,
+      "loss": 1.0116,
+      "step": 11045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5361370226496034e-05,
+      "loss": 0.7479,
+      "step": 11050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.535730610992924e-05,
+      "loss": 0.7617,
+      "step": 11055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.535324039599067e-05,
+      "loss": 0.8237,
+      "step": 11060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5349173084999366e-05,
+      "loss": 0.6832,
+      "step": 11065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.534510417727445e-05,
+      "loss": 0.7419,
+      "step": 11070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5341033673135235e-05,
+      "loss": 0.7406,
+      "step": 11075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5336961572901084e-05,
+      "loss": 0.704,
+      "step": 11080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.533288787689154e-05,
+      "loss": 0.854,
+      "step": 11085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.532881258542625e-05,
+      "loss": 0.8356,
+      "step": 11090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.532473569882498e-05,
+      "loss": 0.7699,
+      "step": 11095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5320657217407645e-05,
+      "loss": 0.7127,
+      "step": 11100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.531657714149427e-05,
+      "loss": 0.7333,
+      "step": 11105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.531249547140498e-05,
+      "loss": 0.8154,
+      "step": 11110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.530841220746008e-05,
+      "loss": 0.7927,
+      "step": 11115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.530432734997996e-05,
+      "loss": 0.6829,
+      "step": 11120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.530024089928513e-05,
+      "loss": 0.6931,
+      "step": 11125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5296152855696275e-05,
+      "loss": 0.6714,
+      "step": 11130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.529206321953413e-05,
+      "loss": 0.8873,
+      "step": 11135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5287971991119626e-05,
+      "loss": 0.7947,
+      "step": 11140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.528387917077378e-05,
+      "loss": 0.7713,
+      "step": 11145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.527978475881774e-05,
+      "loss": 0.851,
+      "step": 11150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.527568875557278e-05,
+      "loss": 0.7505,
+      "step": 11155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.52715911613603e-05,
+      "loss": 0.7439,
+      "step": 11160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5267491976501834e-05,
+      "loss": 0.7772,
+      "step": 11165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5263391201319016e-05,
+      "loss": 0.7568,
+      "step": 11170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5259288836133635e-05,
+      "loss": 0.9697,
+      "step": 11175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.525518488126758e-05,
+      "loss": 0.8261,
+      "step": 11180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.525107933704289e-05,
+      "loss": 0.7863,
+      "step": 11185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.524697220378169e-05,
+      "loss": 0.735,
+      "step": 11190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.524286348180627e-05,
+      "loss": 0.8028,
+      "step": 11195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5238753171439024e-05,
+      "loss": 0.8267,
+      "step": 11200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5234641273002474e-05,
+      "loss": 0.6984,
+      "step": 11205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.523052778681928e-05,
+      "loss": 0.8856,
+      "step": 11210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.522641271321219e-05,
+      "loss": 0.9242,
+      "step": 11215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.522229605250412e-05,
+      "loss": 0.6925,
+      "step": 11220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.521817780501808e-05,
+      "loss": 0.738,
+      "step": 11225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5214057971077216e-05,
+      "loss": 0.837,
+      "step": 11230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.520993655100481e-05,
+      "loss": 0.8385,
+      "step": 11235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5205813545124244e-05,
+      "loss": 0.7381,
+      "step": 11240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.520168895375905e-05,
+      "loss": 0.8823,
+      "step": 11245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.519756277723285e-05,
+      "loss": 0.6183,
+      "step": 11250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.519343501586943e-05,
+      "loss": 0.7952,
+      "step": 11255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.518930566999267e-05,
+      "loss": 0.6961,
+      "step": 11260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.51851747399266e-05,
+      "loss": 0.8361,
+      "step": 11265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5181042225995344e-05,
+      "loss": 0.6481,
+      "step": 11270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.517690812852319e-05,
+      "loss": 0.8791,
+      "step": 11275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.51727724478345e-05,
+      "loss": 0.8789,
+      "step": 11280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5168635184253805e-05,
+      "loss": 0.8676,
+      "step": 11285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5165324233925155e-05,
+      "loss": 0.8545,
+      "step": 11290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.516118412195701e-05,
+      "loss": 0.6939,
+      "step": 11295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5157042428006165e-05,
+      "loss": 0.79,
+      "step": 11300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.515289915239759e-05,
+      "loss": 0.7704,
+      "step": 11305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.514875429545639e-05,
+      "loss": 0.8389,
+      "step": 11310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.514460785750782e-05,
+      "loss": 0.9743,
+      "step": 11315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.514045983887721e-05,
+      "loss": 0.6996,
+      "step": 11320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.513631023989007e-05,
+      "loss": 0.905,
+      "step": 11325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5132159060871985e-05,
+      "loss": 0.6121,
+      "step": 11330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.512800630214869e-05,
+      "loss": 0.7722,
+      "step": 11335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5123851964046045e-05,
+      "loss": 0.7694,
+      "step": 11340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.511969604689001e-05,
+      "loss": 0.8299,
+      "step": 11345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.511553855100671e-05,
+      "loss": 0.8293,
+      "step": 11350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.511137947672236e-05,
+      "loss": 0.662,
+      "step": 11355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5107218824363306e-05,
+      "loss": 0.88,
+      "step": 11360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.510305659425602e-05,
+      "loss": 0.9379,
+      "step": 11365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.509889278672711e-05,
+      "loss": 0.7759,
+      "step": 11370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.509472740210328e-05,
+      "loss": 0.8812,
+      "step": 11375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.509056044071138e-05,
+      "loss": 0.8354,
+      "step": 11380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.508639190287839e-05,
+      "loss": 0.8905,
+      "step": 11385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5082221788931384e-05,
+      "loss": 0.7021,
+      "step": 11390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.507805009919759e-05,
+      "loss": 0.7593,
+      "step": 11395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.507387683400434e-05,
+      "loss": 0.8714,
+      "step": 11400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.506970199367909e-05,
+      "loss": 0.671,
+      "step": 11405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.506552557854945e-05,
+      "loss": 0.6942,
+      "step": 11410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.50613475889431e-05,
+      "loss": 0.6638,
+      "step": 11415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.50571680251879e-05,
+      "loss": 0.7896,
+      "step": 11420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.505298688761179e-05,
+      "loss": 0.7905,
+      "step": 11425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5048804176542855e-05,
+      "loss": 0.8022,
+      "step": 11430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5044619892309295e-05,
+      "loss": 0.7792,
+      "step": 11435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.504043403523944e-05,
+      "loss": 0.6195,
+      "step": 11440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5036246605661754e-05,
+      "loss": 0.7328,
+      "step": 11445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5032057603904785e-05,
+      "loss": 0.757,
+      "step": 11450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5027867030297257e-05,
+      "loss": 0.7619,
+      "step": 11455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5023674885167966e-05,
+      "loss": 0.7739,
+      "step": 11460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.501948116884587e-05,
+      "loss": 0.8146,
+      "step": 11465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.501528588166004e-05,
+      "loss": 0.7016,
+      "step": 11470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5011089023939655e-05,
+      "loss": 0.8032,
+      "step": 11475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.500689059601403e-05,
+      "loss": 0.7545,
+      "step": 11480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5002690598212616e-05,
+      "loss": 0.8386,
+      "step": 11485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.499848903086497e-05,
+      "loss": 0.7125,
+      "step": 11490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.499428589430075e-05,
+      "loss": 0.7925,
+      "step": 11495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.499008118884979e-05,
+      "loss": 0.7435,
+      "step": 11500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4985874914842015e-05,
+      "loss": 0.8457,
+      "step": 11505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.498166707260747e-05,
+      "loss": 0.8124,
+      "step": 11510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.497745766247634e-05,
+      "loss": 0.8022,
+      "step": 11515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4973246684778905e-05,
+      "loss": 0.7935,
+      "step": 11520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.496903413984561e-05,
+      "loss": 0.7948,
+      "step": 11525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.496482002800699e-05,
+      "loss": 0.7123,
+      "step": 11530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.496060434959371e-05,
+      "loss": 0.6831,
+      "step": 11535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.495638710493656e-05,
+      "loss": 0.5735,
+      "step": 11540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.495216829436646e-05,
+      "loss": 0.6868,
+      "step": 11545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4947947918214444e-05,
+      "loss": 0.8139,
+      "step": 11550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4943725976811666e-05,
+      "loss": 0.7321,
+      "step": 11555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.493950247048942e-05,
+      "loss": 0.6833,
+      "step": 11560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.49361225389123e-05,
+      "loss": 0.7848,
+      "step": 11565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4931896216570216e-05,
+      "loss": 0.733,
+      "step": 11570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4927668330236895e-05,
+      "loss": 0.7849,
+      "step": 11575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4923438880244094e-05,
+      "loss": 0.8142,
+      "step": 11580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4919207866923674e-05,
+      "loss": 0.8539,
+      "step": 11585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4914975290607644e-05,
+      "loss": 0.8426,
+      "step": 11590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.49107411516281e-05,
+      "loss": 0.8092,
+      "step": 11595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.49065054503173e-05,
+      "loss": 0.8492,
+      "step": 11600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.49022681870076e-05,
+      "loss": 0.7114,
+      "step": 11605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4898029362031486e-05,
+      "loss": 0.7255,
+      "step": 11610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.489378897572155e-05,
+      "loss": 0.6828,
+      "step": 11615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.488954702841054e-05,
+      "loss": 0.8522,
+      "step": 11620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.48853035204313e-05,
+      "loss": 0.7798,
+      "step": 11625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4881058452116803e-05,
+      "loss": 0.6972,
+      "step": 11630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.487681182380015e-05,
+      "loss": 0.7316,
+      "step": 11635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4872563635814555e-05,
+      "loss": 0.7366,
+      "step": 11640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.486831388849336e-05,
+      "loss": 0.7457,
+      "step": 11645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.486406258217003e-05,
+      "loss": 0.8564,
+      "step": 11650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.485980971717816e-05,
+      "loss": 0.7643,
+      "step": 11655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4855555293851445e-05,
+      "loss": 0.7106,
+      "step": 11660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.485129931252373e-05,
+      "loss": 0.7829,
+      "step": 11665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.484704177352895e-05,
+      "loss": 0.9266,
+      "step": 11670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.484278267720119e-05,
+      "loss": 0.9619,
+      "step": 11675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4838522023874655e-05,
+      "loss": 0.9012,
+      "step": 11680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.483425981388365e-05,
+      "loss": 0.8963,
+      "step": 11685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4829996047562626e-05,
+      "loss": 0.7888,
+      "step": 11690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.482573072524615e-05,
+      "loss": 0.8111,
+      "step": 11695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.482146384726889e-05,
+      "loss": 0.804,
+      "step": 11700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4817195413965684e-05,
+      "loss": 0.7884,
+      "step": 11705
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4812925425671435e-05,
+      "loss": 0.8359,
+      "step": 11710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4808653882721205e-05,
+      "loss": 0.8497,
+      "step": 11715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.480438078545017e-05,
+      "loss": 0.7625,
+      "step": 11720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.480010613419363e-05,
+      "loss": 0.8469,
+      "step": 11725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4795829929286983e-05,
+      "loss": 0.6867,
+      "step": 11730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4791552171065793e-05,
+      "loss": 0.7706,
+      "step": 11735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.478727285986571e-05,
+      "loss": 0.7705,
+      "step": 11740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4782991996022516e-05,
+      "loss": 0.7344,
+      "step": 11745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4778709579872125e-05,
+      "loss": 0.649,
+      "step": 11750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.477442561175056e-05,
+      "loss": 0.7686,
+      "step": 11755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4770140091993975e-05,
+      "loss": 0.7635,
+      "step": 11760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4765853020938616e-05,
+      "loss": 0.8724,
+      "step": 11765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.476156439892092e-05,
+      "loss": 0.7394,
+      "step": 11770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.475727422627736e-05,
+      "loss": 0.878,
+      "step": 11775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.475298250334459e-05,
+      "loss": 0.7403,
+      "step": 11780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.474868923045937e-05,
+      "loss": 0.6866,
+      "step": 11785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.474439440795857e-05,
+      "loss": 0.6778,
+      "step": 11790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.47400980361792e-05,
+      "loss": 0.8495,
+      "step": 11795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4735800115458376e-05,
+      "loss": 0.8818,
+      "step": 11800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4731500646133344e-05,
+      "loss": 0.7179,
+      "step": 11805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4727199628541474e-05,
+      "loss": 0.7991,
+      "step": 11810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4722897063020244e-05,
+      "loss": 0.7647,
+      "step": 11815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.471859294990727e-05,
+      "loss": 0.7148,
+      "step": 11820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.471428728954027e-05,
+      "loss": 0.7418,
+      "step": 11825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.470998008225711e-05,
+      "loss": 0.7474,
+      "step": 11830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.470567132839575e-05,
+      "loss": 0.7828,
+      "step": 11835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.47013610282943e-05,
+      "loss": 0.8215,
+      "step": 11840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.469704918229096e-05,
+      "loss": 0.8326,
+      "step": 11845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.469273579072407e-05,
+      "loss": 0.8397,
+      "step": 11850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.46884208539321e-05,
+      "loss": 0.6773,
+      "step": 11855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.468410437225361e-05,
+      "loss": 0.822,
+      "step": 11860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4679786346027305e-05,
+      "loss": 0.7635,
+      "step": 11865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4675466775592016e-05,
+      "loss": 0.7476,
+      "step": 11870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.467114566128668e-05,
+      "loss": 0.7371,
+      "step": 11875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4666823003450365e-05,
+      "loss": 0.8943,
+      "step": 11880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4662498802422245e-05,
+      "loss": 0.7084,
+      "step": 11885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4658173058541644e-05,
+      "loss": 0.9649,
+      "step": 11890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4653845772147976e-05,
+      "loss": 0.758,
+      "step": 11895
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.464951694358078e-05,
+      "loss": 0.737,
+      "step": 11900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.464518657317974e-05,
+      "loss": 0.9439,
+      "step": 11905
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.464085466128465e-05,
+      "loss": 0.8029,
+      "step": 11910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.463652120823541e-05,
+      "loss": 0.8127,
+      "step": 11915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4632186214372056e-05,
+      "loss": 0.7715,
+      "step": 11920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.462784968003474e-05,
+      "loss": 0.8689,
+      "step": 11925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4623511605563736e-05,
+      "loss": 0.7674,
+      "step": 11930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.461917199129944e-05,
+      "loss": 0.8544,
+      "step": 11935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4614830837582364e-05,
+      "loss": 0.8383,
+      "step": 11940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4610488144753157e-05,
+      "loss": 0.8173,
+      "step": 11945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.460614391315255e-05,
+      "loss": 0.6548,
+      "step": 11950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.460179814312145e-05,
+      "loss": 0.7145,
+      "step": 11955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4597450835000835e-05,
+      "loss": 0.6893,
+      "step": 11960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.459310198913183e-05,
+      "loss": 0.6765,
+      "step": 11965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4588751605855686e-05,
+      "loss": 0.8187,
+      "step": 11970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.458439968551374e-05,
+      "loss": 0.7609,
+      "step": 11975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.45800462284475e-05,
+      "loss": 0.8555,
+      "step": 11980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.457569123499854e-05,
+      "loss": 0.8014,
+      "step": 11985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4571334705508607e-05,
+      "loss": 0.7674,
+      "step": 11990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4566976640319527e-05,
+      "loss": 0.6711,
+      "step": 11995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.456261703977327e-05,
+      "loss": 0.6891,
+      "step": 12000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.455825590421192e-05,
+      "loss": 0.7042,
+      "step": 12005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.455389323397768e-05,
+      "loss": 0.8806,
+      "step": 12010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4549529029412884e-05,
+      "loss": 0.8583,
+      "step": 12015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.454516329085996e-05,
+      "loss": 0.8913,
+      "step": 12020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.454079601866148e-05,
+      "loss": 0.6946,
+      "step": 12025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4536427213160134e-05,
+      "loss": 0.8268,
+      "step": 12030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.453205687469872e-05,
+      "loss": 0.8099,
+      "step": 12035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.452768500362017e-05,
+      "loss": 0.7899,
+      "step": 12040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4523311600267535e-05,
+      "loss": 0.805,
+      "step": 12045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.451893666498397e-05,
+      "loss": 0.6887,
+      "step": 12050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.451456019811277e-05,
+      "loss": 0.6805,
+      "step": 12055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4510182199997344e-05,
+      "loss": 0.7977,
+      "step": 12060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.450580267098121e-05,
+      "loss": 0.8631,
+      "step": 12065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4501421611408024e-05,
+      "loss": 0.707,
+      "step": 12070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.449703902162156e-05,
+      "loss": 0.6694,
+      "step": 12075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.449265490196568e-05,
+      "loss": 0.7346,
+      "step": 12080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.448826925278442e-05,
+      "loss": 0.7558,
+      "step": 12085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.448388207442189e-05,
+      "loss": 0.6895,
+      "step": 12090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4479493367222346e-05,
+      "loss": 0.8271,
+      "step": 12095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.447510313153015e-05,
+      "loss": 0.6803,
+      "step": 12100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.44707113676898e-05,
+      "loss": 0.7462,
+      "step": 12105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.446631807604589e-05,
+      "loss": 0.6744,
+      "step": 12110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4461923256943147e-05,
+      "loss": 0.6925,
+      "step": 12115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4457526910726434e-05,
+      "loss": 0.7863,
+      "step": 12120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.445312903774071e-05,
+      "loss": 0.8851,
+      "step": 12125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4448729638331056e-05,
+      "loss": 0.8196,
+      "step": 12130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.444432871284269e-05,
+      "loss": 0.7028,
+      "step": 12135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.443992626162092e-05,
+      "loss": 0.7868,
+      "step": 12140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.443552228501121e-05,
+      "loss": 0.8979,
+      "step": 12145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.443111678335911e-05,
+      "loss": 0.7723,
+      "step": 12150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4426709757010324e-05,
+      "loss": 0.8113,
+      "step": 12155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.442230120631065e-05,
+      "loss": 0.6194,
+      "step": 12160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4417891131606005e-05,
+      "loss": 0.7462,
+      "step": 12165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.441347953324244e-05,
+      "loss": 0.8843,
+      "step": 12170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4409066411566114e-05,
+      "loss": 0.7431,
+      "step": 12175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.440465176692332e-05,
+      "loss": 0.7888,
+      "step": 12180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.440023559966045e-05,
+      "loss": 0.8514,
+      "step": 12185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.439581791012403e-05,
+      "loss": 0.7558,
+      "step": 12190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.439139869866071e-05,
+      "loss": 0.7756,
+      "step": 12195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.438697796561724e-05,
+      "loss": 0.7798,
+      "step": 12200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.43825557113405e-05,
+      "loss": 0.6846,
+      "step": 12205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.43781319361775e-05,
+      "loss": 0.6656,
+      "step": 12210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.437370664047535e-05,
+      "loss": 0.6589,
+      "step": 12215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.436927982458129e-05,
+      "loss": 0.8864,
+      "step": 12220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.436485148884268e-05,
+      "loss": 0.9256,
+      "step": 12225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4360421633607005e-05,
+      "loss": 0.8187,
+      "step": 12230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.435599025922185e-05,
+      "loss": 0.6863,
+      "step": 12235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4351557366034934e-05,
+      "loss": 0.6832,
+      "step": 12240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.43471229543941e-05,
+      "loss": 0.6344,
+      "step": 12245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.434268702464728e-05,
+      "loss": 0.8072,
+      "step": 12250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4338249577142564e-05,
+      "loss": 0.6251,
+      "step": 12255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.433381061222814e-05,
+      "loss": 0.8505,
+      "step": 12260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.432937013025232e-05,
+      "loss": 0.8873,
+      "step": 12265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4324928131563546e-05,
+      "loss": 0.818,
+      "step": 12270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.432048461651034e-05,
+      "loss": 0.6063,
+      "step": 12275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.43160395854414e-05,
+      "loss": 0.7536,
+      "step": 12280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.431159303870549e-05,
+      "loss": 0.8557,
+      "step": 12285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.430714497665153e-05,
+      "loss": 0.7247,
+      "step": 12290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.430269539962854e-05,
+      "loss": 0.8816,
+      "step": 12295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.429824430798566e-05,
+      "loss": 0.8217,
+      "step": 12300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.429379170207215e-05,
+      "loss": 0.6246,
+      "step": 12305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.428933758223741e-05,
+      "loss": 0.9075,
+      "step": 12310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.428488194883093e-05,
+      "loss": 0.7784,
+      "step": 12315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4280424802202315e-05,
+      "loss": 0.7788,
+      "step": 12320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.427596614270133e-05,
+      "loss": 0.8232,
+      "step": 12325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.427150597067781e-05,
+      "loss": 0.8237,
+      "step": 12330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.426704428648174e-05,
+      "loss": 0.8094,
+      "step": 12335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.426258109046321e-05,
+      "loss": 0.7078,
+      "step": 12340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.425811638297243e-05,
+      "loss": 0.8493,
+      "step": 12345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4253650164359726e-05,
+      "loss": 0.813,
+      "step": 12350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.424918243497557e-05,
+      "loss": 0.8007,
+      "step": 12355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4244713195170515e-05,
+      "loss": 0.8579,
+      "step": 12360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.424024244529524e-05,
+      "loss": 0.9052,
+      "step": 12365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4235770185700575e-05,
+      "loss": 0.7603,
+      "step": 12370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4231296416737425e-05,
+      "loss": 0.7722,
+      "step": 12375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.422682113875683e-05,
+      "loss": 0.8097,
+      "step": 12380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4222344352109955e-05,
+      "loss": 0.7279,
+      "step": 12385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.421786605714808e-05,
+      "loss": 0.7405,
+      "step": 12390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.421338625422261e-05,
+      "loss": 0.6884,
+      "step": 12395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4208904943685045e-05,
+      "loss": 0.854,
+      "step": 12400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.420442212588703e-05,
+      "loss": 0.74,
+      "step": 12405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4199937801180314e-05,
+      "loss": 0.672,
+      "step": 12410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.419545196991677e-05,
+      "loss": 0.8228,
+      "step": 12415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4190964632448384e-05,
+      "loss": 0.8464,
+      "step": 12420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.418647578912726e-05,
+      "loss": 0.8099,
+      "step": 12425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4181985440305626e-05,
+      "loss": 0.7404,
+      "step": 12430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.417749358633582e-05,
+      "loss": 0.6858,
+      "step": 12435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4173000227570315e-05,
+      "loss": 0.7632,
+      "step": 12440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.416850536436168e-05,
+      "loss": 0.8343,
+      "step": 12445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.416400899706261e-05,
+      "loss": 0.8169,
+      "step": 12450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.415951112602593e-05,
+      "loss": 0.7808,
+      "step": 12455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.415501175160458e-05,
+      "loss": 0.759,
+      "step": 12460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.415051087415159e-05,
+      "loss": 0.8992,
+      "step": 12465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4146008494020144e-05,
+      "loss": 0.9015,
+      "step": 12470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.414150461156352e-05,
+      "loss": 0.7866,
+      "step": 12475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4136999227135136e-05,
+      "loss": 0.7988,
+      "step": 12480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.41324923410885e-05,
+      "loss": 0.8147,
+      "step": 12485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.412798395377726e-05,
+      "loss": 0.8618,
+      "step": 12490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.412347406555518e-05,
+      "loss": 0.7601,
+      "step": 12495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.411896267677612e-05,
+      "loss": 0.7719,
+      "step": 12500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.41144497877941e-05,
+      "loss": 0.7976,
+      "step": 12505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.410993539896321e-05,
+      "loss": 0.7467,
+      "step": 12510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.410541951063768e-05,
+      "loss": 0.6516,
+      "step": 12515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.410090212317187e-05,
+      "loss": 0.8148,
+      "step": 12520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.409638323692024e-05,
+      "loss": 0.9414,
+      "step": 12525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4091862852237355e-05,
+      "loss": 0.9168,
+      "step": 12530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.408734096947794e-05,
+      "loss": 0.7589,
+      "step": 12535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4082817588996805e-05,
+      "loss": 0.85,
+      "step": 12540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.407829271114888e-05,
+      "loss": 0.7882,
+      "step": 12545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4073766336289216e-05,
+      "loss": 0.9139,
+      "step": 12550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4069238464772994e-05,
+      "loss": 0.7504,
+      "step": 12555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4065615090206036e-05,
+      "loss": 0.8008,
+      "step": 12560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.406108452560339e-05,
+      "loss": 0.7145,
+      "step": 12565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.405655246533929e-05,
+      "loss": 0.8822,
+      "step": 12570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.405201890976934e-05,
+      "loss": 0.8377,
+      "step": 12575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.404748385924927e-05,
+      "loss": 0.8367,
+      "step": 12580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4042947314134935e-05,
+      "loss": 0.7154,
+      "step": 12585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4038409274782306e-05,
+      "loss": 0.9397,
+      "step": 12590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.403386974154747e-05,
+      "loss": 0.9481,
+      "step": 12595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4029328714786613e-05,
+      "loss": 1.0109,
+      "step": 12600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.402478619485608e-05,
+      "loss": 0.7807,
+      "step": 12605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.402024218211229e-05,
+      "loss": 0.9127,
+      "step": 12610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.40156966769118e-05,
+      "loss": 0.8309,
+      "step": 12615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4011149679611285e-05,
+      "loss": 0.7362,
+      "step": 12620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.400660119056753e-05,
+      "loss": 0.8508,
+      "step": 12625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.400205121013744e-05,
+      "loss": 0.8496,
+      "step": 12630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.399749973867804e-05,
+      "loss": 0.8917,
+      "step": 12635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.399294677654646e-05,
+      "loss": 0.7266,
+      "step": 12640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.398839232409997e-05,
+      "loss": 0.8649,
+      "step": 12645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.398383638169594e-05,
+      "loss": 0.9445,
+      "step": 12650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.397927894969185e-05,
+      "loss": 0.8074,
+      "step": 12655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.397472002844532e-05,
+      "loss": 0.6829,
+      "step": 12660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3970159618314064e-05,
+      "loss": 0.8344,
+      "step": 12665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.396559771965592e-05,
+      "loss": 0.8884,
+      "step": 12670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.396103433282885e-05,
+      "loss": 0.8908,
+      "step": 12675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.395646945819094e-05,
+      "loss": 0.7523,
+      "step": 12680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.395190309610037e-05,
+      "loss": 0.7315,
+      "step": 12685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3947335246915445e-05,
+      "loss": 0.6234,
+      "step": 12690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3942765910994594e-05,
+      "loss": 0.7772,
+      "step": 12695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3938195088696356e-05,
+      "loss": 0.8661,
+      "step": 12700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.393362278037938e-05,
+      "loss": 0.7324,
+      "step": 12705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.392904898640246e-05,
+      "loss": 0.8745,
+      "step": 12710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.392447370712447e-05,
+      "loss": 0.7445,
+      "step": 12715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.391989694290443e-05,
+      "loss": 0.8657,
+      "step": 12720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3915318694101445e-05,
+      "loss": 0.8338,
+      "step": 12725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.391073896107477e-05,
+      "loss": 0.7449,
+      "step": 12730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3906157744183766e-05,
+      "loss": 0.7508,
+      "step": 12735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.39015750437879e-05,
+      "loss": 0.786,
+      "step": 12740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.389699086024676e-05,
+      "loss": 0.6942,
+      "step": 12745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.389240519392005e-05,
+      "loss": 1.0015,
+      "step": 12750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.38878180451676e-05,
+      "loss": 0.7224,
+      "step": 12755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3883229414349334e-05,
+      "loss": 0.7491,
+      "step": 12760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.387863930182532e-05,
+      "loss": 0.7287,
+      "step": 12765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.387404770795573e-05,
+      "loss": 0.7377,
+      "step": 12770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.386945463310085e-05,
+      "loss": 0.8481,
+      "step": 12775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3864860077621074e-05,
+      "loss": 0.7454,
+      "step": 12780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.386026404187693e-05,
+      "loss": 0.7473,
+      "step": 12785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.385566652622906e-05,
+      "loss": 0.6627,
+      "step": 12790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3851067531038206e-05,
+      "loss": 0.8933,
+      "step": 12795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.384646705666524e-05,
+      "loss": 0.8745,
+      "step": 12800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.384186510347114e-05,
+      "loss": 0.6821,
+      "step": 12805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.383726167181702e-05,
+      "loss": 0.7889,
+      "step": 12810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.383265676206408e-05,
+      "loss": 0.8617,
+      "step": 12815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.382805037457367e-05,
+      "loss": 0.8249,
+      "step": 12820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3823442509707225e-05,
+      "loss": 0.7794,
+      "step": 12825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3818833167826316e-05,
+      "loss": 0.9807,
+      "step": 12830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.381422234929262e-05,
+      "loss": 0.7749,
+      "step": 12835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3809610054467934e-05,
+      "loss": 0.8389,
+      "step": 12840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.380499628371417e-05,
+      "loss": 0.7548,
+      "step": 12845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.380038103739335e-05,
+      "loss": 0.7417,
+      "step": 12850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3795764315867625e-05,
+      "loss": 0.7841,
+      "step": 12855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3791146119499246e-05,
+      "loss": 0.8969,
+      "step": 12860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3786526448650614e-05,
+      "loss": 0.7581,
+      "step": 12865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.378190530368418e-05,
+      "loss": 0.8292,
+      "step": 12870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.377728268496257e-05,
+      "loss": 0.7353,
+      "step": 12875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.377265859284851e-05,
+      "loss": 0.7211,
+      "step": 12880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.376803302770483e-05,
+      "loss": 0.7554,
+      "step": 12885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.376340598989448e-05,
+      "loss": 0.815,
+      "step": 12890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3758777479780545e-05,
+      "loss": 0.8335,
+      "step": 12895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.375414749772619e-05,
+      "loss": 0.7592,
+      "step": 12900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.374951604409473e-05,
+      "loss": 0.8748,
+      "step": 12905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.374488311924958e-05,
+      "loss": 0.7128,
+      "step": 12910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3740248723554256e-05,
+      "loss": 0.7926,
+      "step": 12915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.37356128573724e-05,
+      "loss": 0.7336,
+      "step": 12920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3730975521067805e-05,
+      "loss": 0.679,
+      "step": 12925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.372633671500431e-05,
+      "loss": 0.7649,
+      "step": 12930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3721696439545936e-05,
+      "loss": 0.8468,
+      "step": 12935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.371705469505678e-05,
+      "loss": 0.6405,
+      "step": 12940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.371241148190107e-05,
+      "loss": 0.703,
+      "step": 12945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.370776680044313e-05,
+      "loss": 0.7934,
+      "step": 12950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.370312065104741e-05,
+      "loss": 0.9499,
+      "step": 12955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.369847303407851e-05,
+      "loss": 0.8789,
+      "step": 12960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.369382394990108e-05,
+      "loss": 0.8279,
+      "step": 12965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.368917339887993e-05,
+      "loss": 0.8488,
+      "step": 12970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.368452138137998e-05,
+      "loss": 0.8183,
+      "step": 12975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3679867897766255e-05,
+      "loss": 0.888,
+      "step": 12980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3675212948403896e-05,
+      "loss": 0.7248,
+      "step": 12985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.367055653365816e-05,
+      "loss": 0.9176,
+      "step": 12990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366589865389443e-05,
+      "loss": 0.8361,
+      "step": 12995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3661239309478186e-05,
+      "loss": 0.7936,
+      "step": 13000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365657850077504e-05,
+      "loss": 0.7791,
+      "step": 13005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3651916228150694e-05,
+      "loss": 0.7412,
+      "step": 13010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3647252491971e-05,
+      "loss": 0.8163,
+      "step": 13015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3642587292601886e-05,
+      "loss": 0.778,
+      "step": 13020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363792063040945e-05,
+      "loss": 0.8507,
+      "step": 13025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3633252505759825e-05,
+      "loss": 0.7303,
+      "step": 13030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362858291901933e-05,
+      "loss": 0.8076,
+      "step": 13035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362391187055438e-05,
+      "loss": 0.8979,
+      "step": 13040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361923936073148e-05,
+      "loss": 0.7571,
+      "step": 13045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361456538991727e-05,
+      "loss": 0.6578,
+      "step": 13050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36098899584785e-05,
+      "loss": 0.6153,
+      "step": 13055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3605213066782045e-05,
+      "loss": 0.7902,
+      "step": 13060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.360053471519489e-05,
+      "loss": 0.7363,
+      "step": 13065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35958549040841e-05,
+      "loss": 0.8214,
+      "step": 13070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359117363381691e-05,
+      "loss": 0.7418,
+      "step": 13075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358649090476065e-05,
+      "loss": 0.7491,
+      "step": 13080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358180671728274e-05,
+      "loss": 0.7573,
+      "step": 13085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357712107175073e-05,
+      "loss": 0.6123,
+      "step": 13090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357243396853231e-05,
+      "loss": 0.6387,
+      "step": 13095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356774540799524e-05,
+      "loss": 0.8532,
+      "step": 13100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356305539050744e-05,
+      "loss": 0.7031,
+      "step": 13105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3558363916436894e-05,
+      "loss": 0.6158,
+      "step": 13110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355367098615174e-05,
+      "loss": 0.7412,
+      "step": 13115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354897660002022e-05,
+      "loss": 0.7006,
+      "step": 13120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3544280758410676e-05,
+      "loss": 0.7137,
+      "step": 13125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353958346169159e-05,
+      "loss": 0.854,
+      "step": 13130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353488471023153e-05,
+      "loss": 0.8121,
+      "step": 13135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3530184504399204e-05,
+      "loss": 0.5475,
+      "step": 13140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.352548284456341e-05,
+      "loss": 0.7884,
+      "step": 13145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3520779731093084e-05,
+      "loss": 0.6686,
+      "step": 13150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351607516435725e-05,
+      "loss": 0.8095,
+      "step": 13155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3511369144725066e-05,
+      "loss": 0.6401,
+      "step": 13160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350666167256581e-05,
+      "loss": 0.8313,
+      "step": 13165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350195274824885e-05,
+      "loss": 0.7963,
+      "step": 13170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349724237214368e-05,
+      "loss": 0.6575,
+      "step": 13175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.349253054461991e-05,
+      "loss": 0.8806,
+      "step": 13180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.348781726604726e-05,
+      "loss": 0.7534,
+      "step": 13185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3483102536795566e-05,
+      "loss": 0.8731,
+      "step": 13190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3478386357234786e-05,
+      "loss": 0.7056,
+      "step": 13195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3473668727734966e-05,
+      "loss": 0.9734,
+      "step": 13200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.34689496486663e-05,
+      "loss": 0.833,
+      "step": 13205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3464229120399075e-05,
+      "loss": 0.7495,
+      "step": 13210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.345950714330369e-05,
+      "loss": 0.7892,
+      "step": 13215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3454783717750664e-05,
+      "loss": 0.7466,
+      "step": 13220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.345005884411063e-05,
+      "loss": 0.7017,
+      "step": 13225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3445332522754335e-05,
+      "loss": 0.907,
+      "step": 13230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.344060475405264e-05,
+      "loss": 0.6775,
+      "step": 13235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.343587553837651e-05,
+      "loss": 0.7486,
+      "step": 13240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.343114487609704e-05,
+      "loss": 0.7794,
+      "step": 13245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.342641276758543e-05,
+      "loss": 0.7655,
+      "step": 13250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3421679213212986e-05,
+      "loss": 0.8388,
+      "step": 13255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.341694421335114e-05,
+      "loss": 0.7333,
+      "step": 13260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.341220776837143e-05,
+      "loss": 0.9167,
+      "step": 13265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.34074698786455e-05,
+      "loss": 0.8251,
+      "step": 13270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3402730544545135e-05,
+      "loss": 0.7719,
+      "step": 13275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3397989766442204e-05,
+      "loss": 0.9258,
+      "step": 13280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3393247544708706e-05,
+      "loss": 0.9782,
+      "step": 13285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.338850387971674e-05,
+      "loss": 0.7782,
+      "step": 13290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3383758771838534e-05,
+      "loss": 0.8658,
+      "step": 13295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3379012221446415e-05,
+      "loss": 0.6974,
+      "step": 13300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.337426422891283e-05,
+      "loss": 0.7487,
+      "step": 13305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.336951479461035e-05,
+      "loss": 0.7499,
+      "step": 13310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3364763918911635e-05,
+      "loss": 0.8276,
+      "step": 13315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.336001160218947e-05,
+      "loss": 0.8311,
+      "step": 13320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.335525784481676e-05,
+      "loss": 0.7503,
+      "step": 13325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.335050264716652e-05,
+      "loss": 0.74,
+      "step": 13330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3345746009611865e-05,
+      "loss": 0.6921,
+      "step": 13335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.334098793252604e-05,
+      "loss": 0.8065,
+      "step": 13340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.333622841628239e-05,
+      "loss": 0.8301,
+      "step": 13345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3331467461254387e-05,
+      "loss": 0.7791,
+      "step": 13350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3326705067815606e-05,
+      "loss": 0.6827,
+      "step": 13355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3321941236339724e-05,
+      "loss": 0.658,
+      "step": 13360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.331717596720056e-05,
+      "loss": 0.7526,
+      "step": 13365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.331240926077202e-05,
+      "loss": 0.8567,
+      "step": 13370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3307641117428146e-05,
+      "loss": 0.8012,
+      "step": 13375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3302871537543055e-05,
+      "loss": 0.7934,
+      "step": 13380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.329810052149101e-05,
+      "loss": 0.5949,
+      "step": 13385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.329332806964639e-05,
+      "loss": 0.6297,
+      "step": 13390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.328855418238366e-05,
+      "loss": 0.8759,
+      "step": 13395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.328377886007742e-05,
+      "loss": 0.6256,
+      "step": 13400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3279002103102364e-05,
+      "loss": 0.8063,
+      "step": 13405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.327422391183331e-05,
+      "loss": 0.7435,
+      "step": 13410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3269444286645195e-05,
+      "loss": 0.8356,
+      "step": 13415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.326466322791306e-05,
+      "loss": 0.8603,
+      "step": 13420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.325988073601205e-05,
+      "loss": 0.7352,
+      "step": 13425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.325509681131744e-05,
+      "loss": 0.7229,
+      "step": 13430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3250311454204604e-05,
+      "loss": 0.7984,
+      "step": 13435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.324552466504904e-05,
+      "loss": 0.854,
+      "step": 13440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.324073644422635e-05,
+      "loss": 0.9319,
+      "step": 13445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.323594679211224e-05,
+      "loss": 0.7814,
+      "step": 13450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.323115570908255e-05,
+      "loss": 0.7654,
+      "step": 13455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.322636319551322e-05,
+      "loss": 0.7388,
+      "step": 13460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3221569251780294e-05,
+      "loss": 0.9102,
+      "step": 13465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.321677387825995e-05,
+      "loss": 0.7954,
+      "step": 13470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.321197707532846e-05,
+      "loss": 0.7261,
+      "step": 13475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3207178843362206e-05,
+      "loss": 0.8319,
+      "step": 13480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3202379182737696e-05,
+      "loss": 0.812,
+      "step": 13485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3197578093831546e-05,
+      "loss": 0.6999,
+      "step": 13490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.319277557702048e-05,
+      "loss": 0.7062,
+      "step": 13495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3187971632681334e-05,
+      "loss": 0.7148,
+      "step": 13500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.318316626119106e-05,
+      "loss": 0.7924,
+      "step": 13505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.317835946292673e-05,
+      "loss": 0.8971,
+      "step": 13510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.31735512382655e-05,
+      "loss": 0.8842,
+      "step": 13515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.316874158758465e-05,
+      "loss": 0.7511,
+      "step": 13520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.316393051126161e-05,
+      "loss": 0.826,
+      "step": 13525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.315911800967386e-05,
+      "loss": 0.7259,
+      "step": 13530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.315430408319903e-05,
+      "loss": 0.7994,
+      "step": 13535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.314948873221486e-05,
+      "loss": 0.7357,
+      "step": 13540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3144671957099195e-05,
+      "loss": 0.8327,
+      "step": 13545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3139853758229974e-05,
+      "loss": 0.7013,
+      "step": 13550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3135034135985284e-05,
+      "loss": 0.7645,
+      "step": 13555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.31302130907433e-05,
+      "loss": 0.7555,
+      "step": 13560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.312539062288232e-05,
+      "loss": 0.6446,
+      "step": 13565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3120566732780734e-05,
+      "loss": 0.9196,
+      "step": 13570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.311574142081706e-05,
+      "loss": 0.7571,
+      "step": 13575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.311091468736993e-05,
+      "loss": 0.8022,
+      "step": 13580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.310608653281807e-05,
+      "loss": 0.7824,
+      "step": 13585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.310125695754036e-05,
+      "loss": 0.7804,
+      "step": 13590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3096425961915726e-05,
+      "loss": 0.9084,
+      "step": 13595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.309159354632326e-05,
+      "loss": 0.7308,
+      "step": 13600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.308675971114215e-05,
+      "loss": 0.8479,
+      "step": 13605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3081924456751665e-05,
+      "loss": 0.7876,
+      "step": 13610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.307708778353124e-05,
+      "loss": 0.7154,
+      "step": 13615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.307224969186038e-05,
+      "loss": 0.7687,
+      "step": 13620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.306741018211872e-05,
+      "loss": 0.8058,
+      "step": 13625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3062569254685994e-05,
+      "loss": 0.8055,
+      "step": 13630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3057726909942054e-05,
+      "loss": 0.7865,
+      "step": 13635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.305288314826688e-05,
+      "loss": 0.8011,
+      "step": 13640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3048037970040515e-05,
+      "loss": 0.6767,
+      "step": 13645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.304319137564318e-05,
+      "loss": 0.9095,
+      "step": 13650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.303834336545514e-05,
+      "loss": 0.6939,
+      "step": 13655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3033493939856816e-05,
+      "loss": 0.7084,
+      "step": 13660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.302864309922874e-05,
+      "loss": 0.8821,
+      "step": 13665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3023790843951515e-05,
+      "loss": 0.7774,
+      "step": 13670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.30189371744059e-05,
+      "loss": 0.8113,
+      "step": 13675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.301408209097274e-05,
+      "loss": 0.7243,
+      "step": 13680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.300922559403301e-05,
+      "loss": 0.7637,
+      "step": 13685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.300436768396776e-05,
+      "loss": 0.8162,
+      "step": 13690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.299950836115819e-05,
+      "loss": 0.7728,
+      "step": 13695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.29946476259856e-05,
+      "loss": 0.7161,
+      "step": 13700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.298978547883138e-05,
+      "loss": 0.7888,
+      "step": 13705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.298492192007707e-05,
+      "loss": 0.8428,
+      "step": 13710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.298005695010427e-05,
+      "loss": 0.7264,
+      "step": 13715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.297519056929474e-05,
+      "loss": 0.6785,
+      "step": 13720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.297032277803032e-05,
+      "loss": 0.7387,
+      "step": 13725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.296545357669297e-05,
+      "loss": 0.8192,
+      "step": 13730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.296058296566476e-05,
+      "loss": 0.682,
+      "step": 13735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2955710945327875e-05,
+      "loss": 0.7063,
+      "step": 13740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2950837516064605e-05,
+      "loss": 0.8201,
+      "step": 13745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2945962678257344e-05,
+      "loss": 0.9147,
+      "step": 13750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.294108643228862e-05,
+      "loss": 0.8736,
+      "step": 13755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.293620877854104e-05,
+      "loss": 0.8974,
+      "step": 13760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.293132971739736e-05,
+      "loss": 0.7415,
+      "step": 13765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.29264492492404e-05,
+      "loss": 0.8989,
+      "step": 13770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.292156737445312e-05,
+      "loss": 0.7945,
+      "step": 13775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2916684093418605e-05,
+      "loss": 0.6267,
+      "step": 13780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.291179940652e-05,
+      "loss": 0.8909,
+      "step": 13785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.290691331414061e-05,
+      "loss": 0.7708,
+      "step": 13790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2902025816663826e-05,
+      "loss": 0.9017,
+      "step": 13795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.289713691447316e-05,
+      "loss": 0.7676,
+      "step": 13800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2892246607952214e-05,
+      "loss": 0.8595,
+      "step": 13805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.288735489748473e-05,
+      "loss": 0.7769,
+      "step": 13810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2882461783454534e-05,
+      "loss": 0.8757,
+      "step": 13815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.287756726624557e-05,
+      "loss": 0.7026,
+      "step": 13820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.287267134624191e-05,
+      "loss": 0.7234,
+      "step": 13825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2867774023827706e-05,
+      "loss": 0.6693,
+      "step": 13830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.286287529938724e-05,
+      "loss": 0.7171,
+      "step": 13835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2857975173304906e-05,
+      "loss": 0.6785,
+      "step": 13840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.28530736459652e-05,
+      "loss": 0.7787,
+      "step": 13845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.284817071775271e-05,
+      "loss": 0.7244,
+      "step": 13850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.284326638905218e-05,
+      "loss": 0.8608,
+      "step": 13855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.283836066024841e-05,
+      "loss": 0.7805,
+      "step": 13860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.283345353172636e-05,
+      "loss": 0.8232,
+      "step": 13865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.282854500387107e-05,
+      "loss": 0.7503,
+      "step": 13870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.282363507706769e-05,
+      "loss": 0.78,
+      "step": 13875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.281872375170148e-05,
+      "loss": 0.7393,
+      "step": 13880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.281381102815784e-05,
+      "loss": 0.674,
+      "step": 13885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.280889690682223e-05,
+      "loss": 0.806,
+      "step": 13890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2803981388080254e-05,
+      "loss": 0.7558,
+      "step": 13895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.279906447231763e-05,
+      "loss": 0.8335,
+      "step": 13900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.279414615992014e-05,
+      "loss": 0.7716,
+      "step": 13905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.278922645127375e-05,
+      "loss": 0.9745,
+      "step": 13910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2784305346764464e-05,
+      "loss": 0.8848,
+      "step": 13915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2779382846778434e-05,
+      "loss": 0.8083,
+      "step": 13920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.277445895170191e-05,
+      "loss": 0.7511,
+      "step": 13925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.276953366192126e-05,
+      "loss": 0.9569,
+      "step": 13930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.276460697782295e-05,
+      "loss": 0.832,
+      "step": 13935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.275967889979356e-05,
+      "loss": 0.7719,
+      "step": 13940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.275474942821978e-05,
+      "loss": 0.8437,
+      "step": 13945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.274981856348842e-05,
+      "loss": 0.8975,
+      "step": 13950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2744886305986376e-05,
+      "loss": 0.77,
+      "step": 13955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.273995265610068e-05,
+      "loss": 0.8018,
+      "step": 13960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2735017614218444e-05,
+      "loss": 0.6188,
+      "step": 13965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.273008118072691e-05,
+      "loss": 0.8172,
+      "step": 13970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.272514335601343e-05,
+      "loss": 0.8165,
+      "step": 13975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.272020414046546e-05,
+      "loss": 0.7788,
+      "step": 13980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2715263534470545e-05,
+      "loss": 0.8823,
+      "step": 13985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.271032153841638e-05,
+      "loss": 0.8042,
+      "step": 13990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2705378152690746e-05,
+      "loss": 0.7676,
+      "step": 13995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2700433377681514e-05,
+      "loss": 0.6615,
+      "step": 14000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2695487213776705e-05,
+      "loss": 0.8033,
+      "step": 14005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.269053966136443e-05,
+      "loss": 0.7983,
+      "step": 14010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.268559072083289e-05,
+      "loss": 0.7349,
+      "step": 14015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2680640392570425e-05,
+      "loss": 0.9242,
+      "step": 14020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.267568867696548e-05,
+      "loss": 0.891,
+      "step": 14025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.267073557440657e-05,
+      "loss": 0.8082,
+      "step": 14030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2665781085282376e-05,
+      "loss": 0.8487,
+      "step": 14035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.266082520998165e-05,
+      "loss": 0.8148,
+      "step": 14040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.265586794889327e-05,
+      "loss": 0.6315,
+      "step": 14045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.26509093024062e-05,
+      "loss": 0.6232,
+      "step": 14050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.264594927090955e-05,
+      "loss": 0.7678,
+      "step": 14055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.26409878547925e-05,
+      "loss": 0.7228,
+      "step": 14060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.263602505444437e-05,
+      "loss": 0.6994,
+      "step": 14065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.263106087025456e-05,
+      "loss": 0.7507,
+      "step": 14070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.262609530261262e-05,
+      "loss": 0.6821,
+      "step": 14075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.262112835190815e-05,
+      "loss": 0.6878,
+      "step": 14080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.26161600185309e-05,
+      "loss": 0.6651,
+      "step": 14085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.261119030287074e-05,
+      "loss": 0.9296,
+      "step": 14090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2606219205317606e-05,
+      "loss": 0.6205,
+      "step": 14095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.260124672626156e-05,
+      "loss": 0.646,
+      "step": 14100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.25962728660928e-05,
+      "loss": 0.7434,
+      "step": 14105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.259129762520159e-05,
+      "loss": 0.8019,
+      "step": 14110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.258632100397831e-05,
+      "loss": 0.6602,
+      "step": 14115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.258134300281349e-05,
+      "loss": 0.6979,
+      "step": 14120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.257636362209772e-05,
+      "loss": 0.7679,
+      "step": 14125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.257138286222172e-05,
+      "loss": 0.6943,
+      "step": 14130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.25664007235763e-05,
+      "loss": 0.6629,
+      "step": 14135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2561417206552415e-05,
+      "loss": 0.8641,
+      "step": 14140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2556432311541095e-05,
+      "loss": 0.7435,
+      "step": 14145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.255144603893348e-05,
+      "loss": 0.7812,
+      "step": 14150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2546458389120846e-05,
+      "loss": 0.7806,
+      "step": 14155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.254146936249455e-05,
+      "loss": 0.7192,
+      "step": 14160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2536478959446046e-05,
+      "loss": 0.8489,
+      "step": 14165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2531487180366934e-05,
+      "loss": 0.9311,
+      "step": 14170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.25264940256489e-05,
+      "loss": 0.6379,
+      "step": 14175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.252149949568374e-05,
+      "loss": 0.8005,
+      "step": 14180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.251650359086336e-05,
+      "loss": 0.8579,
+      "step": 14185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.251150631157977e-05,
+      "loss": 0.7688,
+      "step": 14190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.250650765822509e-05,
+      "loss": 0.8505,
+      "step": 14195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.250150763119155e-05,
+      "loss": 0.7682,
+      "step": 14200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.249650623087148e-05,
+      "loss": 0.6978,
+      "step": 14205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2491503457657335e-05,
+      "loss": 0.7054,
+      "step": 14210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.248649931194165e-05,
+      "loss": 0.8373,
+      "step": 14215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.24814937941171e-05,
+      "loss": 0.784,
+      "step": 14220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.247648690457645e-05,
+      "loss": 0.7168,
+      "step": 14225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.247147864371256e-05,
+      "loss": 0.7629,
+      "step": 14230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.246646901191843e-05,
+      "loss": 0.7039,
+      "step": 14235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.246145800958714e-05,
+      "loss": 0.8141,
+      "step": 14240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.245644563711189e-05,
+      "loss": 0.9522,
+      "step": 14245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.245143189488598e-05,
+      "loss": 0.8157,
+      "step": 14250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.244641678330282e-05,
+      "loss": 0.8473,
+      "step": 14255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2441400302755945e-05,
+      "loss": 0.711,
+      "step": 14260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.243638245363897e-05,
+      "loss": 0.7445,
+      "step": 14265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2431363236345625e-05,
+      "loss": 0.7772,
+      "step": 14270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.242634265126977e-05,
+      "loss": 0.7294,
+      "step": 14275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.242132069880533e-05,
+      "loss": 0.6996,
+      "step": 14280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2416297379346376e-05,
+      "loss": 0.6972,
+      "step": 14285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2411272693287064e-05,
+      "loss": 0.7665,
+      "step": 14290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.240624664102167e-05,
+      "loss": 0.739,
+      "step": 14295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.240121922294459e-05,
+      "loss": 0.7974,
+      "step": 14300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.239619043945027e-05,
+      "loss": 0.9701,
+      "step": 14305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.239116029093333e-05,
+      "loss": 0.9068,
+      "step": 14310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2386128777788465e-05,
+      "loss": 0.947,
+      "step": 14315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.238109590041047e-05,
+      "loss": 0.6394,
+      "step": 14320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.237606165919428e-05,
+      "loss": 0.6713,
+      "step": 14325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2371026054534904e-05,
+      "loss": 0.6775,
+      "step": 14330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2365989086827454e-05,
+      "loss": 0.8097,
+      "step": 14335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.236095075646719e-05,
+      "loss": 0.8135,
+      "step": 14340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.235591106384944e-05,
+      "loss": 0.7608,
+      "step": 14345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2350870009369654e-05,
+      "loss": 0.8754,
+      "step": 14350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.234582759342339e-05,
+      "loss": 0.8942,
+      "step": 14355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.234078381640631e-05,
+      "loss": 0.7907,
+      "step": 14360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.233573867871418e-05,
+      "loss": 0.7991,
+      "step": 14365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.233069218074287e-05,
+      "loss": 0.7618,
+      "step": 14370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.232564432288838e-05,
+      "loss": 0.7194,
+      "step": 14375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.232059510554678e-05,
+      "loss": 0.6555,
+      "step": 14380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.231554452911427e-05,
+      "loss": 0.6782,
+      "step": 14385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.231049259398716e-05,
+      "loss": 0.7596,
+      "step": 14390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.230543930056186e-05,
+      "loss": 0.76,
+      "step": 14395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.230038464923488e-05,
+      "loss": 0.6658,
+      "step": 14400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2295328640402836e-05,
+      "loss": 0.8056,
+      "step": 14405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2290271274462464e-05,
+      "loss": 0.6643,
+      "step": 14410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2285212551810604e-05,
+      "loss": 0.7615,
+      "step": 14415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2280152472844194e-05,
+      "loss": 0.8005,
+      "step": 14420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2275091037960276e-05,
+      "loss": 0.7706,
+      "step": 14425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2270028247556e-05,
+      "loss": 0.7255,
+      "step": 14430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2264964102028646e-05,
+      "loss": 0.8158,
+      "step": 14435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2259898601775567e-05,
+      "loss": 0.7915,
+      "step": 14440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.225483174719424e-05,
+      "loss": 1.0766,
+      "step": 14445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.224976353868224e-05,
+      "loss": 0.8752,
+      "step": 14450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.224469397663726e-05,
+      "loss": 0.7725,
+      "step": 14455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.223962306145709e-05,
+      "loss": 0.7454,
+      "step": 14460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.223455079353963e-05,
+      "loss": 0.7165,
+      "step": 14465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.222947717328287e-05,
+      "loss": 0.7469,
+      "step": 14470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2224402201084945e-05,
+      "loss": 0.8593,
+      "step": 14475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2219325877344054e-05,
+      "loss": 0.7903,
+      "step": 14480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2214248202458524e-05,
+      "loss": 0.8492,
+      "step": 14485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2209169176826785e-05,
+      "loss": 0.6709,
+      "step": 14490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.220408880084737e-05,
+      "loss": 0.7305,
+      "step": 14495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.219900707491892e-05,
+      "loss": 0.6952,
+      "step": 14500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.219392399944018e-05,
+      "loss": 0.7016,
+      "step": 14505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2188839574810014e-05,
+      "loss": 0.9051,
+      "step": 14510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2183753801427364e-05,
+      "loss": 0.8284,
+      "step": 14515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.217866667969129e-05,
+      "loss": 0.7009,
+      "step": 14520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.217357821000099e-05,
+      "loss": 0.7906,
+      "step": 14525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2168488392755715e-05,
+      "loss": 0.7416,
+      "step": 14530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.216339722835486e-05,
+      "loss": 0.8503,
+      "step": 14535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.215830471719789e-05,
+      "loss": 0.8094,
+      "step": 14540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.215321085968443e-05,
+      "loss": 0.7569,
+      "step": 14545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.214811565621416e-05,
+      "loss": 0.7285,
+      "step": 14550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.214301910718688e-05,
+      "loss": 0.7024,
+      "step": 14555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.213792121300252e-05,
+      "loss": 0.6515,
+      "step": 14560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2132821974061064e-05,
+      "loss": 0.7869,
+      "step": 14565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.212772139076266e-05,
+      "loss": 0.7572,
+      "step": 14570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2122619463507516e-05,
+      "loss": 0.8004,
+      "step": 14575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2117516192695986e-05,
+      "loss": 0.8827,
+      "step": 14580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.211241157872848e-05,
+      "loss": 0.8307,
+      "step": 14585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.210730562200557e-05,
+      "loss": 0.7506,
+      "step": 14590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.210219832292787e-05,
+      "loss": 0.7545,
+      "step": 14595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.209708968189615e-05,
+      "loss": 0.7133,
+      "step": 14600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.209197969931128e-05,
+      "loss": 0.8743,
+      "step": 14605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.208686837557421e-05,
+      "loss": 0.7308,
+      "step": 14610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2081755711086014e-05,
+      "loss": 0.8478,
+      "step": 14615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.207664170624786e-05,
+      "loss": 0.9201,
+      "step": 14620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2071526361461034e-05,
+      "loss": 0.7675,
+      "step": 14625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.206640967712691e-05,
+      "loss": 0.6525,
+      "step": 14630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2061291653646996e-05,
+      "loss": 0.8383,
+      "step": 14635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.205617229142287e-05,
+      "loss": 0.791,
+      "step": 14640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.205105159085624e-05,
+      "loss": 0.8871,
+      "step": 14645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2045929552348914e-05,
+      "loss": 0.8507,
+      "step": 14650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2040806176302795e-05,
+      "loss": 0.6648,
+      "step": 14655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.203568146311989e-05,
+      "loss": 0.8827,
+      "step": 14660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.203055541320233e-05,
+      "loss": 0.8537,
+      "step": 14665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.202542802695235e-05,
+      "loss": 0.6253,
+      "step": 14670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.202029930477226e-05,
+      "loss": 0.8452,
+      "step": 14675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2015169247064494e-05,
+      "loss": 0.8675,
+      "step": 14680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.20100378542316e-05,
+      "loss": 0.7464,
+      "step": 14685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2004905126676225e-05,
+      "loss": 0.7281,
+      "step": 14690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.199977106480111e-05,
+      "loss": 0.7011,
+      "step": 14695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.199463566900911e-05,
+      "loss": 0.8669,
+      "step": 14700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1989498939703186e-05,
+      "loss": 0.5759,
+      "step": 14705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.19843608772864e-05,
+      "loss": 0.8512,
+      "step": 14710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.197922148216191e-05,
+      "loss": 0.6725,
+      "step": 14715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1974080754732994e-05,
+      "loss": 0.7492,
+      "step": 14720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1968938695403026e-05,
+      "loss": 0.7745,
+      "step": 14725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1963795304575497e-05,
+      "loss": 0.7565,
+      "step": 14730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1958650582653986e-05,
+      "loss": 0.7457,
+      "step": 14735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.195350453004218e-05,
+      "loss": 0.7437,
+      "step": 14740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.194835714714386e-05,
+      "loss": 0.7795,
+      "step": 14745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.194320843436296e-05,
+      "loss": 0.8125,
+      "step": 14750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.193805839210344e-05,
+      "loss": 0.7636,
+      "step": 14755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.193290702076945e-05,
+      "loss": 0.7778,
+      "step": 14760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1927754320765166e-05,
+      "loss": 0.8949,
+      "step": 14765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.192260029249492e-05,
+      "loss": 0.8191,
+      "step": 14770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.191744493636313e-05,
+      "loss": 0.6705,
+      "step": 14775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1912288252774326e-05,
+      "loss": 0.7548,
+      "step": 14780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.190713024213312e-05,
+      "loss": 0.6672,
+      "step": 14785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.190197090484426e-05,
+      "loss": 0.8458,
+      "step": 14790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.189681024131258e-05,
+      "loss": 0.8437,
+      "step": 14795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1891648251943006e-05,
+      "loss": 0.7822,
+      "step": 14800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.18864849371406e-05,
+      "loss": 0.8932,
+      "step": 14805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.18813202973105e-05,
+      "loss": 0.6989,
+      "step": 14810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.187615433285797e-05,
+      "loss": 0.8448,
+      "step": 14815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.187098704418836e-05,
+      "loss": 0.7325,
+      "step": 14820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1865818431707124e-05,
+      "loss": 0.8346,
+      "step": 14825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.186064849581983e-05,
+      "loss": 0.7526,
+      "step": 14830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.185547723693215e-05,
+      "loss": 0.7775,
+      "step": 14835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1850304655449855e-05,
+      "loss": 0.8915,
+      "step": 14840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1845130751778826e-05,
+      "loss": 0.739,
+      "step": 14845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1839955526325026e-05,
+      "loss": 0.7897,
+      "step": 14850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1834778979494556e-05,
+      "loss": 0.657,
+      "step": 14855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.182960111169359e-05,
+      "loss": 0.6768,
+      "step": 14860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1824421923328427e-05,
+      "loss": 0.7218,
+      "step": 14865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.181924141480545e-05,
+      "loss": 0.815,
+      "step": 14870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1814059586531174e-05,
+      "loss": 0.7547,
+      "step": 14875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.180887643891218e-05,
+      "loss": 0.7784,
+      "step": 14880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1803691972355195e-05,
+      "loss": 0.6969,
+      "step": 14885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1798506187267004e-05,
+      "loss": 0.8073,
+      "step": 14890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.179331908405454e-05,
+      "loss": 0.8073,
+      "step": 14895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1788130663124804e-05,
+      "loss": 0.6123,
+      "step": 14900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.178294092488492e-05,
+      "loss": 0.8278,
+      "step": 14905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.177774986974211e-05,
+      "loss": 0.8543,
+      "step": 14910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.177255749810369e-05,
+      "loss": 0.8314,
+      "step": 14915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.176736381037712e-05,
+      "loss": 0.8078,
+      "step": 14920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.176216880696988e-05,
+      "loss": 0.7023,
+      "step": 14925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1756972488289656e-05,
+      "loss": 0.7758,
+      "step": 14930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.175177485474415e-05,
+      "loss": 0.8373,
+      "step": 14935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.174657590674122e-05,
+      "loss": 0.8394,
+      "step": 14940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.174137564468881e-05,
+      "loss": 0.766,
+      "step": 14945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.173617406899496e-05,
+      "loss": 0.8331,
+      "step": 14950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.173097118006783e-05,
+      "loss": 0.8118,
+      "step": 14955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1725766978315675e-05,
+      "loss": 0.7548,
+      "step": 14960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.172056146414684e-05,
+      "loss": 0.7587,
+      "step": 14965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.17153546379698e-05,
+      "loss": 0.7882,
+      "step": 14970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1710146500193106e-05,
+      "loss": 0.8631,
+      "step": 14975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.170493705122543e-05,
+      "loss": 0.6468,
+      "step": 14980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1699726291475524e-05,
+      "loss": 0.7515,
+      "step": 14985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.169451422135229e-05,
+      "loss": 0.9514,
+      "step": 14990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.168930084126468e-05,
+      "loss": 0.7948,
+      "step": 14995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.168408615162178e-05,
+      "loss": 0.8938,
+      "step": 15000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.167887015283276e-05,
+      "loss": 0.6644,
+      "step": 15005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.167365284530691e-05,
+      "loss": 0.784,
+      "step": 15010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.166843422945362e-05,
+      "loss": 0.7293,
+      "step": 15015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.166321430568236e-05,
+      "loss": 0.8173,
+      "step": 15020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1657993074402745e-05,
+      "loss": 0.6368,
+      "step": 15025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1652770536024445e-05,
+      "loss": 0.7605,
+      "step": 15030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.164754669095727e-05,
+      "loss": 0.8674,
+      "step": 15035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.164232153961112e-05,
+      "loss": 0.8138,
+      "step": 15040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1637095082395985e-05,
+      "loss": 0.6695,
+      "step": 15045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.163186731972197e-05,
+      "loss": 0.7654,
+      "step": 15050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.162663825199929e-05,
+      "loss": 0.7337,
+      "step": 15055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.162140787963824e-05,
+      "loss": 0.7798,
+      "step": 15060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.161617620304924e-05,
+      "loss": 0.7638,
+      "step": 15065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.16109432226428e-05,
+      "loss": 0.8025,
+      "step": 15070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1605708938829535e-05,
+      "loss": 0.714,
+      "step": 15075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1600473352020166e-05,
+      "loss": 0.7884,
+      "step": 15080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.15952364626255e-05,
+      "loss": 0.7055,
+      "step": 15085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1589998271056473e-05,
+      "loss": 0.7577,
+      "step": 15090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.15847587777241e-05,
+      "loss": 0.6966,
+      "step": 15095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1579517983039514e-05,
+      "loss": 0.7704,
+      "step": 15100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.157427588741394e-05,
+      "loss": 0.8016,
+      "step": 15105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1569032491258695e-05,
+      "loss": 0.6717,
+      "step": 15110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.156378779498524e-05,
+      "loss": 0.804,
+      "step": 15115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.155854179900508e-05,
+      "loss": 0.7728,
+      "step": 15120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1553294503729875e-05,
+      "loss": 0.745,
+      "step": 15125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1548045909571354e-05,
+      "loss": 0.8904,
+      "step": 15130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1542796016941344e-05,
+      "loss": 0.6619,
+      "step": 15135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.153754482625181e-05,
+      "loss": 0.6701,
+      "step": 15140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1532292337914775e-05,
+      "loss": 0.6635,
+      "step": 15145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1527038552342394e-05,
+      "loss": 0.6441,
+      "step": 15150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.152178346994692e-05,
+      "loss": 0.6902,
+      "step": 15155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.15165270911407e-05,
+      "loss": 0.8804,
+      "step": 15160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.151126941633619e-05,
+      "loss": 0.9651,
+      "step": 15165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.150601044594591e-05,
+      "loss": 0.8558,
+      "step": 15170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1500750180382555e-05,
+      "loss": 0.8583,
+      "step": 15175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1495488620058865e-05,
+      "loss": 0.9248,
+      "step": 15180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.149022576538769e-05,
+      "loss": 0.6452,
+      "step": 15185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1484961616782016e-05,
+      "loss": 0.746,
+      "step": 15190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.147969617465487e-05,
+      "loss": 0.7649,
+      "step": 15195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1474429439419426e-05,
+      "loss": 0.8337,
+      "step": 15200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.146916141148896e-05,
+      "loss": 0.8712,
+      "step": 15205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.146389209127682e-05,
+      "loss": 0.6799,
+      "step": 15210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.145862147919648e-05,
+      "loss": 0.6875,
+      "step": 15215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.145334957566151e-05,
+      "loss": 0.8359,
+      "step": 15220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.144807638108558e-05,
+      "loss": 0.6971,
+      "step": 15225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1442801895882454e-05,
+      "loss": 0.8152,
+      "step": 15230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.143752612046601e-05,
+      "loss": 0.7915,
+      "step": 15235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.143224905525021e-05,
+      "loss": 0.8422,
+      "step": 15240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1426970700649147e-05,
+      "loss": 0.6184,
+      "step": 15245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1421691057076975e-05,
+      "loss": 0.7222,
+      "step": 15250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.141641012494799e-05,
+      "loss": 0.8185,
+      "step": 15255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1411127904676556e-05,
+      "loss": 0.7646,
+      "step": 15260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1405844396677153e-05,
+      "loss": 0.856,
+      "step": 15265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.140055960136437e-05,
+      "loss": 0.774,
+      "step": 15270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.139527351915288e-05,
+      "loss": 0.8261,
+      "step": 15275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.138998615045747e-05,
+      "loss": 0.6593,
+      "step": 15280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1384697495693014e-05,
+      "loss": 0.7459,
+      "step": 15285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1379407555274507e-05,
+      "loss": 0.739,
+      "step": 15290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.137411632961702e-05,
+      "loss": 0.7252,
+      "step": 15295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.136882381913575e-05,
+      "loss": 0.7608,
+      "step": 15300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1363530024245986e-05,
+      "loss": 0.8032,
+      "step": 15305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.13582349453631e-05,
+      "loss": 0.8075,
+      "step": 15310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.135293858290258e-05,
+      "loss": 0.8385,
+      "step": 15315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.134764093728003e-05,
+      "loss": 0.9119,
+      "step": 15320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1342342008911126e-05,
+      "loss": 0.8657,
+      "step": 15325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1337041798211675e-05,
+      "loss": 0.7458,
+      "step": 15330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1331740305597546e-05,
+      "loss": 0.8361,
+      "step": 15335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1326437531484734e-05,
+      "loss": 0.9129,
+      "step": 15340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.132113347628934e-05,
+      "loss": 0.8046,
+      "step": 15345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.131582814042755e-05,
+      "loss": 0.7866,
+      "step": 15350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.131052152431566e-05,
+      "loss": 0.6996,
+      "step": 15355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1305213628370065e-05,
+      "loss": 0.7152,
+      "step": 15360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1299904453007245e-05,
+      "loss": 0.7719,
+      "step": 15365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1294593998643805e-05,
+      "loss": 0.7073,
+      "step": 15370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1289282265696436e-05,
+      "loss": 0.635,
+      "step": 15375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.128396925458194e-05,
+      "loss": 0.7967,
+      "step": 15380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.12786549657172e-05,
+      "loss": 0.8185,
+      "step": 15385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.127333939951922e-05,
+      "loss": 0.7898,
+      "step": 15390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1268022556405086e-05,
+      "loss": 0.8471,
+      "step": 15395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1262704436792006e-05,
+      "loss": 0.7575,
+      "step": 15400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.125738504109726e-05,
+      "loss": 0.7044,
+      "step": 15405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1252064369738256e-05,
+      "loss": 0.7422,
+      "step": 15410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.124674242313249e-05,
+      "loss": 0.8324,
+      "step": 15415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.124141920169755e-05,
+      "loss": 0.7179,
+      "step": 15420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1236094705851136e-05,
+      "loss": 0.986,
+      "step": 15425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1230768936011045e-05,
+      "loss": 0.6335,
+      "step": 15430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.122544189259517e-05,
+      "loss": 0.8705,
+      "step": 15435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.122011357602151e-05,
+      "loss": 0.7458,
+      "step": 15440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1214783986708156e-05,
+      "loss": 0.7942,
+      "step": 15445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.12094531250733e-05,
+      "loss": 0.7531,
+      "step": 15450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.120412099153525e-05,
+      "loss": 0.6637,
+      "step": 15455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.119878758651241e-05,
+      "loss": 0.8635,
+      "step": 15460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1193452910423246e-05,
+      "loss": 0.8071,
+      "step": 15465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.118811696368637e-05,
+      "loss": 0.6934,
+      "step": 15470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.118277974672047e-05,
+      "loss": 0.7153,
+      "step": 15475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.117744125994435e-05,
+      "loss": 0.7298,
+      "step": 15480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.117210150377689e-05,
+      "loss": 0.7669,
+      "step": 15485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.116676047863709e-05,
+      "loss": 0.6456,
+      "step": 15490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.116141818494406e-05,
+      "loss": 0.9272,
+      "step": 15495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.115607462311696e-05,
+      "loss": 0.7576,
+      "step": 15500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1150729793575104e-05,
+      "loss": 0.7664,
+      "step": 15505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.114538369673787e-05,
+      "loss": 0.7221,
+      "step": 15510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.114003633302476e-05,
+      "loss": 0.6396,
+      "step": 15515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1134687702855365e-05,
+      "loss": 0.766,
+      "step": 15520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1129337806649365e-05,
+      "loss": 0.8143,
+      "step": 15525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.112398664482656e-05,
+      "loss": 0.7454,
+      "step": 15530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.111863421780683e-05,
+      "loss": 0.7416,
+      "step": 15535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.111328052601017e-05,
+      "loss": 0.6742,
+      "step": 15540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.110792556985666e-05,
+      "loss": 0.8823,
+      "step": 15545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.110256934976647e-05,
+      "loss": 0.8482,
+      "step": 15550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.109721186615992e-05,
+      "loss": 0.692,
+      "step": 15555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.109185311945738e-05,
+      "loss": 0.7132,
+      "step": 15560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1086493110079326e-05,
+      "loss": 0.8627,
+      "step": 15565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.108113183844634e-05,
+      "loss": 0.8395,
+      "step": 15570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.107576930497912e-05,
+      "loss": 0.8132,
+      "step": 15575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.107040551009843e-05,
+      "loss": 0.8542,
+      "step": 15580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.106504045422515e-05,
+      "loss": 0.7814,
+      "step": 15585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1059674137780275e-05,
+      "loss": 0.8744,
+      "step": 15590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.105430656118486e-05,
+      "loss": 0.7,
+      "step": 15595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.104893772486011e-05,
+      "loss": 0.6878,
+      "step": 15600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1043567629227265e-05,
+      "loss": 0.807,
+      "step": 15605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.103819627470772e-05,
+      "loss": 0.723,
+      "step": 15610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.103282366172295e-05,
+      "loss": 0.6832,
+      "step": 15615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.102744979069452e-05,
+      "loss": 0.8209,
+      "step": 15620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.10220746620441e-05,
+      "loss": 0.8223,
+      "step": 15625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.101669827619346e-05,
+      "loss": 0.8338,
+      "step": 15630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.101132063356447e-05,
+      "loss": 0.7741,
+      "step": 15635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.100594173457909e-05,
+      "loss": 0.8217,
+      "step": 15640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1000561579659396e-05,
+      "loss": 0.7858,
+      "step": 15645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0995180169227536e-05,
+      "loss": 0.7079,
+      "step": 15650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0989797503705784e-05,
+      "loss": 0.8236,
+      "step": 15655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.098441358351649e-05,
+      "loss": 0.7493,
+      "step": 15660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.097902840908213e-05,
+      "loss": 0.6868,
+      "step": 15665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.097364198082524e-05,
+      "loss": 0.8257,
+      "step": 15670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.096825429916849e-05,
+      "loss": 0.8279,
+      "step": 15675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.096286536453463e-05,
+      "loss": 0.7511,
+      "step": 15680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.095747517734651e-05,
+      "loss": 0.8088,
+      "step": 15685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.095208373802708e-05,
+      "loss": 0.8327,
+      "step": 15690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.09466910469994e-05,
+      "loss": 0.708,
+      "step": 15695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0941297104686597e-05,
+      "loss": 0.9236,
+      "step": 15700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.093590191151193e-05,
+      "loss": 0.9189,
+      "step": 15705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.093050546789874e-05,
+      "loss": 0.6486,
+      "step": 15710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.092510777427048e-05,
+      "loss": 0.7899,
+      "step": 15715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.091970883105066e-05,
+      "loss": 0.6896,
+      "step": 15720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0914308638662935e-05,
+      "loss": 0.8377,
+      "step": 15725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0908907197531054e-05,
+      "loss": 0.7677,
+      "step": 15730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0903504508078825e-05,
+      "loss": 0.8673,
+      "step": 15735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.08981005707302e-05,
+      "loss": 0.7684,
+      "step": 15740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0892695385909195e-05,
+      "loss": 0.743,
+      "step": 15745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0887288954039945e-05,
+      "loss": 0.6505,
+      "step": 15750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.088188127554666e-05,
+      "loss": 0.7623,
+      "step": 15755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0876472350853695e-05,
+      "loss": 0.8144,
+      "step": 15760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.087106218038544e-05,
+      "loss": 0.8852,
+      "step": 15765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.086565076456643e-05,
+      "loss": 0.7938,
+      "step": 15770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.086023810382127e-05,
+      "loss": 0.8009,
+      "step": 15775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0854824198574684e-05,
+      "loss": 0.6205,
+      "step": 15780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0849409049251476e-05,
+      "loss": 0.6615,
+      "step": 15785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.084399265627656e-05,
+      "loss": 0.9059,
+      "step": 15790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.083857502007494e-05,
+      "loss": 0.7041,
+      "step": 15795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0833156141071725e-05,
+      "loss": 0.7699,
+      "step": 15800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.082773601969212e-05,
+      "loss": 0.7804,
+      "step": 15805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.08223146563614e-05,
+      "loss": 0.8664,
+      "step": 15810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0816892051504994e-05,
+      "loss": 0.797,
+      "step": 15815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.081146820554839e-05,
+      "loss": 0.9085,
+      "step": 15820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.080604311891716e-05,
+      "loss": 0.6532,
+      "step": 15825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0800616792037e-05,
+      "loss": 0.6988,
+      "step": 15830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.079518922533371e-05,
+      "loss": 0.9728,
+      "step": 15835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.078976041923316e-05,
+      "loss": 0.8549,
+      "step": 15840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.078433037416133e-05,
+      "loss": 0.9004,
+      "step": 15845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0778899090544306e-05,
+      "loss": 0.7381,
+      "step": 15850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0773466568808264e-05,
+      "loss": 0.7542,
+      "step": 15855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0768032809379474e-05,
+      "loss": 0.6511,
+      "step": 15860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.07625978126843e-05,
+      "loss": 0.8358,
+      "step": 15865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.075716157914922e-05,
+      "loss": 0.8453,
+      "step": 15870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0751724109200786e-05,
+      "loss": 0.6953,
+      "step": 15875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.074628540326566e-05,
+      "loss": 0.8037,
+      "step": 15880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.074084546177061e-05,
+      "loss": 0.6102,
+      "step": 15885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.073540428514247e-05,
+      "loss": 0.7115,
+      "step": 15890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0729961873808206e-05,
+      "loss": 0.7593,
+      "step": 15895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.072451822819487e-05,
+      "loss": 0.7675,
+      "step": 15900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.071907334872961e-05,
+      "loss": 0.7334,
+      "step": 15905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.071362723583966e-05,
+      "loss": 0.7047,
+      "step": 15910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0708179889952344e-05,
+      "loss": 0.7953,
+      "step": 15915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0702731311495124e-05,
+      "loss": 0.7965,
+      "step": 15920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.069728150089552e-05,
+      "loss": 0.7737,
+      "step": 15925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.069183045858116e-05,
+      "loss": 0.6713,
+      "step": 15930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0686378184979775e-05,
+      "loss": 0.7556,
+      "step": 15935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.068092468051918e-05,
+      "loss": 0.8424,
+      "step": 15940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.06754699456273e-05,
+      "loss": 0.6981,
+      "step": 15945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.067001398073214e-05,
+      "loss": 0.8353,
+      "step": 15950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.066455678626183e-05,
+      "loss": 0.8526,
+      "step": 15955
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0659098362644566e-05,
+      "loss": 0.8427,
+      "step": 15960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0653638710308654e-05,
+      "loss": 0.8564,
+      "step": 15965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0648177829682486e-05,
+      "loss": 0.6733,
+      "step": 15970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.064271572119458e-05,
+      "loss": 0.7721,
+      "step": 15975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.063725238527352e-05,
+      "loss": 0.7879,
+      "step": 15980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0631787822347985e-05,
+      "loss": 0.8468,
+      "step": 15985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0626322032846794e-05,
+      "loss": 0.6828,
+      "step": 15990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.062085501719879e-05,
+      "loss": 0.6688,
+      "step": 15995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0615386775832976e-05,
+      "loss": 0.8081,
+      "step": 16000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.060991730917842e-05,
+      "loss": 0.7331,
+      "step": 16005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.060444661766429e-05,
+      "loss": 0.9095,
+      "step": 16010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.059897470171987e-05,
+      "loss": 0.7211,
+      "step": 16015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0593501561774503e-05,
+      "loss": 0.9216,
+      "step": 16020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.058802719825766e-05,
+      "loss": 0.7649,
+      "step": 16025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.058255161159889e-05,
+      "loss": 0.9022,
+      "step": 16030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.057707480222785e-05,
+      "loss": 0.6374,
+      "step": 16035
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0571596770574284e-05,
+      "loss": 0.7034,
+      "step": 16040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.056611751706804e-05,
+      "loss": 0.7342,
+      "step": 16045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0560637042139056e-05,
+      "loss": 0.9441,
+      "step": 16050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.055515534621736e-05,
+      "loss": 0.7188,
+      "step": 16055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0549672429733085e-05,
+      "loss": 0.7402,
+      "step": 16060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.054418829311647e-05,
+      "loss": 0.7769,
+      "step": 16065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0538702936797824e-05,
+      "loss": 0.6366,
+      "step": 16070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.053321636120757e-05,
+      "loss": 0.7457,
+      "step": 16075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0527728566776225e-05,
+      "loss": 0.7776,
+      "step": 16080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.052223955393439e-05,
+      "loss": 0.71,
+      "step": 16085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.051674932311277e-05,
+      "loss": 0.6044,
+      "step": 16090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0511257874742175e-05,
+      "loss": 0.7029,
+      "step": 16095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.05057652092535e-05,
+      "loss": 0.8178,
+      "step": 16100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.050027132707773e-05,
+      "loss": 0.7541,
+      "step": 16105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.049477622864595e-05,
+      "loss": 0.7106,
+      "step": 16110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0489279914389354e-05,
+      "loss": 0.7993,
+      "step": 16115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.048378238473921e-05,
+      "loss": 0.6655,
+      "step": 16120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0478283640126897e-05,
+      "loss": 0.6923,
+      "step": 16125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0472783680983885e-05,
+      "loss": 0.7375,
+      "step": 16130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0467282507741725e-05,
+      "loss": 0.7332,
+      "step": 16135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0461780120832094e-05,
+      "loss": 0.8744,
+      "step": 16140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.045627652068673e-05,
+      "loss": 0.8498,
+      "step": 16145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.04507717077375e-05,
+      "loss": 0.8059,
+      "step": 16150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.044526568241633e-05,
+      "loss": 0.8051,
+      "step": 16155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.043975844515528e-05,
+      "loss": 0.7595,
+      "step": 16160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.043424999638647e-05,
+      "loss": 0.8861,
+      "step": 16165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.042874033654214e-05,
+      "loss": 0.655,
+      "step": 16170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.04232294660546e-05,
+      "loss": 0.7882,
+      "step": 16175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.041771738535628e-05,
+      "loss": 0.6727,
+      "step": 16180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.04122040948797e-05,
+      "loss": 0.8422,
+      "step": 16185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.040668959505747e-05,
+      "loss": 0.7551,
+      "step": 16190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.040117388632228e-05,
+      "loss": 0.8242,
+      "step": 16195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0395656969106946e-05,
+      "loss": 0.7644,
+      "step": 16200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.039013884384435e-05,
+      "loss": 0.6685,
+      "step": 16205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0384619510967494e-05,
+      "loss": 0.7384,
+      "step": 16210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.037909897090946e-05,
+      "loss": 0.7545,
+      "step": 16215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.037357722410341e-05,
+      "loss": 0.8711,
+      "step": 16220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0368054270982636e-05,
+      "loss": 0.7537,
+      "step": 16225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.03625301119805e-05,
+      "loss": 0.7159,
+      "step": 16230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.035700474753047e-05,
+      "loss": 0.8496,
+      "step": 16235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.035147817806609e-05,
+      "loss": 0.7534,
+      "step": 16240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.034595040402104e-05,
+      "loss": 0.7851,
+      "step": 16245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.034042142582903e-05,
+      "loss": 0.7378,
+      "step": 16250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.033489124392392e-05,
+      "loss": 0.8197,
+      "step": 16255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.032935985873965e-05,
+      "loss": 0.8229,
+      "step": 16260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0323827270710235e-05,
+      "loss": 0.8895,
+      "step": 16265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0318293480269806e-05,
+      "loss": 0.7482,
+      "step": 16270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.03127584878526e-05,
+      "loss": 0.9005,
+      "step": 16275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.03072222938929e-05,
+      "loss": 0.7047,
+      "step": 16280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.030168489882512e-05,
+      "loss": 0.8361,
+      "step": 16285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.029614630308378e-05,
+      "loss": 0.8218,
+      "step": 16290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.029060650710346e-05,
+      "loss": 0.6176,
+      "step": 16295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0285065511318854e-05,
+      "loss": 0.7183,
+      "step": 16300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.027952331616475e-05,
+      "loss": 0.8603,
+      "step": 16305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.027397992207601e-05,
+      "loss": 0.7843,
+      "step": 16310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.026843532948762e-05,
+      "loss": 0.6523,
+      "step": 16315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.026288953883465e-05,
+      "loss": 0.7879,
+      "step": 16320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.025734255055225e-05,
+      "loss": 0.9354,
+      "step": 16325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0251794365075676e-05,
+      "loss": 0.6322,
+      "step": 16330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.024624498284029e-05,
+      "loss": 0.9286,
+      "step": 16335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.024069440428151e-05,
+      "loss": 0.9716,
+      "step": 16340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.023514262983489e-05,
+      "loss": 0.9005,
+      "step": 16345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.022958965993605e-05,
+      "loss": 0.7018,
+      "step": 16350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.022403549502072e-05,
+      "loss": 0.8839,
+      "step": 16355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.021848013552471e-05,
+      "loss": 0.7201,
+      "step": 16360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0212923581883946e-05,
+      "loss": 0.7888,
+      "step": 16365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.020736583453441e-05,
+      "loss": 0.8112,
+      "step": 16370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.020180689391222e-05,
+      "loss": 0.7379,
+      "step": 16375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0196246760453555e-05,
+      "loss": 0.8324,
+      "step": 16380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.019068543459471e-05,
+      "loss": 0.7707,
+      "step": 16385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0185122916772066e-05,
+      "loss": 0.9081,
+      "step": 16390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.017955920742208e-05,
+      "loss": 0.789,
+      "step": 16395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.017399430698133e-05,
+      "loss": 0.7814,
+      "step": 16400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.016842821588648e-05,
+      "loss": 0.8245,
+      "step": 16405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0162860934574275e-05,
+      "loss": 0.7647,
+      "step": 16410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.015729246348157e-05,
+      "loss": 0.8532,
+      "step": 16415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.01517228030453e-05,
+      "loss": 0.7159,
+      "step": 16420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.014615195370248e-05,
+      "loss": 0.7297,
+      "step": 16425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0140579915890264e-05,
+      "loss": 0.74,
+      "step": 16430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0135006690045864e-05,
+      "loss": 0.763,
+      "step": 16435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.012943227660659e-05,
+      "loss": 0.7939,
+      "step": 16440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.012385667600985e-05,
+      "loss": 0.8639,
+      "step": 16445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.011827988869313e-05,
+      "loss": 0.8524,
+      "step": 16450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.011270191509404e-05,
+      "loss": 0.6646,
+      "step": 16455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0107122755650264e-05,
+      "loss": 0.6719,
+      "step": 16460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.010154241079957e-05,
+      "loss": 0.7728,
+      "step": 16465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.009596088097985e-05,
+      "loss": 0.6417,
+      "step": 16470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.009037816662904e-05,
+      "loss": 0.5898,
+      "step": 16475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.008479426818521e-05,
+      "loss": 0.9297,
+      "step": 16480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.007920918608652e-05,
+      "loss": 0.7191,
+      "step": 16485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.007362292077119e-05,
+      "loss": 0.8802,
+      "step": 16490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.006803547267759e-05,
+      "loss": 0.7134,
+      "step": 16495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.006244684224412e-05,
+      "loss": 0.8336,
+      "step": 16500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.005685702990932e-05,
+      "loss": 0.6985,
+      "step": 16505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.005126603611179e-05,
+      "loss": 0.7269,
+      "step": 16510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.004567386129025e-05,
+      "loss": 0.7239,
+      "step": 16515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0040080505883484e-05,
+      "loss": 0.8284,
+      "step": 16520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0034485970330394e-05,
+      "loss": 0.7905,
+      "step": 16525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.002889025506997e-05,
+      "loss": 0.8063,
+      "step": 16530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.002329336054128e-05,
+      "loss": 0.7231,
+      "step": 16535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.001769528718351e-05,
+      "loss": 0.8226,
+      "step": 16540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.00120960354359e-05,
+      "loss": 0.7639,
+      "step": 16545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.0006495605737815e-05,
+      "loss": 0.8261,
+      "step": 16550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.000089399852871e-05,
+      "loss": 0.864,
+      "step": 16555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.999529121424812e-05,
+      "loss": 0.8983,
+      "step": 16560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9989687253335674e-05,
+      "loss": 0.7784,
+      "step": 16565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.99840821162311e-05,
+      "loss": 0.8629,
+      "step": 16570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9978475803374215e-05,
+      "loss": 0.7185,
+      "step": 16575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9972868315204924e-05,
+      "loss": 0.7348,
+      "step": 16580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.996725965216323e-05,
+      "loss": 0.7805,
+      "step": 16585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.996164981468923e-05,
+      "loss": 0.8034,
+      "step": 16590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9956038803223115e-05,
+      "loss": 0.7638,
+      "step": 16595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.995042661820515e-05,
+      "loss": 0.7141,
+      "step": 16600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9944813260075706e-05,
+      "loss": 0.8057,
+      "step": 16605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.993919872927525e-05,
+      "loss": 0.6932,
+      "step": 16610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9933583026244333e-05,
+      "loss": 0.7585,
+      "step": 16615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.992796615142362e-05,
+      "loss": 0.7144,
+      "step": 16620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.992234810525381e-05,
+      "loss": 0.8344,
+      "step": 16625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.991672888817578e-05,
+      "loss": 0.7852,
+      "step": 16630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.991110850063041e-05,
+      "loss": 0.6898,
+      "step": 16635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9905486943058736e-05,
+      "loss": 0.7883,
+      "step": 16640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.989986421590185e-05,
+      "loss": 0.8072,
+      "step": 16645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9894240319600965e-05,
+      "loss": 0.7579,
+      "step": 16650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.988861525459736e-05,
+      "loss": 0.6968,
+      "step": 16655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.988298902133242e-05,
+      "loss": 0.8827,
+      "step": 16660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.987736162024762e-05,
+      "loss": 0.7285,
+      "step": 16665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.987173305178451e-05,
+      "loss": 0.8232,
+      "step": 16670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.986610331638476e-05,
+      "loss": 0.7227,
+      "step": 16675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.98604724144901e-05,
+      "loss": 0.8194,
+      "step": 16680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9854840346542395e-05,
+      "loss": 0.7903,
+      "step": 16685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9849207112983553e-05,
+      "loss": 0.7442,
+      "step": 16690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.98435727142556e-05,
+      "loss": 0.8516,
+      "step": 16695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.983793715080066e-05,
+      "loss": 0.6535,
+      "step": 16700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.983230042306093e-05,
+      "loss": 0.661,
+      "step": 16705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9826662531478696e-05,
+      "loss": 0.6785,
+      "step": 16710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.982102347649636e-05,
+      "loss": 0.7567,
+      "step": 16715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9815383258556385e-05,
+      "loss": 0.6886,
+      "step": 16720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.980974187810136e-05,
+      "loss": 0.805,
+      "step": 16725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.980409933557393e-05,
+      "loss": 0.8739,
+      "step": 16730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9798455631416854e-05,
+      "loss": 0.737,
+      "step": 16735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.979281076607297e-05,
+      "loss": 0.8015,
+      "step": 16740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.978716473998523e-05,
+      "loss": 0.7946,
+      "step": 16745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.978151755359663e-05,
+      "loss": 0.7053,
+      "step": 16750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.977586920735031e-05,
+      "loss": 0.8706,
+      "step": 16755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.977021970168947e-05,
+      "loss": 0.8629,
+      "step": 16760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.976456903705741e-05,
+      "loss": 0.7652,
+      "step": 16765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9758917213897506e-05,
+      "loss": 0.8781,
+      "step": 16770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.975326423265325e-05,
+      "loss": 0.6837,
+      "step": 16775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.974761009376822e-05,
+      "loss": 0.7895,
+      "step": 16780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.974195479768607e-05,
+      "loss": 0.7482,
+      "step": 16785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9736298344850554e-05,
+      "loss": 0.7473,
+      "step": 16790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.973064073570551e-05,
+      "loss": 0.8753,
+      "step": 16795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9724981970694883e-05,
+      "loss": 0.8102,
+      "step": 16800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.971932205026268e-05,
+      "loss": 0.8776,
+      "step": 16805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.971366097485304e-05,
+      "loss": 0.8675,
+      "step": 16810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.970799874491014e-05,
+      "loss": 0.8738,
+      "step": 16815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9702335360878316e-05,
+      "loss": 0.8247,
+      "step": 16820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.969667082320193e-05,
+      "loss": 0.7889,
+      "step": 16825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9691005132325456e-05,
+      "loss": 0.7191,
+      "step": 16830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9685338288693475e-05,
+      "loss": 0.9174,
+      "step": 16835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.967967029275063e-05,
+      "loss": 0.732,
+      "step": 16840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.96740011449417e-05,
+      "loss": 0.6945,
+      "step": 16845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.966833084571149e-05,
+      "loss": 0.7886,
+      "step": 16850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.966265939550495e-05,
+      "loss": 0.7362,
+      "step": 16855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.965698679476709e-05,
+      "loss": 0.8332,
+      "step": 16860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.965131304394304e-05,
+      "loss": 0.7917,
+      "step": 16865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.964563814347798e-05,
+      "loss": 0.7757,
+      "step": 16870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.963996209381721e-05,
+      "loss": 0.7305,
+      "step": 16875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.96342848954061e-05,
+      "loss": 0.6734,
+      "step": 16880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962860654869014e-05,
+      "loss": 0.7232,
+      "step": 16885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9622927054114885e-05,
+      "loss": 0.8664,
+      "step": 16890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.961724641212598e-05,
+      "loss": 0.7922,
+      "step": 16895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9611564623169166e-05,
+      "loss": 0.7069,
+      "step": 16900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.960588168769029e-05,
+      "loss": 0.8122,
+      "step": 16905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.960019760613525e-05,
+      "loss": 0.6678,
+      "step": 16910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.959451237895008e-05,
+      "loss": 0.7489,
+      "step": 16915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9588826006580864e-05,
+      "loss": 0.7602,
+      "step": 16920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9583138489473806e-05,
+      "loss": 0.7805,
+      "step": 16925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9577449828075177e-05,
+      "loss": 0.6593,
+      "step": 16930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.957176002283136e-05,
+      "loss": 0.774,
+      "step": 16935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.95660690741888e-05,
+      "loss": 0.8005,
+      "step": 16940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9560376982594054e-05,
+      "loss": 0.6867,
+      "step": 16945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.955468374849377e-05,
+      "loss": 0.9174,
+      "step": 16950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9548989372334664e-05,
+      "loss": 0.7258,
+      "step": 16955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.954329385456357e-05,
+      "loss": 0.7968,
+      "step": 16960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.953759719562738e-05,
+      "loss": 0.7154,
+      "step": 16965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.953189939597311e-05,
+      "loss": 0.9012,
+      "step": 16970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9526200456047825e-05,
+      "loss": 0.8471,
+      "step": 16975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.952050037629873e-05,
+      "loss": 0.7506,
+      "step": 16980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.951479915717307e-05,
+      "loss": 0.7195,
+      "step": 16985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.950909679911822e-05,
+      "loss": 0.8235,
+      "step": 16990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.95033933025816e-05,
+      "loss": 0.7116,
+      "step": 16995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9497688668010765e-05,
+      "loss": 0.7941,
+      "step": 17000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9491982895853336e-05,
+      "loss": 0.8068,
+      "step": 17005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.948627598655702e-05,
+      "loss": 0.7387,
+      "step": 17010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.948056794056963e-05,
+      "loss": 0.6297,
+      "step": 17015
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.947485875833905e-05,
+      "loss": 0.7021,
+      "step": 17020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.946914844031326e-05,
+      "loss": 0.6078,
+      "step": 17025
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.946343698694034e-05,
+      "loss": 0.8473,
+      "step": 17030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.945772439866843e-05,
+      "loss": 0.7613,
+      "step": 17035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.945201067594579e-05,
+      "loss": 0.8674,
+      "step": 17040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.944629581922077e-05,
+      "loss": 0.8318,
+      "step": 17045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.944057982894178e-05,
+      "loss": 0.8046,
+      "step": 17050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.943486270555734e-05,
+      "loss": 0.6783,
+      "step": 17055
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.942914444951604e-05,
+      "loss": 0.745,
+      "step": 17060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9423425061266606e-05,
+      "loss": 0.7864,
+      "step": 17065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9417704541257785e-05,
+      "loss": 0.693,
+      "step": 17070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.941198288993847e-05,
+      "loss": 0.7757,
+      "step": 17075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.940626010775761e-05,
+      "loss": 0.7796,
+      "step": 17080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.940053619516426e-05,
+      "loss": 0.5972,
+      "step": 17085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.939481115260755e-05,
+      "loss": 0.7691,
+      "step": 17090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9389084980536705e-05,
+      "loss": 0.8229,
+      "step": 17095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.938335767940105e-05,
+      "loss": 0.6784,
+      "step": 17100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9377629249649985e-05,
+      "loss": 0.6522,
+      "step": 17105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9371899691732986e-05,
+      "loss": 0.8549,
+      "step": 17110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.936616900609964e-05,
+      "loss": 0.7314,
+      "step": 17115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.936043719319963e-05,
+      "loss": 0.6969,
+      "step": 17120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9354704253482696e-05,
+      "loss": 0.6828,
+      "step": 17125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9348970187398684e-05,
+      "loss": 0.9116,
+      "step": 17130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.934323499539755e-05,
+      "loss": 0.791,
+      "step": 17135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9337498677929286e-05,
+      "loss": 0.742,
+      "step": 17140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.933176123544401e-05,
+      "loss": 0.8269,
+      "step": 17145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.932602266839193e-05,
+      "loss": 0.671,
+      "step": 17150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9320282977223335e-05,
+      "loss": 0.8394,
+      "step": 17155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.931454216238858e-05,
+      "loss": 0.7843,
+      "step": 17160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.930880022433815e-05,
+      "loss": 0.7456,
+      "step": 17165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9303057163522586e-05,
+      "loss": 0.7608,
+      "step": 17170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.929731298039252e-05,
+      "loss": 0.7318,
+      "step": 17175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.92915676753987e-05,
+      "loss": 0.7529,
+      "step": 17180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9285821248991915e-05,
+      "loss": 0.6915,
+      "step": 17185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.92800737016231e-05,
+      "loss": 0.6232,
+      "step": 17190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.927432503374322e-05,
+      "loss": 0.7671,
+      "step": 17195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.926857524580336e-05,
+      "loss": 0.7864,
+      "step": 17200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.92628243382547e-05,
+      "loss": 0.7925,
+      "step": 17205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.925707231154848e-05,
+      "loss": 0.8217,
+      "step": 17210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9251319166136046e-05,
+      "loss": 0.705,
+      "step": 17215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.924556490246884e-05,
+      "loss": 0.776,
+      "step": 17220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.923980952099836e-05,
+      "loss": 0.798,
+      "step": 17225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.923405302217623e-05,
+      "loss": 0.6944,
+      "step": 17230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.922829540645414e-05,
+      "loss": 0.6899,
+      "step": 17235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.922253667428387e-05,
+      "loss": 0.7824,
+      "step": 17240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9216776826117286e-05,
+      "loss": 0.875,
+      "step": 17245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.921101586240634e-05,
+      "loss": 0.9044,
+      "step": 17250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.920525378360309e-05,
+      "loss": 0.7314,
+      "step": 17255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.919949059015966e-05,
+      "loss": 0.7633,
+      "step": 17260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.919372628252827e-05,
+      "loss": 0.8406,
+      "step": 17265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.918796086116122e-05,
+      "loss": 0.8061,
+      "step": 17270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9182194326510916e-05,
+      "loss": 0.7262,
+      "step": 17275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9176426679029835e-05,
+      "loss": 0.7744,
+      "step": 17280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.917065791917053e-05,
+      "loss": 0.8709,
+      "step": 17285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.916488804738568e-05,
+      "loss": 0.9183,
+      "step": 17290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9159117064128027e-05,
+      "loss": 0.8939,
+      "step": 17295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9153344969850384e-05,
+      "loss": 0.6784,
+      "step": 17300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.914757176500567e-05,
+      "loss": 0.9736,
+      "step": 17305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.914179745004691e-05,
+      "loss": 0.8132,
+      "step": 17310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.913602202542718e-05,
+      "loss": 0.6945,
+      "step": 17315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.913024549159966e-05,
+      "loss": 0.8214,
+      "step": 17320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.912446784901762e-05,
+      "loss": 0.6764,
+      "step": 17325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.911868909813441e-05,
+      "loss": 0.6392,
+      "step": 17330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9112909239403475e-05,
+      "loss": 0.8495,
+      "step": 17335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.910712827327833e-05,
+      "loss": 0.5846,
+      "step": 17340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.91013462002126e-05,
+      "loss": 0.6663,
+      "step": 17345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.909556302065998e-05,
+      "loss": 0.8245,
+      "step": 17350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.908977873507425e-05,
+      "loss": 0.6627,
+      "step": 17355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.908399334390931e-05,
+      "loss": 0.7817,
+      "step": 17360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.907820684761909e-05,
+      "loss": 0.7928,
+      "step": 17365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9072419246657655e-05,
+      "loss": 0.932,
+      "step": 17370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.906663054147913e-05,
+      "loss": 0.7943,
+      "step": 17375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.906084073253775e-05,
+      "loss": 0.8446,
+      "step": 17380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.905504982028781e-05,
+      "loss": 0.7757,
+      "step": 17385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9049257805183714e-05,
+      "loss": 0.7323,
+      "step": 17390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.904346468767993e-05,
+      "loss": 0.6872,
+      "step": 17395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9037670468231037e-05,
+      "loss": 0.5631,
+      "step": 17400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9031875147291684e-05,
+      "loss": 0.7096,
+      "step": 17405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9026078725316605e-05,
+      "loss": 0.8894,
+      "step": 17410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.902028120276063e-05,
+      "loss": 0.72,
+      "step": 17415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9014482580078684e-05,
+      "loss": 0.8377,
+      "step": 17420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.900868285772575e-05,
+      "loss": 0.7026,
+      "step": 17425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.900288203615692e-05,
+      "loss": 0.7654,
+      "step": 17430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.899708011582737e-05,
+      "loss": 0.8032,
+      "step": 17435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.899127709719234e-05,
+      "loss": 0.7333,
+      "step": 17440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.898547298070719e-05,
+      "loss": 0.8479,
+      "step": 17445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8979667766827355e-05,
+      "loss": 0.6394,
+      "step": 17450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.897386145600834e-05,
+      "loss": 0.8133,
+      "step": 17455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.896805404870575e-05,
+      "loss": 0.8142,
+      "step": 17460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.896224554537527e-05,
+      "loss": 0.6381,
+      "step": 17465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.895643594647268e-05,
+      "loss": 0.5959,
+      "step": 17470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.895062525245384e-05,
+      "loss": 0.742,
+      "step": 17475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8944813463774705e-05,
+      "loss": 0.6821,
+      "step": 17480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.893900058089128e-05,
+      "loss": 0.808,
+      "step": 17485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8933186604259715e-05,
+      "loss": 0.6793,
+      "step": 17490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8927371534336196e-05,
+      "loss": 0.7619,
+      "step": 17495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8921555371577e-05,
+      "loss": 0.6483,
+      "step": 17500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.891573811643854e-05,
+      "loss": 0.899,
+      "step": 17505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.890991976937725e-05,
+      "loss": 0.7156,
+      "step": 17510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.890410033084968e-05,
+      "loss": 0.7189,
+      "step": 17515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.889827980131246e-05,
+      "loss": 0.9347,
+      "step": 17520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.889245818122232e-05,
+      "loss": 0.6657,
+      "step": 17525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8886635471036056e-05,
+      "loss": 0.7726,
+      "step": 17530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8880811671210556e-05,
+      "loss": 0.7855,
+      "step": 17535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.887498678220279e-05,
+      "loss": 0.7795,
+      "step": 17540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8869160804469834e-05,
+      "loss": 0.7495,
+      "step": 17545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8863333738468824e-05,
+      "loss": 0.8478,
+      "step": 17550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.885750558465698e-05,
+      "loss": 0.8729,
+      "step": 17555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.885167634349165e-05,
+      "loss": 0.9165,
+      "step": 17560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8845846015430195e-05,
+      "loss": 0.7312,
+      "step": 17565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.884001460093013e-05,
+      "loss": 0.7507,
+      "step": 17570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.883418210044901e-05,
+      "loss": 0.7741,
+      "step": 17575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.882834851444451e-05,
+      "loss": 0.7969,
+      "step": 17580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8822513843374367e-05,
+      "loss": 0.6521,
+      "step": 17585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.881667808769641e-05,
+      "loss": 0.7637,
+      "step": 17590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8810841247868535e-05,
+      "loss": 0.7294,
+      "step": 17595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8805003324348757e-05,
+      "loss": 0.7753,
+      "step": 17600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.879916431759516e-05,
+      "loss": 0.8253,
+      "step": 17605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.87933242280659e-05,
+      "loss": 0.7932,
+      "step": 17610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.878748305621923e-05,
+      "loss": 0.9105,
+      "step": 17615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.878164080251351e-05,
+      "loss": 0.7556,
+      "step": 17620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.877579746740714e-05,
+      "loss": 0.7761,
+      "step": 17625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.876995305135863e-05,
+      "loss": 0.8138,
+      "step": 17630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.876410755482658e-05,
+      "loss": 0.7631,
+      "step": 17635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.875826097826966e-05,
+      "loss": 0.8511,
+      "step": 17640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.875241332214664e-05,
+      "loss": 0.7067,
+      "step": 17645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.874656458691637e-05,
+      "loss": 0.8036,
+      "step": 17650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.874071477303775e-05,
+      "loss": 0.7485,
+      "step": 17655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8734863880969844e-05,
+      "loss": 0.941,
+      "step": 17660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.872901191117172e-05,
+      "loss": 0.7004,
+      "step": 17665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8723158864102566e-05,
+      "loss": 0.7929,
+      "step": 17670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.871730474022166e-05,
+      "loss": 0.8741,
+      "step": 17675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.871144953998835e-05,
+      "loss": 0.6594,
+      "step": 17680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8705593263862085e-05,
+      "loss": 0.7829,
+      "step": 17685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.869973591230237e-05,
+      "loss": 0.7887,
+      "step": 17690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.869387748576884e-05,
+      "loss": 0.699,
+      "step": 17695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.868801798472115e-05,
+      "loss": 0.6287,
+      "step": 17700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8682157409619105e-05,
+      "loss": 0.8502,
+      "step": 17705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8676295760922555e-05,
+      "loss": 0.7976,
+      "step": 17710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.867043303909145e-05,
+      "loss": 0.8013,
+      "step": 17715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.86645692445858e-05,
+      "loss": 0.685,
+      "step": 17720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.865870437786574e-05,
+      "loss": 0.8897,
+      "step": 17725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8652838439391464e-05,
+      "loss": 0.7566,
+      "step": 17730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.864697142962325e-05,
+      "loss": 0.6678,
+      "step": 17735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.864110334902145e-05,
+      "loss": 0.7299,
+      "step": 17740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8635234198046534e-05,
+      "loss": 0.7878,
+      "step": 17745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.862936397715902e-05,
+      "loss": 0.7374,
+      "step": 17750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.862349268681954e-05,
+      "loss": 0.7431,
+      "step": 17755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.861762032748878e-05,
+      "loss": 0.7168,
+      "step": 17760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8611746899627534e-05,
+      "loss": 0.7872,
+      "step": 17765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.860587240369666e-05,
+      "loss": 0.7851,
+      "step": 17770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8599996840157126e-05,
+      "loss": 0.9775,
+      "step": 17775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.859412020946995e-05,
+      "loss": 0.7903,
+      "step": 17780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.858824251209628e-05,
+      "loss": 0.7083,
+      "step": 17785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.85823637484973e-05,
+      "loss": 0.8283,
+      "step": 17790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8576483919134295e-05,
+      "loss": 0.7515,
+      "step": 17795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.857060302446864e-05,
+      "loss": 0.6571,
+      "step": 17800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8564721064961794e-05,
+      "loss": 0.7269,
+      "step": 17805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8558838041075296e-05,
+      "loss": 0.6345,
+      "step": 17810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.855295395327077e-05,
+      "loss": 0.7981,
+      "step": 17815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.85470688020099e-05,
+      "loss": 0.8889,
+      "step": 17820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8541182587754495e-05,
+      "loss": 0.8363,
+      "step": 17825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.853529531096643e-05,
+      "loss": 0.7905,
+      "step": 17830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.852940697210765e-05,
+      "loss": 0.7161,
+      "step": 17835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.85235175716402e-05,
+      "loss": 0.6421,
+      "step": 17840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.85176271100262e-05,
+      "loss": 0.6763,
+      "step": 17845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8511735587727846e-05,
+      "loss": 0.7565,
+      "step": 17850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.850584300520744e-05,
+      "loss": 0.8027,
+      "step": 17855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8499949362927354e-05,
+      "loss": 0.8461,
+      "step": 17860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.849405466135003e-05,
+      "loss": 0.8408,
+      "step": 17865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8488158900938016e-05,
+      "loss": 0.7571,
+      "step": 17870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8482262082153934e-05,
+      "loss": 0.8898,
+      "step": 17875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.847636420546049e-05,
+      "loss": 0.8263,
+      "step": 17880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8470465271320457e-05,
+      "loss": 0.7813,
+      "step": 17885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.846456528019672e-05,
+      "loss": 0.7865,
+      "step": 17890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.845866423255223e-05,
+      "loss": 0.7483,
+      "step": 17895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8452762128850017e-05,
+      "loss": 0.6165,
+      "step": 17900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.84468589695532e-05,
+      "loss": 0.5934,
+      "step": 17905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8440954755124994e-05,
+      "loss": 0.8084,
+      "step": 17910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8435049486028665e-05,
+      "loss": 0.7367,
+      "step": 17915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.842914316272759e-05,
+      "loss": 0.7221,
+      "step": 17920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.842323578568522e-05,
+      "loss": 0.5363,
+      "step": 17925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8417327355365086e-05,
+      "loss": 0.7407,
+      "step": 17930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8411417872230805e-05,
+      "loss": 0.7741,
+      "step": 17935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.840550733674607e-05,
+      "loss": 0.8834,
+      "step": 17940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.839959574937466e-05,
+      "loss": 0.673,
+      "step": 17945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.839368311058045e-05,
+      "loss": 0.811,
+      "step": 17950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8387769420827375e-05,
+      "loss": 0.7714,
+      "step": 17955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.838185468057947e-05,
+      "loss": 0.7051,
+      "step": 17960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.837593889030083e-05,
+      "loss": 0.7434,
+      "step": 17965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.837002205045568e-05,
+      "loss": 0.7249,
+      "step": 17970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8364104161508256e-05,
+      "loss": 0.7987,
+      "step": 17975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.835818522392294e-05,
+      "loss": 0.7849,
+      "step": 17980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.835226523816417e-05,
+      "loss": 0.7924,
+      "step": 17985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8346344204696465e-05,
+      "loss": 0.7745,
+      "step": 17990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8340422123984424e-05,
+      "loss": 0.706,
+      "step": 17995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.833449899649274e-05,
+      "loss": 0.7746,
+      "step": 18000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.832857482268618e-05,
+      "loss": 0.7411,
+      "step": 18005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8322649603029595e-05,
+      "loss": 0.602,
+      "step": 18010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8316723337987906e-05,
+      "loss": 0.8662,
+      "step": 18015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.831079602802616e-05,
+      "loss": 0.7628,
+      "step": 18020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.830486767360941e-05,
+      "loss": 0.6366,
+      "step": 18025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.829893827520287e-05,
+      "loss": 0.7286,
+      "step": 18030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8293007833271786e-05,
+      "loss": 0.8289,
+      "step": 18035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8287076348281505e-05,
+      "loss": 0.7653,
+      "step": 18040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8281143820697446e-05,
+      "loss": 0.8328,
+      "step": 18045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.827521025098512e-05,
+      "loss": 0.7595,
+      "step": 18050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.826927563961012e-05,
+      "loss": 0.6525,
+      "step": 18055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.826333998703809e-05,
+      "loss": 0.7187,
+      "step": 18060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8257403293734816e-05,
+      "loss": 0.7774,
+      "step": 18065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.825146556016611e-05,
+      "loss": 0.7758,
+      "step": 18070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.82455267867979e-05,
+      "loss": 0.7693,
+      "step": 18075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8239586974096164e-05,
+      "loss": 0.9131,
+      "step": 18080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.823364612252699e-05,
+      "loss": 0.7526,
+      "step": 18085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8227704232556546e-05,
+      "loss": 0.7251,
+      "step": 18090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.822176130465105e-05,
+      "loss": 0.8181,
+      "step": 18095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.821581733927685e-05,
+      "loss": 0.7405,
+      "step": 18100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8209872336900346e-05,
+      "loss": 0.7548,
+      "step": 18105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8203926297988e-05,
+      "loss": 0.7436,
+      "step": 18110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.81979792230064e-05,
+      "loss": 0.8826,
+      "step": 18115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8192031112422185e-05,
+      "loss": 0.7785,
+      "step": 18120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8186081966702085e-05,
+      "loss": 0.8562,
+      "step": 18125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.818013178631291e-05,
+      "loss": 0.7341,
+      "step": 18130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.817418057172156e-05,
+      "loss": 0.8296,
+      "step": 18135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.816822832339499e-05,
+      "loss": 0.8363,
+      "step": 18140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.816227504180028e-05,
+      "loss": 0.7421,
+      "step": 18145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.815632072740454e-05,
+      "loss": 0.99,
+      "step": 18150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.815036538067499e-05,
+      "loss": 0.7078,
+      "step": 18155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.814440900207894e-05,
+      "loss": 0.7589,
+      "step": 18160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8138451592083755e-05,
+      "loss": 0.8,
+      "step": 18165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8132493151156894e-05,
+      "loss": 0.7336,
+      "step": 18170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8126533679765894e-05,
+      "loss": 0.6928,
+      "step": 18175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.81205731783784e-05,
+      "loss": 0.8581,
+      "step": 18180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8114611647462084e-05,
+      "loss": 0.6296,
+      "step": 18185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.810864908748474e-05,
+      "loss": 0.611,
+      "step": 18190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8102685498914226e-05,
+      "loss": 0.8231,
+      "step": 18195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.80967208822185e-05,
+      "loss": 0.8017,
+      "step": 18200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8090755237865556e-05,
+      "loss": 0.7067,
+      "step": 18205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8084788566323524e-05,
+      "loss": 0.7478,
+      "step": 18210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.807882086806059e-05,
+      "loss": 0.784,
+      "step": 18215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8072852143545015e-05,
+      "loss": 0.8031,
+      "step": 18220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8066882393245126e-05,
+      "loss": 0.8042,
+      "step": 18225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.806091161762938e-05,
+      "loss": 0.7842,
+      "step": 18230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8054939817166275e-05,
+      "loss": 0.6238,
+      "step": 18235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.804896699232439e-05,
+      "loss": 0.7415,
+      "step": 18240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.804299314357239e-05,
+      "loss": 0.7814,
+      "step": 18245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.803701827137905e-05,
+      "loss": 0.6005,
+      "step": 18250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.803104237621318e-05,
+      "loss": 0.6153,
+      "step": 18255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.802506545854367e-05,
+      "loss": 0.6509,
+      "step": 18260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.801908751883955e-05,
+      "loss": 0.7546,
+      "step": 18265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.801310855756986e-05,
+      "loss": 0.6083,
+      "step": 18270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8007128575203765e-05,
+      "loss": 0.7606,
+      "step": 18275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8001147572210485e-05,
+      "loss": 0.7806,
+      "step": 18280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.799516554905932e-05,
+      "loss": 0.8587,
+      "step": 18285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.798918250621969e-05,
+      "loss": 0.6479,
+      "step": 18290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7983198444161045e-05,
+      "loss": 0.6683,
+      "step": 18295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.797721336335294e-05,
+      "loss": 0.7001,
+      "step": 18300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7971227264264997e-05,
+      "loss": 0.8493,
+      "step": 18305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.796524014736694e-05,
+      "loss": 0.7354,
+      "step": 18310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7959252013128546e-05,
+      "loss": 0.6859,
+      "step": 18315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.795326286201968e-05,
+      "loss": 0.7001,
+      "step": 18320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.794727269451032e-05,
+      "loss": 0.7015,
+      "step": 18325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.794128151107046e-05,
+      "loss": 0.7658,
+      "step": 18330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.793528931217023e-05,
+      "loss": 0.8859,
+      "step": 18335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.792929609827981e-05,
+      "loss": 0.7092,
+      "step": 18340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.792330186986947e-05,
+      "loss": 0.7721,
+      "step": 18345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7917306627409556e-05,
+      "loss": 0.7075,
+      "step": 18350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.79113103713705e-05,
+      "loss": 0.8593,
+      "step": 18355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7905313102222816e-05,
+      "loss": 0.7468,
+      "step": 18360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.789931482043707e-05,
+      "loss": 0.6434,
+      "step": 18365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.789331552648393e-05,
+      "loss": 0.7863,
+      "step": 18370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.788731522083416e-05,
+      "loss": 0.7467,
+      "step": 18375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.788131390395858e-05,
+      "loss": 0.6507,
+      "step": 18380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.787531157632808e-05,
+      "loss": 1.0363,
+      "step": 18385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.786930823841365e-05,
+      "loss": 0.9571,
+      "step": 18390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.786330389068635e-05,
+      "loss": 0.7973,
+      "step": 18395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.785729853361733e-05,
+      "loss": 0.9392,
+      "step": 18400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.785249352155273e-05,
+      "loss": 0.7905,
+      "step": 18405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7846486348856136e-05,
+      "loss": 0.665,
+      "step": 18410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.784047816813743e-05,
+      "loss": 0.7184,
+      "step": 18415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.783446897986806e-05,
+      "loss": 0.6136,
+      "step": 18420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.782845878451955e-05,
+      "loss": 0.7457,
+      "step": 18425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7822447582563505e-05,
+      "loss": 0.7722,
+      "step": 18430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.781643537447158e-05,
+      "loss": 0.5666,
+      "step": 18435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.781042216071555e-05,
+      "loss": 0.7349,
+      "step": 18440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.780440794176727e-05,
+      "loss": 0.6803,
+      "step": 18445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.779839271809863e-05,
+      "loss": 0.8977,
+      "step": 18450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.779237649018163e-05,
+      "loss": 0.712,
+      "step": 18455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7786359258488356e-05,
+      "loss": 0.8406,
+      "step": 18460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.778034102349095e-05,
+      "loss": 0.6713,
+      "step": 18465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.777432178566164e-05,
+      "loss": 0.6923,
+      "step": 18470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.776830154547275e-05,
+      "loss": 0.8148,
+      "step": 18475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.776228030339666e-05,
+      "loss": 0.663,
+      "step": 18480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.775625805990583e-05,
+      "loss": 0.7632,
+      "step": 18485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.775023481547282e-05,
+      "loss": 0.6002,
+      "step": 18490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.774421057057025e-05,
+      "loss": 0.8781,
+      "step": 18495
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7738185325670815e-05,
+      "loss": 0.585,
+      "step": 18500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.773215908124731e-05,
+      "loss": 0.793,
+      "step": 18505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.772613183777258e-05,
+      "loss": 0.7079,
+      "step": 18510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7720103595719586e-05,
+      "loss": 0.8202,
+      "step": 18515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.771407435556131e-05,
+      "loss": 0.6826,
+      "step": 18520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.770804411777088e-05,
+      "loss": 0.7195,
+      "step": 18525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.770201288282145e-05,
+      "loss": 0.7974,
+      "step": 18530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.769598065118628e-05,
+      "loss": 0.6449,
+      "step": 18535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7689947423338686e-05,
+      "loss": 0.7252,
+      "step": 18540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.768391319975211e-05,
+      "loss": 0.8655,
+      "step": 18545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7677877980900004e-05,
+      "loss": 0.7962,
+      "step": 18550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7671841767255935e-05,
+      "loss": 0.9109,
+      "step": 18555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.766580455929355e-05,
+      "loss": 0.802,
+      "step": 18560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.765976635748658e-05,
+      "loss": 0.7244,
+      "step": 18565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.765372716230881e-05,
+      "loss": 0.8154,
+      "step": 18570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7647686974234125e-05,
+      "loss": 0.9034,
+      "step": 18575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.764164579373647e-05,
+      "loss": 0.7674,
+      "step": 18580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.763560362128989e-05,
+      "loss": 0.79,
+      "step": 18585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.762956045736848e-05,
+      "loss": 0.7604,
+      "step": 18590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.762351630244643e-05,
+      "loss": 0.5801,
+      "step": 18595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7617471156998015e-05,
+      "loss": 0.7421,
+      "step": 18600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.761142502149758e-05,
+      "loss": 0.6293,
+      "step": 18605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.760537789641952e-05,
+      "loss": 0.761,
+      "step": 18610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7599329782238366e-05,
+      "loss": 0.7081,
+      "step": 18615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.759328067942867e-05,
+      "loss": 0.6849,
+      "step": 18620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7587230588465095e-05,
+      "loss": 0.7673,
+      "step": 18625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7581179509822376e-05,
+      "loss": 0.6882,
+      "step": 18630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7575127443975314e-05,
+      "loss": 0.6605,
+      "step": 18635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.75690743913988e-05,
+      "loss": 0.6872,
+      "step": 18640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.756302035256779e-05,
+      "loss": 0.9828,
+      "step": 18645
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.755696532795733e-05,
+      "loss": 0.7231,
+      "step": 18650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7550909318042547e-05,
+      "loss": 0.8458,
+      "step": 18655
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.754485232329862e-05,
+      "loss": 0.6933,
+      "step": 18660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7538794344200834e-05,
+      "loss": 0.7562,
+      "step": 18665
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.753273538122453e-05,
+      "loss": 0.7589,
+      "step": 18670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7526675434845146e-05,
+      "loss": 0.8413,
+      "step": 18675
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.752061450553816e-05,
+      "loss": 0.6979,
+      "step": 18680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7514552593779193e-05,
+      "loss": 0.7118,
+      "step": 18685
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.750848970004388e-05,
+      "loss": 0.6427,
+      "step": 18690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7502425824807965e-05,
+      "loss": 0.7137,
+      "step": 18695
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.749636096854724e-05,
+      "loss": 0.7334,
+      "step": 18700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.749029513173763e-05,
+      "loss": 0.6886,
+      "step": 18705
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.748422831485507e-05,
+      "loss": 0.7472,
+      "step": 18710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.747816051837561e-05,
+      "loss": 0.7642,
+      "step": 18715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7472091742775394e-05,
+      "loss": 0.7597,
+      "step": 18720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.746602198853059e-05,
+      "loss": 0.7698,
+      "step": 18725
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7459951256117484e-05,
+      "loss": 0.8039,
+      "step": 18730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.745387954601243e-05,
+      "loss": 0.7821,
+      "step": 18735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7447806858691835e-05,
+      "loss": 0.5738,
+      "step": 18740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.744173319463224e-05,
+      "loss": 0.8147,
+      "step": 18745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7435658554310195e-05,
+      "loss": 0.6916,
+      "step": 18750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7429582938202365e-05,
+      "loss": 0.633,
+      "step": 18755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.742350634678549e-05,
+      "loss": 0.6888,
+      "step": 18760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.741742878053637e-05,
+      "loss": 0.6898,
+      "step": 18765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.741135023993191e-05,
+      "loss": 0.7246,
+      "step": 18770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7405270725449056e-05,
+      "loss": 0.7306,
+      "step": 18775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.739919023756485e-05,
+      "loss": 0.8678,
+      "step": 18780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7393108776756414e-05,
+      "loss": 0.8372,
+      "step": 18785
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7387026343500934e-05,
+      "loss": 0.6038,
+      "step": 18790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.738094293827569e-05,
+      "loss": 0.7692,
+      "step": 18795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.737485856155802e-05,
+      "loss": 0.7382,
+      "step": 18800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.736877321382534e-05,
+      "loss": 0.8407,
+      "step": 18805
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.736268689555516e-05,
+      "loss": 0.7166,
+      "step": 18810
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7356599607225036e-05,
+      "loss": 0.7035,
+      "step": 18815
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.735051134931263e-05,
+      "loss": 0.693,
+      "step": 18820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7344422122295675e-05,
+      "loss": 0.7892,
+      "step": 18825
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7338331926651957e-05,
+      "loss": 0.6281,
+      "step": 18830
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.733224076285936e-05,
+      "loss": 0.6696,
+      "step": 18835
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.732614863139585e-05,
+      "loss": 0.748,
+      "step": 18840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7320055532739426e-05,
+      "loss": 0.7578,
+      "step": 18845
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7313961467368225e-05,
+      "loss": 0.7734,
+      "step": 18850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.730786643576042e-05,
+      "loss": 0.7208,
+      "step": 18855
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.730177043839426e-05,
+      "loss": 0.624,
+      "step": 18860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7295673475748085e-05,
+      "loss": 0.631,
+      "step": 18865
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.72895755483003e-05,
+      "loss": 0.7402,
+      "step": 18870
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7283476656529394e-05,
+      "loss": 0.7745,
+      "step": 18875
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.727737680091392e-05,
+      "loss": 0.7307,
+      "step": 18880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.727127598193252e-05,
+      "loss": 0.6934,
+      "step": 18885
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.726517420006391e-05,
+      "loss": 0.706,
+      "step": 18890
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.725907145578687e-05,
+      "loss": 0.5992,
+      "step": 18895
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.725296774958026e-05,
+      "loss": 0.7091,
+      "step": 18900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7246863081923024e-05,
+      "loss": 0.6441,
+      "step": 18905
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7240757453294174e-05,
+      "loss": 0.8,
+      "step": 18910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7234650864172795e-05,
+      "loss": 0.6456,
+      "step": 18915
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.722854331503806e-05,
+      "loss": 0.7247,
+      "step": 18920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.72224348063692e-05,
+      "loss": 0.6949,
+      "step": 18925
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.721632533864553e-05,
+      "loss": 0.6897,
+      "step": 18930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.721021491234644e-05,
+      "loss": 0.6947,
+      "step": 18935
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.720410352795141e-05,
+      "loss": 0.9082,
+      "step": 18940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.719799118593996e-05,
+      "loss": 0.7506,
+      "step": 18945
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.719187788679172e-05,
+      "loss": 0.6098,
+      "step": 18950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.718576363098637e-05,
+      "loss": 0.6518,
+      "step": 18955
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.717964841900369e-05,
+      "loss": 0.683,
+      "step": 18960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.71735322513235e-05,
+      "loss": 0.5155,
+      "step": 18965
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.716741512842573e-05,
+      "loss": 0.7194,
+      "step": 18970
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.716129705079037e-05,
+      "loss": 0.8037,
+      "step": 18975
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.715517801889749e-05,
+      "loss": 0.7316,
+      "step": 18980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.714905803322722e-05,
+      "loss": 0.5391,
+      "step": 18985
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.714293709425978e-05,
+      "loss": 0.7811,
+      "step": 18990
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.713681520247546e-05,
+      "loss": 0.7145,
+      "step": 18995
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.713069235835463e-05,
+      "loss": 0.7411,
+      "step": 19000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.712456856237772e-05,
+      "loss": 0.7058,
+      "step": 19005
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.711844381502525e-05,
+      "loss": 0.7548,
+      "step": 19010
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.711231811677781e-05,
+      "loss": 0.8202,
+      "step": 19015
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.710619146811606e-05,
+      "loss": 0.7619,
+      "step": 19020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.710006386952074e-05,
+      "loss": 0.748,
+      "step": 19025
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.709393532147267e-05,
+      "loss": 0.7685,
+      "step": 19030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7087805824452724e-05,
+      "loss": 0.6781,
+      "step": 19035
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.708167537894187e-05,
+      "loss": 0.6863,
+      "step": 19040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.707554398542115e-05,
+      "loss": 0.7509,
+      "step": 19045
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7069411644371666e-05,
+      "loss": 0.786,
+      "step": 19050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.70632783562746e-05,
+      "loss": 0.6377,
+      "step": 19055
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.705714412161123e-05,
+      "loss": 0.7119,
+      "step": 19060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.705100894086286e-05,
+      "loss": 0.5874,
+      "step": 19065
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.704487281451093e-05,
+      "loss": 0.772,
+      "step": 19070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.70387357430369e-05,
+      "loss": 0.7138,
+      "step": 19075
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.703259772692233e-05,
+      "loss": 0.7787,
+      "step": 19080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.702645876664886e-05,
+      "loss": 0.8605,
+      "step": 19085
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.702031886269818e-05,
+      "loss": 0.6587,
+      "step": 19090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.701417801555208e-05,
+      "loss": 0.816,
+      "step": 19095
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.70080362256924e-05,
+      "loss": 0.8239,
+      "step": 19100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.700189349360109e-05,
+      "loss": 0.7396,
+      "step": 19105
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6995749819760116e-05,
+      "loss": 0.7019,
+      "step": 19110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.698960520465158e-05,
+      "loss": 0.6598,
+      "step": 19115
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6983459648757615e-05,
+      "loss": 0.8095,
+      "step": 19120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.697731315256046e-05,
+      "loss": 0.7334,
+      "step": 19125
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6971165716542386e-05,
+      "loss": 0.669,
+      "step": 19130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.696501734118577e-05,
+      "loss": 0.6063,
+      "step": 19135
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6958868026973075e-05,
+      "loss": 0.7111,
+      "step": 19140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.69527177743868e-05,
+      "loss": 0.606,
+      "step": 19145
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.694656658390954e-05,
+      "loss": 0.7113,
+      "step": 19150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.694041445602394e-05,
+      "loss": 0.6619,
+      "step": 19155
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6934261391212775e-05,
+      "loss": 0.7227,
+      "step": 19160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6928107389958826e-05,
+      "loss": 0.7112,
+      "step": 19165
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.692195245274499e-05,
+      "loss": 0.721,
+      "step": 19170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6915796580054226e-05,
+      "loss": 0.6568,
+      "step": 19175
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.690963977236956e-05,
+      "loss": 0.5884,
+      "step": 19180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.690348203017409e-05,
+      "loss": 0.668,
+      "step": 19185
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6897323353951006e-05,
+      "loss": 0.7781,
+      "step": 19190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6891163744183566e-05,
+      "loss": 0.7888,
+      "step": 19195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.688500320135507e-05,
+      "loss": 0.585,
+      "step": 19200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6878841725948946e-05,
+      "loss": 0.7379,
+      "step": 19205
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6872679318448636e-05,
+      "loss": 0.6995,
+      "step": 19210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.686651597933771e-05,
+      "loss": 0.7771,
+      "step": 19215
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6860351709099754e-05,
+      "loss": 0.6342,
+      "step": 19220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.685418650821849e-05,
+      "loss": 0.7564,
+      "step": 19225
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6848020377177664e-05,
+      "loss": 0.5561,
+      "step": 19230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.684185331646112e-05,
+      "loss": 0.7206,
+      "step": 19235
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.683568532655276e-05,
+      "loss": 0.7525,
+      "step": 19240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.682951640793657e-05,
+      "loss": 0.7429,
+      "step": 19245
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.682334656109661e-05,
+      "loss": 0.8073,
+      "step": 19250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6817175786517e-05,
+      "loss": 0.6816,
+      "step": 19255
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6811004084681944e-05,
+      "loss": 0.6836,
+      "step": 19260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6804831456075704e-05,
+      "loss": 0.7229,
+      "step": 19265
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.679865790118265e-05,
+      "loss": 0.611,
+      "step": 19270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.679248342048719e-05,
+      "loss": 0.7383,
+      "step": 19275
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.67863080144738e-05,
+      "loss": 0.6316,
+      "step": 19280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6780131683627066e-05,
+      "loss": 0.6876,
+      "step": 19285
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.677395442843162e-05,
+      "loss": 0.7016,
+      "step": 19290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.676777624937216e-05,
+      "loss": 0.6569,
+      "step": 19295
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.676159714693347e-05,
+      "loss": 0.6753,
+      "step": 19300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.675541712160042e-05,
+      "loss": 0.7971,
+      "step": 19305
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.674923617385792e-05,
+      "loss": 0.6965,
+      "step": 19310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.674305430419097e-05,
+      "loss": 0.7058,
+      "step": 19315
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6736871513084656e-05,
+      "loss": 0.715,
+      "step": 19320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.673068780102411e-05,
+      "loss": 0.6447,
+      "step": 19325
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.672450316849454e-05,
+      "loss": 0.6567,
+      "step": 19330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6718317615981255e-05,
+      "loss": 0.7121,
+      "step": 19335
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6712131143969596e-05,
+      "loss": 0.6963,
+      "step": 19340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6705943752945e-05,
+      "loss": 0.8069,
+      "step": 19345
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6699755443392983e-05,
+      "loss": 0.6849,
+      "step": 19350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.669356621579911e-05,
+      "loss": 0.7365,
+      "step": 19355
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6687376070649024e-05,
+      "loss": 0.6641,
+      "step": 19360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.668118500842846e-05,
+      "loss": 0.6893,
+      "step": 19365
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6674993029623207e-05,
+      "loss": 0.7242,
+      "step": 19370
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.666880013471913e-05,
+      "loss": 0.678,
+      "step": 19375
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6662606324202145e-05,
+      "loss": 0.7119,
+      "step": 19380
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.665641159855829e-05,
+      "loss": 0.6684,
+      "step": 19385
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.665021595827364e-05,
+      "loss": 0.6933,
+      "step": 19390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.664401940383433e-05,
+      "loss": 0.6098,
+      "step": 19395
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.663782193572659e-05,
+      "loss": 0.7141,
+      "step": 19400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6631623554436725e-05,
+      "loss": 0.5819,
+      "step": 19405
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6625424260451094e-05,
+      "loss": 0.8023,
+      "step": 19410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6619224054256135e-05,
+      "loss": 0.7077,
+      "step": 19415
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.661302293633836e-05,
+      "loss": 0.8241,
+      "step": 19420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.660682090718435e-05,
+      "loss": 0.721,
+      "step": 19425
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6600617967280756e-05,
+      "loss": 0.7919,
+      "step": 19430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6594414117114314e-05,
+      "loss": 0.7374,
+      "step": 19435
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.65882093571718e-05,
+      "loss": 0.9226,
+      "step": 19440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.65820036879401e-05,
+      "loss": 0.6172,
+      "step": 19445
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.657579710990614e-05,
+      "loss": 0.7177,
+      "step": 19450
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.656958962355693e-05,
+      "loss": 0.5644,
+      "step": 19455
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6563381229379576e-05,
+      "loss": 0.6877,
+      "step": 19460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.655717192786119e-05,
+      "loss": 0.7084,
+      "step": 19465
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.655096171948903e-05,
+      "loss": 0.7819,
+      "step": 19470
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.654475060475037e-05,
+      "loss": 0.8572,
+      "step": 19475
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.653853858413259e-05,
+      "loss": 0.6891,
+      "step": 19480
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.653232565812311e-05,
+      "loss": 0.6481,
+      "step": 19485
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.652611182720946e-05,
+      "loss": 0.6192,
+      "step": 19490
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.651989709187921e-05,
+      "loss": 0.7298,
+      "step": 19495
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6513681452619994e-05,
+      "loss": 0.7542,
+      "step": 19500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.650746490991956e-05,
+      "loss": 0.7275,
+      "step": 19505
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6501247464265674e-05,
+      "loss": 0.7142,
+      "step": 19510
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.649502911614623e-05,
+      "loss": 0.7425,
+      "step": 19515
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6488809866049135e-05,
+      "loss": 0.7094,
+      "step": 19520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.64825897144624e-05,
+      "loss": 0.7883,
+      "step": 19525
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.64763686618741e-05,
+      "loss": 0.6493,
+      "step": 19530
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6470146708772384e-05,
+      "loss": 0.6927,
+      "step": 19535
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.646392385564547e-05,
+      "loss": 0.5466,
+      "step": 19540
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.645770010298164e-05,
+      "loss": 0.6321,
+      "step": 19545
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.645147545126926e-05,
+      "loss": 0.8171,
+      "step": 19550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.644524990099675e-05,
+      "loss": 0.5653,
+      "step": 19555
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.64390234526526e-05,
+      "loss": 0.7726,
+      "step": 19560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6432796106725396e-05,
+      "loss": 0.6603,
+      "step": 19565
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.642656786370378e-05,
+      "loss": 0.6843,
+      "step": 19570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.642033872407645e-05,
+      "loss": 0.6938,
+      "step": 19575
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6414108688332186e-05,
+      "loss": 0.677,
+      "step": 19580
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.640787775695985e-05,
+      "loss": 0.6901,
+      "step": 19585
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6401645930448356e-05,
+      "loss": 0.671,
+      "step": 19590
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6395413209286686e-05,
+      "loss": 0.741,
+      "step": 19595
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6389179593963914e-05,
+      "loss": 0.8225,
+      "step": 19600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6382945084969165e-05,
+      "loss": 0.715,
+      "step": 19605
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.637670968279165e-05,
+      "loss": 0.6918,
+      "step": 19610
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.637047338792063e-05,
+      "loss": 0.6316,
+      "step": 19615
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6364236200845456e-05,
+      "loss": 0.7644,
+      "step": 19620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.635799812205554e-05,
+      "loss": 0.659,
+      "step": 19625
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6351759152040355e-05,
+      "loss": 0.6877,
+      "step": 19630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.634551929128945e-05,
+      "loss": 0.7445,
+      "step": 19635
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.633927854029246e-05,
+      "loss": 0.7579,
+      "step": 19640
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.633303689953907e-05,
+      "loss": 0.8384,
+      "step": 19645
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6326794369519034e-05,
+      "loss": 0.6814,
+      "step": 19650
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6320550950722197e-05,
+      "loss": 0.7199,
+      "step": 19655
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.631430664363846e-05,
+      "loss": 0.7483,
+      "step": 19660
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6308061448757776e-05,
+      "loss": 0.5444,
+      "step": 19665
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.630181536657019e-05,
+      "loss": 0.7758,
+      "step": 19670
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.629556839756583e-05,
+      "loss": 0.7804,
+      "step": 19675
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.628932054223486e-05,
+      "loss": 0.7373,
+      "step": 19680
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6283071801067524e-05,
+      "loss": 0.6756,
+      "step": 19685
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6276822174554156e-05,
+      "loss": 0.6399,
+      "step": 19690
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.627057166318514e-05,
+      "loss": 0.7184,
+      "step": 19695
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.626432026745092e-05,
+      "loss": 0.7802,
+      "step": 19700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6258067987842045e-05,
+      "loss": 0.8092,
+      "step": 19705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.625181482484908e-05,
+      "loss": 0.7534,
+      "step": 19710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.624556077896273e-05,
+      "loss": 0.6718,
+      "step": 19715
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.62393058506737e-05,
+      "loss": 0.5859,
+      "step": 19720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.62330500404728e-05,
+      "loss": 0.7219,
+      "step": 19725
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.622679334885091e-05,
+      "loss": 0.7059,
+      "step": 19730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.622053577629896e-05,
+      "loss": 0.6691,
+      "step": 19735
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.621427732330797e-05,
+      "loss": 0.6074,
+      "step": 19740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6208017990369024e-05,
+      "loss": 0.7223,
+      "step": 19745
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.620175777797328e-05,
+      "loss": 0.6325,
+      "step": 19750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6195496686611926e-05,
+      "loss": 0.7658,
+      "step": 19755
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.618923471677626e-05,
+      "loss": 0.6585,
+      "step": 19760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.618297186895767e-05,
+      "loss": 0.6261,
+      "step": 19765
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.617670814364754e-05,
+      "loss": 0.7079,
+      "step": 19770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6170443541337394e-05,
+      "loss": 0.7239,
+      "step": 19775
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.616417806251877e-05,
+      "loss": 0.6381,
+      "step": 19780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.615791170768332e-05,
+      "loss": 0.6819,
+      "step": 19785
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.615164447732274e-05,
+      "loss": 0.6381,
+      "step": 19790
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.614537637192879e-05,
+      "loss": 0.7217,
+      "step": 19795
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6139107391993316e-05,
+      "loss": 0.7405,
+      "step": 19800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6132837538008225e-05,
+      "loss": 0.6009,
+      "step": 19805
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6126566810465486e-05,
+      "loss": 0.7134,
+      "step": 19810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.612029520985715e-05,
+      "loss": 0.7042,
+      "step": 19815
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6114022736675315e-05,
+      "loss": 0.7595,
+      "step": 19820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6107749391412184e-05,
+      "loss": 0.795,
+      "step": 19825
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.610147517455999e-05,
+      "loss": 0.6147,
+      "step": 19830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.609520008661105e-05,
+      "loss": 0.7308,
+      "step": 19835
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6088924128057766e-05,
+      "loss": 0.6944,
+      "step": 19840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.608264729939257e-05,
+      "loss": 0.5859,
+      "step": 19845
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6076369601108e-05,
+      "loss": 0.5451,
+      "step": 19850
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.607009103369663e-05,
+      "loss": 0.7287,
+      "step": 19855
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.606381159765115e-05,
+      "loss": 0.7377,
+      "step": 19860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.605753129346425e-05,
+      "loss": 0.7201,
+      "step": 19865
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.605125012162875e-05,
+      "loss": 0.5411,
+      "step": 19870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.604496808263751e-05,
+      "loss": 0.6901,
+      "step": 19875
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6038685176983445e-05,
+      "loss": 0.6943,
+      "step": 19880
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.603240140515957e-05,
+      "loss": 0.7507,
+      "step": 19885
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6026116767658954e-05,
+      "loss": 0.7565,
+      "step": 19890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.601983126497472e-05,
+      "loss": 0.6515,
+      "step": 19895
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.601354489760008e-05,
+      "loss": 0.6443,
+      "step": 19900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.600725766602831e-05,
+      "loss": 0.722,
+      "step": 19905
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.600096957075273e-05,
+      "loss": 0.6648,
+      "step": 19910
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.5994680612266756e-05,
+      "loss": 0.6729,
+      "step": 19915
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.598839079106387e-05,
+      "loss": 0.7067,
+      "step": 19920
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.598210010763761e-05,
+      "loss": 0.5048,
+      "step": 19925
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.597580856248157e-05,
+      "loss": 0.7746,
+      "step": 19930
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.596951615608945e-05,
+      "loss": 0.6867,
+      "step": 19935
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.596322288895498e-05,
+      "loss": 0.8095,
+      "step": 19940
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.5956928761571976e-05,
+      "loss": 0.5721,
+      "step": 19945
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.595063377443433e-05,
+      "loss": 0.539,
+      "step": 19950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.5944337928035964e-05,
+      "loss": 0.5423,
+      "step": 19955
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.593804122287091e-05,
+      "loss": 0.7047,
+      "step": 19960
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.5931743659433253e-05,
+      "loss": 0.6379,
+      "step": 19965
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.592544523821712e-05,
+      "loss": 0.671,
+      "step": 19970
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.5919145959716765e-05,
+      "loss": 0.6506,
+      "step": 19975
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.591284582442644e-05,
+      "loss": 0.725,
+      "step": 19980
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.59065448328405e-05,
+      "loss": 0.9172,
+      "step": 19985
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.590024298545338e-05,
+      "loss": 0.6289,
+      "step": 19990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.5893940282759555e-05,
+      "loss": 0.7067,
+      "step": 19995
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.5887636725253574e-05,
+      "loss": 0.6789,
+      "step": 20000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 56076,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 2.3139803180942623e+18,
+  "trial_name": null,
+  "trial_params": null
+}