Mansurbek commited on
Commit
c66a42d
1 Parent(s): c3f73c0

Delete trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +0 -4456
trainer_state.json DELETED
@@ -1,4456 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 2.12855310164704,
5
- "global_step": 740000,
6
- "is_hyper_param_search": false,
7
- "is_local_process_zero": true,
8
- "is_world_process_zero": true,
9
- "log_history": [
10
- {
11
- "epoch": 0.0,
12
- "learning_rate": 4.9985617884448335e-05,
13
- "loss": 8.3025,
14
- "step": 1000
15
- },
16
- {
17
- "epoch": 0.01,
18
- "learning_rate": 4.997123576889666e-05,
19
- "loss": 7.7816,
20
- "step": 2000
21
- },
22
- {
23
- "epoch": 0.01,
24
- "learning_rate": 4.9956853653344994e-05,
25
- "loss": 7.5291,
26
- "step": 3000
27
- },
28
- {
29
- "epoch": 0.01,
30
- "learning_rate": 4.9942471537793326e-05,
31
- "loss": 7.2663,
32
- "step": 4000
33
- },
34
- {
35
- "epoch": 0.01,
36
- "learning_rate": 4.992808942224166e-05,
37
- "loss": 6.9718,
38
- "step": 5000
39
- },
40
- {
41
- "epoch": 0.02,
42
- "learning_rate": 4.991370730668999e-05,
43
- "loss": 6.7375,
44
- "step": 6000
45
- },
46
- {
47
- "epoch": 0.02,
48
- "learning_rate": 4.989932519113832e-05,
49
- "loss": 6.4524,
50
- "step": 7000
51
- },
52
- {
53
- "epoch": 0.02,
54
- "learning_rate": 4.988494307558665e-05,
55
- "loss": 6.2397,
56
- "step": 8000
57
- },
58
- {
59
- "epoch": 0.03,
60
- "learning_rate": 4.9870560960034976e-05,
61
- "loss": 6.0318,
62
- "step": 9000
63
- },
64
- {
65
- "epoch": 0.03,
66
- "learning_rate": 4.985617884448331e-05,
67
- "loss": 5.8272,
68
- "step": 10000
69
- },
70
- {
71
- "epoch": 0.03,
72
- "learning_rate": 4.984179672893164e-05,
73
- "loss": 5.7037,
74
- "step": 11000
75
- },
76
- {
77
- "epoch": 0.03,
78
- "learning_rate": 4.9827414613379974e-05,
79
- "loss": 5.5378,
80
- "step": 12000
81
- },
82
- {
83
- "epoch": 0.04,
84
- "learning_rate": 4.981303249782831e-05,
85
- "loss": 5.3808,
86
- "step": 13000
87
- },
88
- {
89
- "epoch": 0.04,
90
- "learning_rate": 4.979865038227663e-05,
91
- "loss": 5.2818,
92
- "step": 14000
93
- },
94
- {
95
- "epoch": 0.04,
96
- "learning_rate": 4.9784268266724966e-05,
97
- "loss": 5.1759,
98
- "step": 15000
99
- },
100
- {
101
- "epoch": 0.05,
102
- "learning_rate": 4.97698861511733e-05,
103
- "loss": 5.0426,
104
- "step": 16000
105
- },
106
- {
107
- "epoch": 0.05,
108
- "learning_rate": 4.9755504035621624e-05,
109
- "loss": 4.974,
110
- "step": 17000
111
- },
112
- {
113
- "epoch": 0.05,
114
- "learning_rate": 4.974112192006996e-05,
115
- "loss": 4.885,
116
- "step": 18000
117
- },
118
- {
119
- "epoch": 0.05,
120
- "learning_rate": 4.972673980451828e-05,
121
- "loss": 4.8165,
122
- "step": 19000
123
- },
124
- {
125
- "epoch": 0.06,
126
- "learning_rate": 4.971235768896662e-05,
127
- "loss": 4.7407,
128
- "step": 20000
129
- },
130
- {
131
- "epoch": 0.06,
132
- "learning_rate": 4.969797557341495e-05,
133
- "loss": 4.6944,
134
- "step": 21000
135
- },
136
- {
137
- "epoch": 0.06,
138
- "learning_rate": 4.968359345786328e-05,
139
- "loss": 4.6118,
140
- "step": 22000
141
- },
142
- {
143
- "epoch": 0.07,
144
- "learning_rate": 4.9669211342311614e-05,
145
- "loss": 4.5637,
146
- "step": 23000
147
- },
148
- {
149
- "epoch": 0.07,
150
- "learning_rate": 4.965482922675994e-05,
151
- "loss": 4.525,
152
- "step": 24000
153
- },
154
- {
155
- "epoch": 0.07,
156
- "learning_rate": 4.964044711120827e-05,
157
- "loss": 4.4445,
158
- "step": 25000
159
- },
160
- {
161
- "epoch": 0.07,
162
- "learning_rate": 4.9626064995656605e-05,
163
- "loss": 4.4009,
164
- "step": 26000
165
- },
166
- {
167
- "epoch": 0.08,
168
- "learning_rate": 4.961168288010493e-05,
169
- "loss": 4.358,
170
- "step": 27000
171
- },
172
- {
173
- "epoch": 0.08,
174
- "learning_rate": 4.959730076455327e-05,
175
- "loss": 4.3549,
176
- "step": 28000
177
- },
178
- {
179
- "epoch": 0.08,
180
- "learning_rate": 4.9582918649001596e-05,
181
- "loss": 4.3059,
182
- "step": 29000
183
- },
184
- {
185
- "epoch": 0.09,
186
- "learning_rate": 4.956853653344993e-05,
187
- "loss": 4.2294,
188
- "step": 30000
189
- },
190
- {
191
- "epoch": 0.09,
192
- "learning_rate": 4.9554154417898255e-05,
193
- "loss": 4.2289,
194
- "step": 31000
195
- },
196
- {
197
- "epoch": 0.09,
198
- "learning_rate": 4.953977230234659e-05,
199
- "loss": 4.1809,
200
- "step": 32000
201
- },
202
- {
203
- "epoch": 0.09,
204
- "learning_rate": 4.952539018679492e-05,
205
- "loss": 4.1458,
206
- "step": 33000
207
- },
208
- {
209
- "epoch": 0.1,
210
- "learning_rate": 4.9511008071243246e-05,
211
- "loss": 4.1041,
212
- "step": 34000
213
- },
214
- {
215
- "epoch": 0.1,
216
- "learning_rate": 4.9496625955691585e-05,
217
- "loss": 4.0819,
218
- "step": 35000
219
- },
220
- {
221
- "epoch": 0.1,
222
- "learning_rate": 4.948224384013991e-05,
223
- "loss": 4.0453,
224
- "step": 36000
225
- },
226
- {
227
- "epoch": 0.11,
228
- "learning_rate": 4.9467861724588244e-05,
229
- "loss": 4.0294,
230
- "step": 37000
231
- },
232
- {
233
- "epoch": 0.11,
234
- "learning_rate": 4.945347960903658e-05,
235
- "loss": 4.017,
236
- "step": 38000
237
- },
238
- {
239
- "epoch": 0.11,
240
- "learning_rate": 4.94390974934849e-05,
241
- "loss": 3.9947,
242
- "step": 39000
243
- },
244
- {
245
- "epoch": 0.12,
246
- "learning_rate": 4.9424715377933235e-05,
247
- "loss": 3.9406,
248
- "step": 40000
249
- },
250
- {
251
- "epoch": 0.12,
252
- "learning_rate": 4.941033326238156e-05,
253
- "loss": 3.9302,
254
- "step": 41000
255
- },
256
- {
257
- "epoch": 0.12,
258
- "learning_rate": 4.9395951146829894e-05,
259
- "loss": 3.9016,
260
- "step": 42000
261
- },
262
- {
263
- "epoch": 0.12,
264
- "learning_rate": 4.9381569031278227e-05,
265
- "loss": 3.8573,
266
- "step": 43000
267
- },
268
- {
269
- "epoch": 0.13,
270
- "learning_rate": 4.936718691572656e-05,
271
- "loss": 3.8791,
272
- "step": 44000
273
- },
274
- {
275
- "epoch": 0.13,
276
- "learning_rate": 4.935280480017489e-05,
277
- "loss": 3.8259,
278
- "step": 45000
279
- },
280
- {
281
- "epoch": 0.13,
282
- "learning_rate": 4.933842268462322e-05,
283
- "loss": 3.7879,
284
- "step": 46000
285
- },
286
- {
287
- "epoch": 0.14,
288
- "learning_rate": 4.932404056907155e-05,
289
- "loss": 3.8047,
290
- "step": 47000
291
- },
292
- {
293
- "epoch": 0.14,
294
- "learning_rate": 4.930965845351988e-05,
295
- "loss": 3.7862,
296
- "step": 48000
297
- },
298
- {
299
- "epoch": 0.14,
300
- "learning_rate": 4.929527633796821e-05,
301
- "loss": 3.7373,
302
- "step": 49000
303
- },
304
- {
305
- "epoch": 0.14,
306
- "learning_rate": 4.928089422241654e-05,
307
- "loss": 3.7103,
308
- "step": 50000
309
- },
310
- {
311
- "epoch": 0.15,
312
- "learning_rate": 4.9266512106864875e-05,
313
- "loss": 3.7193,
314
- "step": 51000
315
- },
316
- {
317
- "epoch": 0.15,
318
- "learning_rate": 4.925212999131321e-05,
319
- "loss": 3.6898,
320
- "step": 52000
321
- },
322
- {
323
- "epoch": 0.15,
324
- "learning_rate": 4.923774787576153e-05,
325
- "loss": 3.6754,
326
- "step": 53000
327
- },
328
- {
329
- "epoch": 0.16,
330
- "learning_rate": 4.9223365760209866e-05,
331
- "loss": 3.661,
332
- "step": 54000
333
- },
334
- {
335
- "epoch": 0.16,
336
- "learning_rate": 4.92089836446582e-05,
337
- "loss": 3.6634,
338
- "step": 55000
339
- },
340
- {
341
- "epoch": 0.16,
342
- "learning_rate": 4.9194601529106524e-05,
343
- "loss": 3.6258,
344
- "step": 56000
345
- },
346
- {
347
- "epoch": 0.16,
348
- "learning_rate": 4.918021941355486e-05,
349
- "loss": 3.6481,
350
- "step": 57000
351
- },
352
- {
353
- "epoch": 0.17,
354
- "learning_rate": 4.916583729800319e-05,
355
- "loss": 3.6197,
356
- "step": 58000
357
- },
358
- {
359
- "epoch": 0.17,
360
- "learning_rate": 4.915145518245152e-05,
361
- "loss": 3.6065,
362
- "step": 59000
363
- },
364
- {
365
- "epoch": 0.17,
366
- "learning_rate": 4.9137073066899855e-05,
367
- "loss": 3.5744,
368
- "step": 60000
369
- },
370
- {
371
- "epoch": 0.18,
372
- "learning_rate": 4.912269095134818e-05,
373
- "loss": 3.539,
374
- "step": 61000
375
- },
376
- {
377
- "epoch": 0.18,
378
- "learning_rate": 4.9108308835796514e-05,
379
- "loss": 3.5618,
380
- "step": 62000
381
- },
382
- {
383
- "epoch": 0.18,
384
- "learning_rate": 4.909392672024484e-05,
385
- "loss": 3.5485,
386
- "step": 63000
387
- },
388
- {
389
- "epoch": 0.18,
390
- "learning_rate": 4.907954460469317e-05,
391
- "loss": 3.5257,
392
- "step": 64000
393
- },
394
- {
395
- "epoch": 0.19,
396
- "learning_rate": 4.9065162489141505e-05,
397
- "loss": 3.5017,
398
- "step": 65000
399
- },
400
- {
401
- "epoch": 0.19,
402
- "learning_rate": 4.905078037358984e-05,
403
- "loss": 3.5034,
404
- "step": 66000
405
- },
406
- {
407
- "epoch": 0.19,
408
- "learning_rate": 4.903639825803817e-05,
409
- "loss": 3.481,
410
- "step": 67000
411
- },
412
- {
413
- "epoch": 0.2,
414
- "learning_rate": 4.9022016142486496e-05,
415
- "loss": 3.4964,
416
- "step": 68000
417
- },
418
- {
419
- "epoch": 0.2,
420
- "learning_rate": 4.900763402693483e-05,
421
- "loss": 3.4815,
422
- "step": 69000
423
- },
424
- {
425
- "epoch": 0.2,
426
- "learning_rate": 4.899325191138316e-05,
427
- "loss": 3.4469,
428
- "step": 70000
429
- },
430
- {
431
- "epoch": 0.2,
432
- "learning_rate": 4.897886979583149e-05,
433
- "loss": 3.4362,
434
- "step": 71000
435
- },
436
- {
437
- "epoch": 0.21,
438
- "learning_rate": 4.896448768027982e-05,
439
- "loss": 3.4294,
440
- "step": 72000
441
- },
442
- {
443
- "epoch": 0.21,
444
- "learning_rate": 4.8950105564728146e-05,
445
- "loss": 3.4208,
446
- "step": 73000
447
- },
448
- {
449
- "epoch": 0.21,
450
- "learning_rate": 4.8935723449176486e-05,
451
- "loss": 3.3885,
452
- "step": 74000
453
- },
454
- {
455
- "epoch": 0.22,
456
- "learning_rate": 4.892134133362482e-05,
457
- "loss": 3.3929,
458
- "step": 75000
459
- },
460
- {
461
- "epoch": 0.22,
462
- "learning_rate": 4.8906959218073144e-05,
463
- "loss": 3.4022,
464
- "step": 76000
465
- },
466
- {
467
- "epoch": 0.22,
468
- "learning_rate": 4.889257710252148e-05,
469
- "loss": 3.3768,
470
- "step": 77000
471
- },
472
- {
473
- "epoch": 0.22,
474
- "learning_rate": 4.88781949869698e-05,
475
- "loss": 3.3687,
476
- "step": 78000
477
- },
478
- {
479
- "epoch": 0.23,
480
- "learning_rate": 4.8863812871418136e-05,
481
- "loss": 3.3437,
482
- "step": 79000
483
- },
484
- {
485
- "epoch": 0.23,
486
- "learning_rate": 4.884943075586647e-05,
487
- "loss": 3.3396,
488
- "step": 80000
489
- },
490
- {
491
- "epoch": 0.23,
492
- "learning_rate": 4.8835048640314794e-05,
493
- "loss": 3.3368,
494
- "step": 81000
495
- },
496
- {
497
- "epoch": 0.24,
498
- "learning_rate": 4.8820666524763134e-05,
499
- "loss": 3.3354,
500
- "step": 82000
501
- },
502
- {
503
- "epoch": 0.24,
504
- "learning_rate": 4.880628440921146e-05,
505
- "loss": 3.3168,
506
- "step": 83000
507
- },
508
- {
509
- "epoch": 0.24,
510
- "learning_rate": 4.879190229365979e-05,
511
- "loss": 3.2886,
512
- "step": 84000
513
- },
514
- {
515
- "epoch": 0.24,
516
- "learning_rate": 4.877752017810812e-05,
517
- "loss": 3.3108,
518
- "step": 85000
519
- },
520
- {
521
- "epoch": 0.25,
522
- "learning_rate": 4.876313806255645e-05,
523
- "loss": 3.2812,
524
- "step": 86000
525
- },
526
- {
527
- "epoch": 0.25,
528
- "learning_rate": 4.8748755947004783e-05,
529
- "loss": 3.2797,
530
- "step": 87000
531
- },
532
- {
533
- "epoch": 0.25,
534
- "learning_rate": 4.873437383145311e-05,
535
- "loss": 3.2961,
536
- "step": 88000
537
- },
538
- {
539
- "epoch": 0.26,
540
- "learning_rate": 4.871999171590144e-05,
541
- "loss": 3.27,
542
- "step": 89000
543
- },
544
- {
545
- "epoch": 0.26,
546
- "learning_rate": 4.8705609600349775e-05,
547
- "loss": 3.2683,
548
- "step": 90000
549
- },
550
- {
551
- "epoch": 0.26,
552
- "learning_rate": 4.869122748479811e-05,
553
- "loss": 3.2646,
554
- "step": 91000
555
- },
556
- {
557
- "epoch": 0.26,
558
- "learning_rate": 4.867684536924644e-05,
559
- "loss": 3.2471,
560
- "step": 92000
561
- },
562
- {
563
- "epoch": 0.27,
564
- "learning_rate": 4.8662463253694766e-05,
565
- "loss": 3.2278,
566
- "step": 93000
567
- },
568
- {
569
- "epoch": 0.27,
570
- "learning_rate": 4.86480811381431e-05,
571
- "loss": 3.2201,
572
- "step": 94000
573
- },
574
- {
575
- "epoch": 0.27,
576
- "learning_rate": 4.8633699022591425e-05,
577
- "loss": 3.2131,
578
- "step": 95000
579
- },
580
- {
581
- "epoch": 0.28,
582
- "learning_rate": 4.861931690703976e-05,
583
- "loss": 3.2069,
584
- "step": 96000
585
- },
586
- {
587
- "epoch": 0.28,
588
- "learning_rate": 4.86049347914881e-05,
589
- "loss": 3.1964,
590
- "step": 97000
591
- },
592
- {
593
- "epoch": 0.28,
594
- "learning_rate": 4.859055267593642e-05,
595
- "loss": 3.203,
596
- "step": 98000
597
- },
598
- {
599
- "epoch": 0.28,
600
- "learning_rate": 4.8576170560384755e-05,
601
- "loss": 3.1894,
602
- "step": 99000
603
- },
604
- {
605
- "epoch": 0.29,
606
- "learning_rate": 4.856178844483308e-05,
607
- "loss": 3.1747,
608
- "step": 100000
609
- },
610
- {
611
- "epoch": 0.29,
612
- "learning_rate": 4.8547406329281414e-05,
613
- "loss": 3.1698,
614
- "step": 101000
615
- },
616
- {
617
- "epoch": 0.29,
618
- "learning_rate": 4.853302421372975e-05,
619
- "loss": 3.1413,
620
- "step": 102000
621
- },
622
- {
623
- "epoch": 0.3,
624
- "learning_rate": 4.851864209817807e-05,
625
- "loss": 3.1599,
626
- "step": 103000
627
- },
628
- {
629
- "epoch": 0.3,
630
- "learning_rate": 4.8504259982626405e-05,
631
- "loss": 3.1493,
632
- "step": 104000
633
- },
634
- {
635
- "epoch": 0.3,
636
- "learning_rate": 4.848987786707474e-05,
637
- "loss": 3.1273,
638
- "step": 105000
639
- },
640
- {
641
- "epoch": 0.3,
642
- "learning_rate": 4.847549575152307e-05,
643
- "loss": 3.1186,
644
- "step": 106000
645
- },
646
- {
647
- "epoch": 0.31,
648
- "learning_rate": 4.84611136359714e-05,
649
- "loss": 3.1539,
650
- "step": 107000
651
- },
652
- {
653
- "epoch": 0.31,
654
- "learning_rate": 4.844673152041973e-05,
655
- "loss": 3.1386,
656
- "step": 108000
657
- },
658
- {
659
- "epoch": 0.31,
660
- "learning_rate": 4.843234940486806e-05,
661
- "loss": 3.1323,
662
- "step": 109000
663
- },
664
- {
665
- "epoch": 0.32,
666
- "learning_rate": 4.841796728931639e-05,
667
- "loss": 3.121,
668
- "step": 110000
669
- },
670
- {
671
- "epoch": 0.32,
672
- "learning_rate": 4.840358517376472e-05,
673
- "loss": 3.0904,
674
- "step": 111000
675
- },
676
- {
677
- "epoch": 0.32,
678
- "learning_rate": 4.838920305821305e-05,
679
- "loss": 3.0882,
680
- "step": 112000
681
- },
682
- {
683
- "epoch": 0.33,
684
- "learning_rate": 4.8374820942661386e-05,
685
- "loss": 3.0924,
686
- "step": 113000
687
- },
688
- {
689
- "epoch": 0.33,
690
- "learning_rate": 4.836043882710972e-05,
691
- "loss": 3.1044,
692
- "step": 114000
693
- },
694
- {
695
- "epoch": 0.33,
696
- "learning_rate": 4.8346056711558044e-05,
697
- "loss": 3.0805,
698
- "step": 115000
699
- },
700
- {
701
- "epoch": 0.33,
702
- "learning_rate": 4.833167459600638e-05,
703
- "loss": 3.0895,
704
- "step": 116000
705
- },
706
- {
707
- "epoch": 0.34,
708
- "learning_rate": 4.831729248045471e-05,
709
- "loss": 3.0581,
710
- "step": 117000
711
- },
712
- {
713
- "epoch": 0.34,
714
- "learning_rate": 4.8302910364903036e-05,
715
- "loss": 3.0611,
716
- "step": 118000
717
- },
718
- {
719
- "epoch": 0.34,
720
- "learning_rate": 4.828852824935137e-05,
721
- "loss": 3.0496,
722
- "step": 119000
723
- },
724
- {
725
- "epoch": 0.35,
726
- "learning_rate": 4.8274146133799694e-05,
727
- "loss": 3.0739,
728
- "step": 120000
729
- },
730
- {
731
- "epoch": 0.35,
732
- "learning_rate": 4.8259764018248034e-05,
733
- "loss": 3.0475,
734
- "step": 121000
735
- },
736
- {
737
- "epoch": 0.35,
738
- "learning_rate": 4.824538190269636e-05,
739
- "loss": 3.0539,
740
- "step": 122000
741
- },
742
- {
743
- "epoch": 0.35,
744
- "learning_rate": 4.823099978714469e-05,
745
- "loss": 3.026,
746
- "step": 123000
747
- },
748
- {
749
- "epoch": 0.36,
750
- "learning_rate": 4.8216617671593025e-05,
751
- "loss": 3.0164,
752
- "step": 124000
753
- },
754
- {
755
- "epoch": 0.36,
756
- "learning_rate": 4.820223555604135e-05,
757
- "loss": 3.0238,
758
- "step": 125000
759
- },
760
- {
761
- "epoch": 0.36,
762
- "learning_rate": 4.8187853440489684e-05,
763
- "loss": 3.0243,
764
- "step": 126000
765
- },
766
- {
767
- "epoch": 0.37,
768
- "learning_rate": 4.8173471324938016e-05,
769
- "loss": 3.0241,
770
- "step": 127000
771
- },
772
- {
773
- "epoch": 0.37,
774
- "learning_rate": 4.815908920938635e-05,
775
- "loss": 3.021,
776
- "step": 128000
777
- },
778
- {
779
- "epoch": 0.37,
780
- "learning_rate": 4.814470709383468e-05,
781
- "loss": 3.0055,
782
- "step": 129000
783
- },
784
- {
785
- "epoch": 0.37,
786
- "learning_rate": 4.813032497828301e-05,
787
- "loss": 3.0213,
788
- "step": 130000
789
- },
790
- {
791
- "epoch": 0.38,
792
- "learning_rate": 4.811594286273134e-05,
793
- "loss": 3.0181,
794
- "step": 131000
795
- },
796
- {
797
- "epoch": 0.38,
798
- "learning_rate": 4.8101560747179666e-05,
799
- "loss": 2.9918,
800
- "step": 132000
801
- },
802
- {
803
- "epoch": 0.38,
804
- "learning_rate": 4.8087178631628e-05,
805
- "loss": 2.9748,
806
- "step": 133000
807
- },
808
- {
809
- "epoch": 0.39,
810
- "learning_rate": 4.807279651607633e-05,
811
- "loss": 2.9746,
812
- "step": 134000
813
- },
814
- {
815
- "epoch": 0.39,
816
- "learning_rate": 4.805841440052466e-05,
817
- "loss": 2.9883,
818
- "step": 135000
819
- },
820
- {
821
- "epoch": 0.39,
822
- "learning_rate": 4.8044032284973e-05,
823
- "loss": 2.9722,
824
- "step": 136000
825
- },
826
- {
827
- "epoch": 0.39,
828
- "learning_rate": 4.802965016942132e-05,
829
- "loss": 2.9932,
830
- "step": 137000
831
- },
832
- {
833
- "epoch": 0.4,
834
- "learning_rate": 4.8015268053869656e-05,
835
- "loss": 2.9794,
836
- "step": 138000
837
- },
838
- {
839
- "epoch": 0.4,
840
- "learning_rate": 4.800088593831799e-05,
841
- "loss": 2.9695,
842
- "step": 139000
843
- },
844
- {
845
- "epoch": 0.4,
846
- "learning_rate": 4.7986503822766314e-05,
847
- "loss": 2.9509,
848
- "step": 140000
849
- },
850
- {
851
- "epoch": 0.41,
852
- "learning_rate": 4.797212170721465e-05,
853
- "loss": 2.954,
854
- "step": 141000
855
- },
856
- {
857
- "epoch": 0.41,
858
- "learning_rate": 4.795773959166297e-05,
859
- "loss": 2.9467,
860
- "step": 142000
861
- },
862
- {
863
- "epoch": 0.41,
864
- "learning_rate": 4.7943357476111305e-05,
865
- "loss": 2.9329,
866
- "step": 143000
867
- },
868
- {
869
- "epoch": 0.41,
870
- "learning_rate": 4.792897536055964e-05,
871
- "loss": 2.9157,
872
- "step": 144000
873
- },
874
- {
875
- "epoch": 0.42,
876
- "learning_rate": 4.791459324500797e-05,
877
- "loss": 2.9528,
878
- "step": 145000
879
- },
880
- {
881
- "epoch": 0.42,
882
- "learning_rate": 4.7900211129456304e-05,
883
- "loss": 2.952,
884
- "step": 146000
885
- },
886
- {
887
- "epoch": 0.42,
888
- "learning_rate": 4.788582901390463e-05,
889
- "loss": 2.92,
890
- "step": 147000
891
- },
892
- {
893
- "epoch": 0.43,
894
- "learning_rate": 4.787144689835296e-05,
895
- "loss": 2.9207,
896
- "step": 148000
897
- },
898
- {
899
- "epoch": 0.43,
900
- "learning_rate": 4.7857064782801295e-05,
901
- "loss": 2.9031,
902
- "step": 149000
903
- },
904
- {
905
- "epoch": 0.43,
906
- "learning_rate": 4.784268266724962e-05,
907
- "loss": 2.9024,
908
- "step": 150000
909
- },
910
- {
911
- "epoch": 0.43,
912
- "learning_rate": 4.7828300551697953e-05,
913
- "loss": 2.9313,
914
- "step": 151000
915
- },
916
- {
917
- "epoch": 0.44,
918
- "learning_rate": 4.7813918436146286e-05,
919
- "loss": 2.9239,
920
- "step": 152000
921
- },
922
- {
923
- "epoch": 0.44,
924
- "learning_rate": 4.779953632059462e-05,
925
- "loss": 2.9183,
926
- "step": 153000
927
- },
928
- {
929
- "epoch": 0.44,
930
- "learning_rate": 4.7785154205042945e-05,
931
- "loss": 2.883,
932
- "step": 154000
933
- },
934
- {
935
- "epoch": 0.45,
936
- "learning_rate": 4.777077208949128e-05,
937
- "loss": 2.8812,
938
- "step": 155000
939
- },
940
- {
941
- "epoch": 0.45,
942
- "learning_rate": 4.775638997393961e-05,
943
- "loss": 2.8808,
944
- "step": 156000
945
- },
946
- {
947
- "epoch": 0.45,
948
- "learning_rate": 4.7742007858387936e-05,
949
- "loss": 2.8817,
950
- "step": 157000
951
- },
952
- {
953
- "epoch": 0.45,
954
- "learning_rate": 4.772762574283627e-05,
955
- "loss": 2.887,
956
- "step": 158000
957
- },
958
- {
959
- "epoch": 0.46,
960
- "learning_rate": 4.77132436272846e-05,
961
- "loss": 2.8945,
962
- "step": 159000
963
- },
964
- {
965
- "epoch": 0.46,
966
- "learning_rate": 4.7698861511732934e-05,
967
- "loss": 2.8613,
968
- "step": 160000
969
- },
970
- {
971
- "epoch": 0.46,
972
- "learning_rate": 4.768447939618127e-05,
973
- "loss": 2.8734,
974
- "step": 161000
975
- },
976
- {
977
- "epoch": 0.47,
978
- "learning_rate": 4.767009728062959e-05,
979
- "loss": 2.8879,
980
- "step": 162000
981
- },
982
- {
983
- "epoch": 0.47,
984
- "learning_rate": 4.7655715165077925e-05,
985
- "loss": 2.8596,
986
- "step": 163000
987
- },
988
- {
989
- "epoch": 0.47,
990
- "learning_rate": 4.764133304952625e-05,
991
- "loss": 2.8558,
992
- "step": 164000
993
- },
994
- {
995
- "epoch": 0.47,
996
- "learning_rate": 4.7626950933974584e-05,
997
- "loss": 2.8576,
998
- "step": 165000
999
- },
1000
- {
1001
- "epoch": 0.48,
1002
- "learning_rate": 4.7612568818422917e-05,
1003
- "loss": 2.8612,
1004
- "step": 166000
1005
- },
1006
- {
1007
- "epoch": 0.48,
1008
- "learning_rate": 4.759818670287125e-05,
1009
- "loss": 2.8549,
1010
- "step": 167000
1011
- },
1012
- {
1013
- "epoch": 0.48,
1014
- "learning_rate": 4.758380458731958e-05,
1015
- "loss": 2.844,
1016
- "step": 168000
1017
- },
1018
- {
1019
- "epoch": 0.49,
1020
- "learning_rate": 4.756942247176791e-05,
1021
- "loss": 2.864,
1022
- "step": 169000
1023
- },
1024
- {
1025
- "epoch": 0.49,
1026
- "learning_rate": 4.755504035621624e-05,
1027
- "loss": 2.8386,
1028
- "step": 170000
1029
- },
1030
- {
1031
- "epoch": 0.49,
1032
- "learning_rate": 4.754065824066457e-05,
1033
- "loss": 2.8605,
1034
- "step": 171000
1035
- },
1036
- {
1037
- "epoch": 0.49,
1038
- "learning_rate": 4.75262761251129e-05,
1039
- "loss": 2.8197,
1040
- "step": 172000
1041
- },
1042
- {
1043
- "epoch": 0.5,
1044
- "learning_rate": 4.751189400956123e-05,
1045
- "loss": 2.8409,
1046
- "step": 173000
1047
- },
1048
- {
1049
- "epoch": 0.5,
1050
- "learning_rate": 4.749751189400956e-05,
1051
- "loss": 2.8411,
1052
- "step": 174000
1053
- },
1054
- {
1055
- "epoch": 0.5,
1056
- "learning_rate": 4.74831297784579e-05,
1057
- "loss": 2.8358,
1058
- "step": 175000
1059
- },
1060
- {
1061
- "epoch": 0.51,
1062
- "learning_rate": 4.746874766290623e-05,
1063
- "loss": 2.8358,
1064
- "step": 176000
1065
- },
1066
- {
1067
- "epoch": 0.51,
1068
- "learning_rate": 4.7454365547354556e-05,
1069
- "loss": 2.8319,
1070
- "step": 177000
1071
- },
1072
- {
1073
- "epoch": 0.51,
1074
- "learning_rate": 4.743998343180289e-05,
1075
- "loss": 2.8248,
1076
- "step": 178000
1077
- },
1078
- {
1079
- "epoch": 0.51,
1080
- "learning_rate": 4.7425601316251214e-05,
1081
- "loss": 2.8218,
1082
- "step": 179000
1083
- },
1084
- {
1085
- "epoch": 0.52,
1086
- "learning_rate": 4.741121920069955e-05,
1087
- "loss": 2.8383,
1088
- "step": 180000
1089
- },
1090
- {
1091
- "epoch": 0.52,
1092
- "learning_rate": 4.739683708514788e-05,
1093
- "loss": 2.8357,
1094
- "step": 181000
1095
- },
1096
- {
1097
- "epoch": 0.52,
1098
- "learning_rate": 4.7382454969596206e-05,
1099
- "loss": 2.8354,
1100
- "step": 182000
1101
- },
1102
- {
1103
- "epoch": 0.53,
1104
- "learning_rate": 4.7368072854044545e-05,
1105
- "loss": 2.817,
1106
- "step": 183000
1107
- },
1108
- {
1109
- "epoch": 0.53,
1110
- "learning_rate": 4.735369073849287e-05,
1111
- "loss": 2.7872,
1112
- "step": 184000
1113
- },
1114
- {
1115
- "epoch": 0.53,
1116
- "learning_rate": 4.7339308622941204e-05,
1117
- "loss": 2.8032,
1118
- "step": 185000
1119
- },
1120
- {
1121
- "epoch": 0.54,
1122
- "learning_rate": 4.7324926507389536e-05,
1123
- "loss": 2.801,
1124
- "step": 186000
1125
- },
1126
- {
1127
- "epoch": 0.54,
1128
- "learning_rate": 4.731054439183786e-05,
1129
- "loss": 2.7922,
1130
- "step": 187000
1131
- },
1132
- {
1133
- "epoch": 0.54,
1134
- "learning_rate": 4.7296162276286195e-05,
1135
- "loss": 2.7929,
1136
- "step": 188000
1137
- },
1138
- {
1139
- "epoch": 0.54,
1140
- "learning_rate": 4.728178016073452e-05,
1141
- "loss": 2.8169,
1142
- "step": 189000
1143
- },
1144
- {
1145
- "epoch": 0.55,
1146
- "learning_rate": 4.726739804518286e-05,
1147
- "loss": 2.8019,
1148
- "step": 190000
1149
- },
1150
- {
1151
- "epoch": 0.55,
1152
- "learning_rate": 4.7253015929631186e-05,
1153
- "loss": 2.7967,
1154
- "step": 191000
1155
- },
1156
- {
1157
- "epoch": 0.55,
1158
- "learning_rate": 4.723863381407952e-05,
1159
- "loss": 2.7808,
1160
- "step": 192000
1161
- },
1162
- {
1163
- "epoch": 0.56,
1164
- "learning_rate": 4.722425169852785e-05,
1165
- "loss": 2.7989,
1166
- "step": 193000
1167
- },
1168
- {
1169
- "epoch": 0.56,
1170
- "learning_rate": 4.720986958297618e-05,
1171
- "loss": 2.7846,
1172
- "step": 194000
1173
- },
1174
- {
1175
- "epoch": 0.56,
1176
- "learning_rate": 4.719548746742451e-05,
1177
- "loss": 2.7647,
1178
- "step": 195000
1179
- },
1180
- {
1181
- "epoch": 0.56,
1182
- "learning_rate": 4.718110535187284e-05,
1183
- "loss": 2.7715,
1184
- "step": 196000
1185
- },
1186
- {
1187
- "epoch": 0.57,
1188
- "learning_rate": 4.716672323632117e-05,
1189
- "loss": 2.7617,
1190
- "step": 197000
1191
- },
1192
- {
1193
- "epoch": 0.57,
1194
- "learning_rate": 4.715234112076951e-05,
1195
- "loss": 2.7564,
1196
- "step": 198000
1197
- },
1198
- {
1199
- "epoch": 0.57,
1200
- "learning_rate": 4.7137959005217834e-05,
1201
- "loss": 2.7703,
1202
- "step": 199000
1203
- },
1204
- {
1205
- "epoch": 0.58,
1206
- "learning_rate": 4.712357688966617e-05,
1207
- "loss": 2.7559,
1208
- "step": 200000
1209
- },
1210
- {
1211
- "epoch": 0.58,
1212
- "learning_rate": 4.710919477411449e-05,
1213
- "loss": 2.7788,
1214
- "step": 201000
1215
- },
1216
- {
1217
- "epoch": 0.58,
1218
- "learning_rate": 4.7094812658562826e-05,
1219
- "loss": 2.7475,
1220
- "step": 202000
1221
- },
1222
- {
1223
- "epoch": 0.58,
1224
- "learning_rate": 4.708043054301116e-05,
1225
- "loss": 2.7479,
1226
- "step": 203000
1227
- },
1228
- {
1229
- "epoch": 0.59,
1230
- "learning_rate": 4.7066048427459484e-05,
1231
- "loss": 2.7623,
1232
- "step": 204000
1233
- },
1234
- {
1235
- "epoch": 0.59,
1236
- "learning_rate": 4.705166631190782e-05,
1237
- "loss": 2.7569,
1238
- "step": 205000
1239
- },
1240
- {
1241
- "epoch": 0.59,
1242
- "learning_rate": 4.703728419635615e-05,
1243
- "loss": 2.7482,
1244
- "step": 206000
1245
- },
1246
- {
1247
- "epoch": 0.6,
1248
- "learning_rate": 4.702290208080448e-05,
1249
- "loss": 2.7618,
1250
- "step": 207000
1251
- },
1252
- {
1253
- "epoch": 0.6,
1254
- "learning_rate": 4.7008519965252815e-05,
1255
- "loss": 2.7649,
1256
- "step": 208000
1257
- },
1258
- {
1259
- "epoch": 0.6,
1260
- "learning_rate": 4.699413784970114e-05,
1261
- "loss": 2.7468,
1262
- "step": 209000
1263
- },
1264
- {
1265
- "epoch": 0.6,
1266
- "learning_rate": 4.6979755734149473e-05,
1267
- "loss": 2.7403,
1268
- "step": 210000
1269
- },
1270
- {
1271
- "epoch": 0.61,
1272
- "learning_rate": 4.69653736185978e-05,
1273
- "loss": 2.7505,
1274
- "step": 211000
1275
- },
1276
- {
1277
- "epoch": 0.61,
1278
- "learning_rate": 4.695099150304613e-05,
1279
- "loss": 2.7218,
1280
- "step": 212000
1281
- },
1282
- {
1283
- "epoch": 0.61,
1284
- "learning_rate": 4.6936609387494465e-05,
1285
- "loss": 2.7156,
1286
- "step": 213000
1287
- },
1288
- {
1289
- "epoch": 0.62,
1290
- "learning_rate": 4.69222272719428e-05,
1291
- "loss": 2.7158,
1292
- "step": 214000
1293
- },
1294
- {
1295
- "epoch": 0.62,
1296
- "learning_rate": 4.690784515639113e-05,
1297
- "loss": 2.7019,
1298
- "step": 215000
1299
- },
1300
- {
1301
- "epoch": 0.62,
1302
- "learning_rate": 4.6893463040839456e-05,
1303
- "loss": 2.7253,
1304
- "step": 216000
1305
- },
1306
- {
1307
- "epoch": 0.62,
1308
- "learning_rate": 4.687908092528779e-05,
1309
- "loss": 2.7182,
1310
- "step": 217000
1311
- },
1312
- {
1313
- "epoch": 0.63,
1314
- "learning_rate": 4.686469880973612e-05,
1315
- "loss": 2.7288,
1316
- "step": 218000
1317
- },
1318
- {
1319
- "epoch": 0.63,
1320
- "learning_rate": 4.685031669418445e-05,
1321
- "loss": 2.7296,
1322
- "step": 219000
1323
- },
1324
- {
1325
- "epoch": 0.63,
1326
- "learning_rate": 4.683593457863278e-05,
1327
- "loss": 2.7019,
1328
- "step": 220000
1329
- },
1330
- {
1331
- "epoch": 0.64,
1332
- "learning_rate": 4.682155246308111e-05,
1333
- "loss": 2.6918,
1334
- "step": 221000
1335
- },
1336
- {
1337
- "epoch": 0.64,
1338
- "learning_rate": 4.6807170347529445e-05,
1339
- "loss": 2.7087,
1340
- "step": 222000
1341
- },
1342
- {
1343
- "epoch": 0.64,
1344
- "learning_rate": 4.679278823197777e-05,
1345
- "loss": 2.71,
1346
- "step": 223000
1347
- },
1348
- {
1349
- "epoch": 0.64,
1350
- "learning_rate": 4.6778406116426104e-05,
1351
- "loss": 2.6998,
1352
- "step": 224000
1353
- },
1354
- {
1355
- "epoch": 0.65,
1356
- "learning_rate": 4.676402400087444e-05,
1357
- "loss": 2.7024,
1358
- "step": 225000
1359
- },
1360
- {
1361
- "epoch": 0.65,
1362
- "learning_rate": 4.674964188532276e-05,
1363
- "loss": 2.7129,
1364
- "step": 226000
1365
- },
1366
- {
1367
- "epoch": 0.65,
1368
- "learning_rate": 4.6735259769771095e-05,
1369
- "loss": 2.7066,
1370
- "step": 227000
1371
- },
1372
- {
1373
- "epoch": 0.66,
1374
- "learning_rate": 4.672087765421943e-05,
1375
- "loss": 2.7125,
1376
- "step": 228000
1377
- },
1378
- {
1379
- "epoch": 0.66,
1380
- "learning_rate": 4.670649553866776e-05,
1381
- "loss": 2.6963,
1382
- "step": 229000
1383
- },
1384
- {
1385
- "epoch": 0.66,
1386
- "learning_rate": 4.669211342311609e-05,
1387
- "loss": 2.7268,
1388
- "step": 230000
1389
- },
1390
- {
1391
- "epoch": 0.66,
1392
- "learning_rate": 4.667773130756442e-05,
1393
- "loss": 2.7021,
1394
- "step": 231000
1395
- },
1396
- {
1397
- "epoch": 0.67,
1398
- "learning_rate": 4.666334919201275e-05,
1399
- "loss": 2.6985,
1400
- "step": 232000
1401
- },
1402
- {
1403
- "epoch": 0.67,
1404
- "learning_rate": 4.664896707646108e-05,
1405
- "loss": 2.6941,
1406
- "step": 233000
1407
- },
1408
- {
1409
- "epoch": 0.67,
1410
- "learning_rate": 4.663458496090941e-05,
1411
- "loss": 2.6878,
1412
- "step": 234000
1413
- },
1414
- {
1415
- "epoch": 0.68,
1416
- "learning_rate": 4.662020284535774e-05,
1417
- "loss": 2.6874,
1418
- "step": 235000
1419
- },
1420
- {
1421
- "epoch": 0.68,
1422
- "learning_rate": 4.660582072980607e-05,
1423
- "loss": 2.6842,
1424
- "step": 236000
1425
- },
1426
- {
1427
- "epoch": 0.68,
1428
- "learning_rate": 4.659143861425441e-05,
1429
- "loss": 2.6723,
1430
- "step": 237000
1431
- },
1432
- {
1433
- "epoch": 0.68,
1434
- "learning_rate": 4.6577056498702734e-05,
1435
- "loss": 2.7055,
1436
- "step": 238000
1437
- },
1438
- {
1439
- "epoch": 0.69,
1440
- "learning_rate": 4.656267438315107e-05,
1441
- "loss": 2.6746,
1442
- "step": 239000
1443
- },
1444
- {
1445
- "epoch": 0.69,
1446
- "learning_rate": 4.65482922675994e-05,
1447
- "loss": 2.6872,
1448
- "step": 240000
1449
- },
1450
- {
1451
- "epoch": 0.69,
1452
- "learning_rate": 4.6533910152047726e-05,
1453
- "loss": 2.6865,
1454
- "step": 241000
1455
- },
1456
- {
1457
- "epoch": 0.7,
1458
- "learning_rate": 4.651952803649606e-05,
1459
- "loss": 2.6728,
1460
- "step": 242000
1461
- },
1462
- {
1463
- "epoch": 0.7,
1464
- "learning_rate": 4.6505145920944384e-05,
1465
- "loss": 2.6626,
1466
- "step": 243000
1467
- },
1468
- {
1469
- "epoch": 0.7,
1470
- "learning_rate": 4.6490763805392724e-05,
1471
- "loss": 2.6654,
1472
- "step": 244000
1473
- },
1474
- {
1475
- "epoch": 0.7,
1476
- "learning_rate": 4.6476381689841056e-05,
1477
- "loss": 2.653,
1478
- "step": 245000
1479
- },
1480
- {
1481
- "epoch": 0.71,
1482
- "learning_rate": 4.646199957428938e-05,
1483
- "loss": 2.671,
1484
- "step": 246000
1485
- },
1486
- {
1487
- "epoch": 0.71,
1488
- "learning_rate": 4.6447617458737715e-05,
1489
- "loss": 2.6642,
1490
- "step": 247000
1491
- },
1492
- {
1493
- "epoch": 0.71,
1494
- "learning_rate": 4.643323534318604e-05,
1495
- "loss": 2.6554,
1496
- "step": 248000
1497
- },
1498
- {
1499
- "epoch": 0.72,
1500
- "learning_rate": 4.6418853227634374e-05,
1501
- "loss": 2.6669,
1502
- "step": 249000
1503
- },
1504
- {
1505
- "epoch": 0.72,
1506
- "learning_rate": 4.6404471112082706e-05,
1507
- "loss": 2.6781,
1508
- "step": 250000
1509
- },
1510
- {
1511
- "epoch": 0.72,
1512
- "learning_rate": 4.639008899653103e-05,
1513
- "loss": 2.6485,
1514
- "step": 251000
1515
- },
1516
- {
1517
- "epoch": 0.72,
1518
- "learning_rate": 4.637570688097937e-05,
1519
- "loss": 2.6709,
1520
- "step": 252000
1521
- },
1522
- {
1523
- "epoch": 0.73,
1524
- "learning_rate": 4.63613247654277e-05,
1525
- "loss": 2.6579,
1526
- "step": 253000
1527
- },
1528
- {
1529
- "epoch": 0.73,
1530
- "learning_rate": 4.634694264987603e-05,
1531
- "loss": 2.6438,
1532
- "step": 254000
1533
- },
1534
- {
1535
- "epoch": 0.73,
1536
- "learning_rate": 4.6332560534324356e-05,
1537
- "loss": 2.6413,
1538
- "step": 255000
1539
- },
1540
- {
1541
- "epoch": 0.74,
1542
- "learning_rate": 4.631817841877269e-05,
1543
- "loss": 2.6721,
1544
- "step": 256000
1545
- },
1546
- {
1547
- "epoch": 0.74,
1548
- "learning_rate": 4.630379630322102e-05,
1549
- "loss": 2.6642,
1550
- "step": 257000
1551
- },
1552
- {
1553
- "epoch": 0.74,
1554
- "learning_rate": 4.628941418766935e-05,
1555
- "loss": 2.6535,
1556
- "step": 258000
1557
- },
1558
- {
1559
- "epoch": 0.74,
1560
- "learning_rate": 4.627503207211768e-05,
1561
- "loss": 2.6321,
1562
- "step": 259000
1563
- },
1564
- {
1565
- "epoch": 0.75,
1566
- "learning_rate": 4.626064995656601e-05,
1567
- "loss": 2.6421,
1568
- "step": 260000
1569
- },
1570
- {
1571
- "epoch": 0.75,
1572
- "learning_rate": 4.6246267841014346e-05,
1573
- "loss": 2.6419,
1574
- "step": 261000
1575
- },
1576
- {
1577
- "epoch": 0.75,
1578
- "learning_rate": 4.623188572546268e-05,
1579
- "loss": 2.6055,
1580
- "step": 262000
1581
- },
1582
- {
1583
- "epoch": 0.76,
1584
- "learning_rate": 4.6217503609911004e-05,
1585
- "loss": 2.6555,
1586
- "step": 263000
1587
- },
1588
- {
1589
- "epoch": 0.76,
1590
- "learning_rate": 4.620312149435934e-05,
1591
- "loss": 2.6568,
1592
- "step": 264000
1593
- },
1594
- {
1595
- "epoch": 0.76,
1596
- "learning_rate": 4.618873937880766e-05,
1597
- "loss": 2.6382,
1598
- "step": 265000
1599
- },
1600
- {
1601
- "epoch": 0.77,
1602
- "learning_rate": 4.6174357263255995e-05,
1603
- "loss": 2.6517,
1604
- "step": 266000
1605
- },
1606
- {
1607
- "epoch": 0.77,
1608
- "learning_rate": 4.615997514770433e-05,
1609
- "loss": 2.6121,
1610
- "step": 267000
1611
- },
1612
- {
1613
- "epoch": 0.77,
1614
- "learning_rate": 4.614559303215266e-05,
1615
- "loss": 2.6113,
1616
- "step": 268000
1617
- },
1618
- {
1619
- "epoch": 0.77,
1620
- "learning_rate": 4.6131210916600994e-05,
1621
- "loss": 2.6238,
1622
- "step": 269000
1623
- },
1624
- {
1625
- "epoch": 0.78,
1626
- "learning_rate": 4.611682880104932e-05,
1627
- "loss": 2.6347,
1628
- "step": 270000
1629
- },
1630
- {
1631
- "epoch": 0.78,
1632
- "learning_rate": 4.610244668549765e-05,
1633
- "loss": 2.6399,
1634
- "step": 271000
1635
- },
1636
- {
1637
- "epoch": 0.78,
1638
- "learning_rate": 4.6088064569945985e-05,
1639
- "loss": 2.6161,
1640
- "step": 272000
1641
- },
1642
- {
1643
- "epoch": 0.79,
1644
- "learning_rate": 4.607368245439431e-05,
1645
- "loss": 2.6291,
1646
- "step": 273000
1647
- },
1648
- {
1649
- "epoch": 0.79,
1650
- "learning_rate": 4.6059300338842643e-05,
1651
- "loss": 2.6609,
1652
- "step": 274000
1653
- },
1654
- {
1655
- "epoch": 0.79,
1656
- "learning_rate": 4.6044918223290976e-05,
1657
- "loss": 2.6113,
1658
- "step": 275000
1659
- },
1660
- {
1661
- "epoch": 0.79,
1662
- "learning_rate": 4.603053610773931e-05,
1663
- "loss": 2.6169,
1664
- "step": 276000
1665
- },
1666
- {
1667
- "epoch": 0.8,
1668
- "learning_rate": 4.601615399218764e-05,
1669
- "loss": 2.5953,
1670
- "step": 277000
1671
- },
1672
- {
1673
- "epoch": 0.8,
1674
- "learning_rate": 4.600177187663597e-05,
1675
- "loss": 2.624,
1676
- "step": 278000
1677
- },
1678
- {
1679
- "epoch": 0.8,
1680
- "learning_rate": 4.59873897610843e-05,
1681
- "loss": 2.6063,
1682
- "step": 279000
1683
- },
1684
- {
1685
- "epoch": 0.81,
1686
- "learning_rate": 4.5973007645532626e-05,
1687
- "loss": 2.6139,
1688
- "step": 280000
1689
- },
1690
- {
1691
- "epoch": 0.81,
1692
- "learning_rate": 4.595862552998096e-05,
1693
- "loss": 2.6096,
1694
- "step": 281000
1695
- },
1696
- {
1697
- "epoch": 0.81,
1698
- "learning_rate": 4.594424341442929e-05,
1699
- "loss": 2.6377,
1700
- "step": 282000
1701
- },
1702
- {
1703
- "epoch": 0.81,
1704
- "learning_rate": 4.5929861298877624e-05,
1705
- "loss": 2.6175,
1706
- "step": 283000
1707
- },
1708
- {
1709
- "epoch": 0.82,
1710
- "learning_rate": 4.591547918332596e-05,
1711
- "loss": 2.6049,
1712
- "step": 284000
1713
- },
1714
- {
1715
- "epoch": 0.82,
1716
- "learning_rate": 4.590109706777428e-05,
1717
- "loss": 2.6157,
1718
- "step": 285000
1719
- },
1720
- {
1721
- "epoch": 0.82,
1722
- "learning_rate": 4.5886714952222615e-05,
1723
- "loss": 2.6097,
1724
- "step": 286000
1725
- },
1726
- {
1727
- "epoch": 0.83,
1728
- "learning_rate": 4.587233283667095e-05,
1729
- "loss": 2.6071,
1730
- "step": 287000
1731
- },
1732
- {
1733
- "epoch": 0.83,
1734
- "learning_rate": 4.5857950721119274e-05,
1735
- "loss": 2.587,
1736
- "step": 288000
1737
- },
1738
- {
1739
- "epoch": 0.83,
1740
- "learning_rate": 4.5843568605567607e-05,
1741
- "loss": 2.6143,
1742
- "step": 289000
1743
- },
1744
- {
1745
- "epoch": 0.83,
1746
- "learning_rate": 4.582918649001593e-05,
1747
- "loss": 2.5819,
1748
- "step": 290000
1749
- },
1750
- {
1751
- "epoch": 0.84,
1752
- "learning_rate": 4.581480437446427e-05,
1753
- "loss": 2.5935,
1754
- "step": 291000
1755
- },
1756
- {
1757
- "epoch": 0.84,
1758
- "learning_rate": 4.58004222589126e-05,
1759
- "loss": 2.6032,
1760
- "step": 292000
1761
- },
1762
- {
1763
- "epoch": 0.84,
1764
- "learning_rate": 4.578604014336093e-05,
1765
- "loss": 2.5783,
1766
- "step": 293000
1767
- },
1768
- {
1769
- "epoch": 0.85,
1770
- "learning_rate": 4.577165802780926e-05,
1771
- "loss": 2.6006,
1772
- "step": 294000
1773
- },
1774
- {
1775
- "epoch": 0.85,
1776
- "learning_rate": 4.575727591225759e-05,
1777
- "loss": 2.588,
1778
- "step": 295000
1779
- },
1780
- {
1781
- "epoch": 0.85,
1782
- "learning_rate": 4.574289379670592e-05,
1783
- "loss": 2.5968,
1784
- "step": 296000
1785
- },
1786
- {
1787
- "epoch": 0.85,
1788
- "learning_rate": 4.5728511681154255e-05,
1789
- "loss": 2.5952,
1790
- "step": 297000
1791
- },
1792
- {
1793
- "epoch": 0.86,
1794
- "learning_rate": 4.571412956560258e-05,
1795
- "loss": 2.5735,
1796
- "step": 298000
1797
- },
1798
- {
1799
- "epoch": 0.86,
1800
- "learning_rate": 4.569974745005092e-05,
1801
- "loss": 2.5927,
1802
- "step": 299000
1803
- },
1804
- {
1805
- "epoch": 0.86,
1806
- "learning_rate": 4.5685365334499246e-05,
1807
- "loss": 2.5798,
1808
- "step": 300000
1809
- },
1810
- {
1811
- "epoch": 0.87,
1812
- "learning_rate": 4.567098321894758e-05,
1813
- "loss": 2.5779,
1814
- "step": 301000
1815
- },
1816
- {
1817
- "epoch": 0.87,
1818
- "learning_rate": 4.5656601103395904e-05,
1819
- "loss": 2.5829,
1820
- "step": 302000
1821
- },
1822
- {
1823
- "epoch": 0.87,
1824
- "learning_rate": 4.564221898784424e-05,
1825
- "loss": 2.5909,
1826
- "step": 303000
1827
- },
1828
- {
1829
- "epoch": 0.87,
1830
- "learning_rate": 4.562783687229257e-05,
1831
- "loss": 2.5908,
1832
- "step": 304000
1833
- },
1834
- {
1835
- "epoch": 0.88,
1836
- "learning_rate": 4.5613454756740896e-05,
1837
- "loss": 2.5774,
1838
- "step": 305000
1839
- },
1840
- {
1841
- "epoch": 0.88,
1842
- "learning_rate": 4.5599072641189235e-05,
1843
- "loss": 2.5621,
1844
- "step": 306000
1845
- },
1846
- {
1847
- "epoch": 0.88,
1848
- "learning_rate": 4.558469052563756e-05,
1849
- "loss": 2.5782,
1850
- "step": 307000
1851
- },
1852
- {
1853
- "epoch": 0.89,
1854
- "learning_rate": 4.5570308410085894e-05,
1855
- "loss": 2.5892,
1856
- "step": 308000
1857
- },
1858
- {
1859
- "epoch": 0.89,
1860
- "learning_rate": 4.5555926294534226e-05,
1861
- "loss": 2.5812,
1862
- "step": 309000
1863
- },
1864
- {
1865
- "epoch": 0.89,
1866
- "learning_rate": 4.554154417898255e-05,
1867
- "loss": 2.5722,
1868
- "step": 310000
1869
- },
1870
- {
1871
- "epoch": 0.89,
1872
- "learning_rate": 4.5527162063430885e-05,
1873
- "loss": 2.5602,
1874
- "step": 311000
1875
- },
1876
- {
1877
- "epoch": 0.9,
1878
- "learning_rate": 4.551277994787921e-05,
1879
- "loss": 2.5647,
1880
- "step": 312000
1881
- },
1882
- {
1883
- "epoch": 0.9,
1884
- "learning_rate": 4.5498397832327544e-05,
1885
- "loss": 2.5753,
1886
- "step": 313000
1887
- },
1888
- {
1889
- "epoch": 0.9,
1890
- "learning_rate": 4.5484015716775876e-05,
1891
- "loss": 2.5702,
1892
- "step": 314000
1893
- },
1894
- {
1895
- "epoch": 0.91,
1896
- "learning_rate": 4.546963360122421e-05,
1897
- "loss": 2.5631,
1898
- "step": 315000
1899
- },
1900
- {
1901
- "epoch": 0.91,
1902
- "learning_rate": 4.545525148567254e-05,
1903
- "loss": 2.5714,
1904
- "step": 316000
1905
- },
1906
- {
1907
- "epoch": 0.91,
1908
- "learning_rate": 4.544086937012087e-05,
1909
- "loss": 2.5669,
1910
- "step": 317000
1911
- },
1912
- {
1913
- "epoch": 0.91,
1914
- "learning_rate": 4.54264872545692e-05,
1915
- "loss": 2.5696,
1916
- "step": 318000
1917
- },
1918
- {
1919
- "epoch": 0.92,
1920
- "learning_rate": 4.541210513901753e-05,
1921
- "loss": 2.5964,
1922
- "step": 319000
1923
- },
1924
- {
1925
- "epoch": 0.92,
1926
- "learning_rate": 4.539772302346586e-05,
1927
- "loss": 2.5503,
1928
- "step": 320000
1929
- },
1930
- {
1931
- "epoch": 0.92,
1932
- "learning_rate": 4.538334090791419e-05,
1933
- "loss": 2.5611,
1934
- "step": 321000
1935
- },
1936
- {
1937
- "epoch": 0.93,
1938
- "learning_rate": 4.5368958792362524e-05,
1939
- "loss": 2.5782,
1940
- "step": 322000
1941
- },
1942
- {
1943
- "epoch": 0.93,
1944
- "learning_rate": 4.535457667681086e-05,
1945
- "loss": 2.543,
1946
- "step": 323000
1947
- },
1948
- {
1949
- "epoch": 0.93,
1950
- "learning_rate": 4.534019456125918e-05,
1951
- "loss": 2.5482,
1952
- "step": 324000
1953
- },
1954
- {
1955
- "epoch": 0.93,
1956
- "learning_rate": 4.5325812445707516e-05,
1957
- "loss": 2.5554,
1958
- "step": 325000
1959
- },
1960
- {
1961
- "epoch": 0.94,
1962
- "learning_rate": 4.531143033015585e-05,
1963
- "loss": 2.5776,
1964
- "step": 326000
1965
- },
1966
- {
1967
- "epoch": 0.94,
1968
- "learning_rate": 4.5297048214604174e-05,
1969
- "loss": 2.5432,
1970
- "step": 327000
1971
- },
1972
- {
1973
- "epoch": 0.94,
1974
- "learning_rate": 4.528266609905251e-05,
1975
- "loss": 2.5333,
1976
- "step": 328000
1977
- },
1978
- {
1979
- "epoch": 0.95,
1980
- "learning_rate": 4.526828398350084e-05,
1981
- "loss": 2.5624,
1982
- "step": 329000
1983
- },
1984
- {
1985
- "epoch": 0.95,
1986
- "learning_rate": 4.525390186794917e-05,
1987
- "loss": 2.5308,
1988
- "step": 330000
1989
- },
1990
- {
1991
- "epoch": 0.95,
1992
- "learning_rate": 4.5239519752397505e-05,
1993
- "loss": 2.5311,
1994
- "step": 331000
1995
- },
1996
- {
1997
- "epoch": 0.95,
1998
- "learning_rate": 4.522513763684583e-05,
1999
- "loss": 2.5386,
2000
- "step": 332000
2001
- },
2002
- {
2003
- "epoch": 0.96,
2004
- "learning_rate": 4.5210755521294163e-05,
2005
- "loss": 2.5355,
2006
- "step": 333000
2007
- },
2008
- {
2009
- "epoch": 0.96,
2010
- "learning_rate": 4.519637340574249e-05,
2011
- "loss": 2.5306,
2012
- "step": 334000
2013
- },
2014
- {
2015
- "epoch": 0.96,
2016
- "learning_rate": 4.518199129019082e-05,
2017
- "loss": 2.5379,
2018
- "step": 335000
2019
- },
2020
- {
2021
- "epoch": 0.97,
2022
- "learning_rate": 4.5167609174639155e-05,
2023
- "loss": 2.5422,
2024
- "step": 336000
2025
- },
2026
- {
2027
- "epoch": 0.97,
2028
- "learning_rate": 4.515322705908749e-05,
2029
- "loss": 2.5435,
2030
- "step": 337000
2031
- },
2032
- {
2033
- "epoch": 0.97,
2034
- "learning_rate": 4.513884494353582e-05,
2035
- "loss": 2.5258,
2036
- "step": 338000
2037
- },
2038
- {
2039
- "epoch": 0.98,
2040
- "learning_rate": 4.5124462827984146e-05,
2041
- "loss": 2.5131,
2042
- "step": 339000
2043
- },
2044
- {
2045
- "epoch": 0.98,
2046
- "learning_rate": 4.511008071243248e-05,
2047
- "loss": 2.5621,
2048
- "step": 340000
2049
- },
2050
- {
2051
- "epoch": 0.98,
2052
- "learning_rate": 4.509569859688081e-05,
2053
- "loss": 2.5329,
2054
- "step": 341000
2055
- },
2056
- {
2057
- "epoch": 0.98,
2058
- "learning_rate": 4.508131648132914e-05,
2059
- "loss": 2.5371,
2060
- "step": 342000
2061
- },
2062
- {
2063
- "epoch": 0.99,
2064
- "learning_rate": 4.506693436577747e-05,
2065
- "loss": 2.5115,
2066
- "step": 343000
2067
- },
2068
- {
2069
- "epoch": 0.99,
2070
- "learning_rate": 4.5052552250225796e-05,
2071
- "loss": 2.5354,
2072
- "step": 344000
2073
- },
2074
- {
2075
- "epoch": 0.99,
2076
- "learning_rate": 4.5038170134674135e-05,
2077
- "loss": 2.5223,
2078
- "step": 345000
2079
- },
2080
- {
2081
- "epoch": 1.0,
2082
- "learning_rate": 4.502378801912247e-05,
2083
- "loss": 2.5349,
2084
- "step": 346000
2085
- },
2086
- {
2087
- "epoch": 1.0,
2088
- "learning_rate": 4.5009405903570794e-05,
2089
- "loss": 2.5495,
2090
- "step": 347000
2091
- },
2092
- {
2093
- "epoch": 1.0,
2094
- "learning_rate": 4.499502378801913e-05,
2095
- "loss": 2.5201,
2096
- "step": 348000
2097
- },
2098
- {
2099
- "epoch": 1.0,
2100
- "learning_rate": 4.498064167246745e-05,
2101
- "loss": 2.5252,
2102
- "step": 349000
2103
- },
2104
- {
2105
- "epoch": 1.01,
2106
- "learning_rate": 4.4966259556915785e-05,
2107
- "loss": 2.5061,
2108
- "step": 350000
2109
- },
2110
- {
2111
- "epoch": 1.01,
2112
- "learning_rate": 4.495187744136412e-05,
2113
- "loss": 2.5005,
2114
- "step": 351000
2115
- },
2116
- {
2117
- "epoch": 1.01,
2118
- "learning_rate": 4.4937495325812444e-05,
2119
- "loss": 2.5299,
2120
- "step": 352000
2121
- },
2122
- {
2123
- "epoch": 1.02,
2124
- "learning_rate": 4.492311321026078e-05,
2125
- "loss": 2.5187,
2126
- "step": 353000
2127
- },
2128
- {
2129
- "epoch": 1.02,
2130
- "learning_rate": 4.490873109470911e-05,
2131
- "loss": 2.5216,
2132
- "step": 354000
2133
- },
2134
- {
2135
- "epoch": 1.02,
2136
- "learning_rate": 4.489434897915744e-05,
2137
- "loss": 2.5011,
2138
- "step": 355000
2139
- },
2140
- {
2141
- "epoch": 1.02,
2142
- "learning_rate": 4.4879966863605775e-05,
2143
- "loss": 2.5166,
2144
- "step": 356000
2145
- },
2146
- {
2147
- "epoch": 1.03,
2148
- "learning_rate": 4.48655847480541e-05,
2149
- "loss": 2.5133,
2150
- "step": 357000
2151
- },
2152
- {
2153
- "epoch": 1.03,
2154
- "learning_rate": 4.485120263250243e-05,
2155
- "loss": 2.5075,
2156
- "step": 358000
2157
- },
2158
- {
2159
- "epoch": 1.03,
2160
- "learning_rate": 4.483682051695076e-05,
2161
- "loss": 2.4966,
2162
- "step": 359000
2163
- },
2164
- {
2165
- "epoch": 1.04,
2166
- "learning_rate": 4.482243840139909e-05,
2167
- "loss": 2.5041,
2168
- "step": 360000
2169
- },
2170
- {
2171
- "epoch": 1.04,
2172
- "learning_rate": 4.4808056285847424e-05,
2173
- "loss": 2.4995,
2174
- "step": 361000
2175
- },
2176
- {
2177
- "epoch": 1.04,
2178
- "learning_rate": 4.479367417029576e-05,
2179
- "loss": 2.496,
2180
- "step": 362000
2181
- },
2182
- {
2183
- "epoch": 1.04,
2184
- "learning_rate": 4.477929205474409e-05,
2185
- "loss": 2.4993,
2186
- "step": 363000
2187
- },
2188
- {
2189
- "epoch": 1.05,
2190
- "learning_rate": 4.4764909939192416e-05,
2191
- "loss": 2.4995,
2192
- "step": 364000
2193
- },
2194
- {
2195
- "epoch": 1.05,
2196
- "learning_rate": 4.475052782364075e-05,
2197
- "loss": 2.5134,
2198
- "step": 365000
2199
- },
2200
- {
2201
- "epoch": 1.05,
2202
- "learning_rate": 4.473614570808908e-05,
2203
- "loss": 2.5108,
2204
- "step": 366000
2205
- },
2206
- {
2207
- "epoch": 1.06,
2208
- "learning_rate": 4.472176359253741e-05,
2209
- "loss": 2.4982,
2210
- "step": 367000
2211
- },
2212
- {
2213
- "epoch": 1.06,
2214
- "learning_rate": 4.4707381476985747e-05,
2215
- "loss": 2.4773,
2216
- "step": 368000
2217
- },
2218
- {
2219
- "epoch": 1.06,
2220
- "learning_rate": 4.469299936143407e-05,
2221
- "loss": 2.4935,
2222
- "step": 369000
2223
- },
2224
- {
2225
- "epoch": 1.06,
2226
- "learning_rate": 4.4678617245882405e-05,
2227
- "loss": 2.5007,
2228
- "step": 370000
2229
- },
2230
- {
2231
- "epoch": 1.07,
2232
- "learning_rate": 4.466423513033073e-05,
2233
- "loss": 2.4817,
2234
- "step": 371000
2235
- },
2236
- {
2237
- "epoch": 1.07,
2238
- "learning_rate": 4.4649853014779064e-05,
2239
- "loss": 2.4933,
2240
- "step": 372000
2241
- },
2242
- {
2243
- "epoch": 1.07,
2244
- "learning_rate": 4.4635470899227396e-05,
2245
- "loss": 2.51,
2246
- "step": 373000
2247
- },
2248
- {
2249
- "epoch": 1.08,
2250
- "learning_rate": 4.462108878367572e-05,
2251
- "loss": 2.4925,
2252
- "step": 374000
2253
- },
2254
- {
2255
- "epoch": 1.08,
2256
- "learning_rate": 4.4606706668124055e-05,
2257
- "loss": 2.4853,
2258
- "step": 375000
2259
- },
2260
- {
2261
- "epoch": 1.08,
2262
- "learning_rate": 4.459232455257239e-05,
2263
- "loss": 2.4628,
2264
- "step": 376000
2265
- },
2266
- {
2267
- "epoch": 1.08,
2268
- "learning_rate": 4.457794243702072e-05,
2269
- "loss": 2.4939,
2270
- "step": 377000
2271
- },
2272
- {
2273
- "epoch": 1.09,
2274
- "learning_rate": 4.456356032146905e-05,
2275
- "loss": 2.5029,
2276
- "step": 378000
2277
- },
2278
- {
2279
- "epoch": 1.09,
2280
- "learning_rate": 4.454917820591738e-05,
2281
- "loss": 2.4981,
2282
- "step": 379000
2283
- },
2284
- {
2285
- "epoch": 1.09,
2286
- "learning_rate": 4.453479609036571e-05,
2287
- "loss": 2.5255,
2288
- "step": 380000
2289
- },
2290
- {
2291
- "epoch": 1.1,
2292
- "learning_rate": 4.452041397481404e-05,
2293
- "loss": 2.4915,
2294
- "step": 381000
2295
- },
2296
- {
2297
- "epoch": 1.1,
2298
- "learning_rate": 4.450603185926237e-05,
2299
- "loss": 2.4583,
2300
- "step": 382000
2301
- },
2302
- {
2303
- "epoch": 1.1,
2304
- "learning_rate": 4.44916497437107e-05,
2305
- "loss": 2.5027,
2306
- "step": 383000
2307
- },
2308
- {
2309
- "epoch": 1.1,
2310
- "learning_rate": 4.4477267628159036e-05,
2311
- "loss": 2.4858,
2312
- "step": 384000
2313
- },
2314
- {
2315
- "epoch": 1.11,
2316
- "learning_rate": 4.446288551260737e-05,
2317
- "loss": 2.4606,
2318
- "step": 385000
2319
- },
2320
- {
2321
- "epoch": 1.11,
2322
- "learning_rate": 4.4448503397055694e-05,
2323
- "loss": 2.4854,
2324
- "step": 386000
2325
- },
2326
- {
2327
- "epoch": 1.11,
2328
- "learning_rate": 4.443412128150403e-05,
2329
- "loss": 2.4883,
2330
- "step": 387000
2331
- },
2332
- {
2333
- "epoch": 1.12,
2334
- "learning_rate": 4.441973916595236e-05,
2335
- "loss": 2.4898,
2336
- "step": 388000
2337
- },
2338
- {
2339
- "epoch": 1.12,
2340
- "learning_rate": 4.4405357050400685e-05,
2341
- "loss": 2.4708,
2342
- "step": 389000
2343
- },
2344
- {
2345
- "epoch": 1.12,
2346
- "learning_rate": 4.439097493484902e-05,
2347
- "loss": 2.4783,
2348
- "step": 390000
2349
- },
2350
- {
2351
- "epoch": 1.12,
2352
- "learning_rate": 4.4376592819297344e-05,
2353
- "loss": 2.4862,
2354
- "step": 391000
2355
- },
2356
- {
2357
- "epoch": 1.13,
2358
- "learning_rate": 4.4362210703745684e-05,
2359
- "loss": 2.4691,
2360
- "step": 392000
2361
- },
2362
- {
2363
- "epoch": 1.13,
2364
- "learning_rate": 4.434782858819401e-05,
2365
- "loss": 2.4725,
2366
- "step": 393000
2367
- },
2368
- {
2369
- "epoch": 1.13,
2370
- "learning_rate": 4.433344647264234e-05,
2371
- "loss": 2.4866,
2372
- "step": 394000
2373
- },
2374
- {
2375
- "epoch": 1.14,
2376
- "learning_rate": 4.4319064357090675e-05,
2377
- "loss": 2.4885,
2378
- "step": 395000
2379
- },
2380
- {
2381
- "epoch": 1.14,
2382
- "learning_rate": 4.4304682241539e-05,
2383
- "loss": 2.4785,
2384
- "step": 396000
2385
- },
2386
- {
2387
- "epoch": 1.14,
2388
- "learning_rate": 4.4290300125987333e-05,
2389
- "loss": 2.4874,
2390
- "step": 397000
2391
- },
2392
- {
2393
- "epoch": 1.14,
2394
- "learning_rate": 4.4275918010435666e-05,
2395
- "loss": 2.4974,
2396
- "step": 398000
2397
- },
2398
- {
2399
- "epoch": 1.15,
2400
- "learning_rate": 4.4261535894884e-05,
2401
- "loss": 2.4739,
2402
- "step": 399000
2403
- },
2404
- {
2405
- "epoch": 1.15,
2406
- "learning_rate": 4.424715377933233e-05,
2407
- "loss": 2.4512,
2408
- "step": 400000
2409
- },
2410
- {
2411
- "epoch": 1.15,
2412
- "learning_rate": 4.423277166378066e-05,
2413
- "loss": 2.462,
2414
- "step": 401000
2415
- },
2416
- {
2417
- "epoch": 1.16,
2418
- "learning_rate": 4.421838954822899e-05,
2419
- "loss": 2.4745,
2420
- "step": 402000
2421
- },
2422
- {
2423
- "epoch": 1.16,
2424
- "learning_rate": 4.4204007432677316e-05,
2425
- "loss": 2.481,
2426
- "step": 403000
2427
- },
2428
- {
2429
- "epoch": 1.16,
2430
- "learning_rate": 4.418962531712565e-05,
2431
- "loss": 2.4491,
2432
- "step": 404000
2433
- },
2434
- {
2435
- "epoch": 1.16,
2436
- "learning_rate": 4.417524320157398e-05,
2437
- "loss": 2.476,
2438
- "step": 405000
2439
- },
2440
- {
2441
- "epoch": 1.17,
2442
- "learning_rate": 4.416086108602231e-05,
2443
- "loss": 2.4731,
2444
- "step": 406000
2445
- },
2446
- {
2447
- "epoch": 1.17,
2448
- "learning_rate": 4.414647897047065e-05,
2449
- "loss": 2.484,
2450
- "step": 407000
2451
- },
2452
- {
2453
- "epoch": 1.17,
2454
- "learning_rate": 4.413209685491897e-05,
2455
- "loss": 2.47,
2456
- "step": 408000
2457
- },
2458
- {
2459
- "epoch": 1.18,
2460
- "learning_rate": 4.4117714739367305e-05,
2461
- "loss": 2.4504,
2462
- "step": 409000
2463
- },
2464
- {
2465
- "epoch": 1.18,
2466
- "learning_rate": 4.410333262381564e-05,
2467
- "loss": 2.4609,
2468
- "step": 410000
2469
- },
2470
- {
2471
- "epoch": 1.18,
2472
- "learning_rate": 4.4088950508263964e-05,
2473
- "loss": 2.4675,
2474
- "step": 411000
2475
- },
2476
- {
2477
- "epoch": 1.19,
2478
- "learning_rate": 4.4074568392712297e-05,
2479
- "loss": 2.4523,
2480
- "step": 412000
2481
- },
2482
- {
2483
- "epoch": 1.19,
2484
- "learning_rate": 4.406018627716062e-05,
2485
- "loss": 2.4336,
2486
- "step": 413000
2487
- },
2488
- {
2489
- "epoch": 1.19,
2490
- "learning_rate": 4.4045804161608955e-05,
2491
- "loss": 2.4539,
2492
- "step": 414000
2493
- },
2494
- {
2495
- "epoch": 1.19,
2496
- "learning_rate": 4.403142204605729e-05,
2497
- "loss": 2.459,
2498
- "step": 415000
2499
- },
2500
- {
2501
- "epoch": 1.2,
2502
- "learning_rate": 4.401703993050562e-05,
2503
- "loss": 2.4674,
2504
- "step": 416000
2505
- },
2506
- {
2507
- "epoch": 1.2,
2508
- "learning_rate": 4.400265781495395e-05,
2509
- "loss": 2.4455,
2510
- "step": 417000
2511
- },
2512
- {
2513
- "epoch": 1.2,
2514
- "learning_rate": 4.398827569940228e-05,
2515
- "loss": 2.4561,
2516
- "step": 418000
2517
- },
2518
- {
2519
- "epoch": 1.21,
2520
- "learning_rate": 4.397389358385061e-05,
2521
- "loss": 2.4597,
2522
- "step": 419000
2523
- },
2524
- {
2525
- "epoch": 1.21,
2526
- "learning_rate": 4.3959511468298945e-05,
2527
- "loss": 2.4555,
2528
- "step": 420000
2529
- },
2530
- {
2531
- "epoch": 1.21,
2532
- "learning_rate": 4.394512935274727e-05,
2533
- "loss": 2.4495,
2534
- "step": 421000
2535
- },
2536
- {
2537
- "epoch": 1.21,
2538
- "learning_rate": 4.39307472371956e-05,
2539
- "loss": 2.4346,
2540
- "step": 422000
2541
- },
2542
- {
2543
- "epoch": 1.22,
2544
- "learning_rate": 4.3916365121643936e-05,
2545
- "loss": 2.4472,
2546
- "step": 423000
2547
- },
2548
- {
2549
- "epoch": 1.22,
2550
- "learning_rate": 4.390198300609227e-05,
2551
- "loss": 2.4263,
2552
- "step": 424000
2553
- },
2554
- {
2555
- "epoch": 1.22,
2556
- "learning_rate": 4.3887600890540594e-05,
2557
- "loss": 2.4298,
2558
- "step": 425000
2559
- },
2560
- {
2561
- "epoch": 1.23,
2562
- "learning_rate": 4.387321877498893e-05,
2563
- "loss": 2.4524,
2564
- "step": 426000
2565
- },
2566
- {
2567
- "epoch": 1.23,
2568
- "learning_rate": 4.385883665943726e-05,
2569
- "loss": 2.4418,
2570
- "step": 427000
2571
- },
2572
- {
2573
- "epoch": 1.23,
2574
- "learning_rate": 4.3844454543885586e-05,
2575
- "loss": 2.4382,
2576
- "step": 428000
2577
- },
2578
- {
2579
- "epoch": 1.23,
2580
- "learning_rate": 4.383007242833392e-05,
2581
- "loss": 2.471,
2582
- "step": 429000
2583
- },
2584
- {
2585
- "epoch": 1.24,
2586
- "learning_rate": 4.381569031278225e-05,
2587
- "loss": 2.4435,
2588
- "step": 430000
2589
- },
2590
- {
2591
- "epoch": 1.24,
2592
- "learning_rate": 4.3801308197230584e-05,
2593
- "loss": 2.4456,
2594
- "step": 431000
2595
- },
2596
- {
2597
- "epoch": 1.24,
2598
- "learning_rate": 4.3786926081678916e-05,
2599
- "loss": 2.4342,
2600
- "step": 432000
2601
- },
2602
- {
2603
- "epoch": 1.25,
2604
- "learning_rate": 4.377254396612724e-05,
2605
- "loss": 2.4396,
2606
- "step": 433000
2607
- },
2608
- {
2609
- "epoch": 1.25,
2610
- "learning_rate": 4.3758161850575575e-05,
2611
- "loss": 2.4427,
2612
- "step": 434000
2613
- },
2614
- {
2615
- "epoch": 1.25,
2616
- "learning_rate": 4.37437797350239e-05,
2617
- "loss": 2.4555,
2618
- "step": 435000
2619
- },
2620
- {
2621
- "epoch": 1.25,
2622
- "learning_rate": 4.3729397619472234e-05,
2623
- "loss": 2.4377,
2624
- "step": 436000
2625
- },
2626
- {
2627
- "epoch": 1.26,
2628
- "learning_rate": 4.3715015503920566e-05,
2629
- "loss": 2.4502,
2630
- "step": 437000
2631
- },
2632
- {
2633
- "epoch": 1.26,
2634
- "learning_rate": 4.37006333883689e-05,
2635
- "loss": 2.4395,
2636
- "step": 438000
2637
- },
2638
- {
2639
- "epoch": 1.26,
2640
- "learning_rate": 4.368625127281723e-05,
2641
- "loss": 2.4351,
2642
- "step": 439000
2643
- },
2644
- {
2645
- "epoch": 1.27,
2646
- "learning_rate": 4.367186915726556e-05,
2647
- "loss": 2.4404,
2648
- "step": 440000
2649
- },
2650
- {
2651
- "epoch": 1.27,
2652
- "learning_rate": 4.365748704171389e-05,
2653
- "loss": 2.4291,
2654
- "step": 441000
2655
- },
2656
- {
2657
- "epoch": 1.27,
2658
- "learning_rate": 4.364310492616222e-05,
2659
- "loss": 2.4372,
2660
- "step": 442000
2661
- },
2662
- {
2663
- "epoch": 1.27,
2664
- "learning_rate": 4.362872281061055e-05,
2665
- "loss": 2.4499,
2666
- "step": 443000
2667
- },
2668
- {
2669
- "epoch": 1.28,
2670
- "learning_rate": 4.361434069505888e-05,
2671
- "loss": 2.4378,
2672
- "step": 444000
2673
- },
2674
- {
2675
- "epoch": 1.28,
2676
- "learning_rate": 4.359995857950721e-05,
2677
- "loss": 2.4231,
2678
- "step": 445000
2679
- },
2680
- {
2681
- "epoch": 1.28,
2682
- "learning_rate": 4.358557646395555e-05,
2683
- "loss": 2.4418,
2684
- "step": 446000
2685
- },
2686
- {
2687
- "epoch": 1.29,
2688
- "learning_rate": 4.357119434840388e-05,
2689
- "loss": 2.4457,
2690
- "step": 447000
2691
- },
2692
- {
2693
- "epoch": 1.29,
2694
- "learning_rate": 4.3556812232852206e-05,
2695
- "loss": 2.4349,
2696
- "step": 448000
2697
- },
2698
- {
2699
- "epoch": 1.29,
2700
- "learning_rate": 4.354243011730054e-05,
2701
- "loss": 2.4468,
2702
- "step": 449000
2703
- },
2704
- {
2705
- "epoch": 1.29,
2706
- "learning_rate": 4.3528048001748864e-05,
2707
- "loss": 2.4302,
2708
- "step": 450000
2709
- },
2710
- {
2711
- "epoch": 1.3,
2712
- "learning_rate": 4.35136658861972e-05,
2713
- "loss": 2.4558,
2714
- "step": 451000
2715
- },
2716
- {
2717
- "epoch": 1.3,
2718
- "learning_rate": 4.349928377064553e-05,
2719
- "loss": 2.4168,
2720
- "step": 452000
2721
- },
2722
- {
2723
- "epoch": 1.3,
2724
- "learning_rate": 4.3484901655093855e-05,
2725
- "loss": 2.4373,
2726
- "step": 453000
2727
- },
2728
- {
2729
- "epoch": 1.31,
2730
- "learning_rate": 4.3470519539542195e-05,
2731
- "loss": 2.4372,
2732
- "step": 454000
2733
- },
2734
- {
2735
- "epoch": 1.31,
2736
- "learning_rate": 4.345613742399052e-05,
2737
- "loss": 2.428,
2738
- "step": 455000
2739
- },
2740
- {
2741
- "epoch": 1.31,
2742
- "learning_rate": 4.3441755308438853e-05,
2743
- "loss": 2.4319,
2744
- "step": 456000
2745
- },
2746
- {
2747
- "epoch": 1.31,
2748
- "learning_rate": 4.3427373192887186e-05,
2749
- "loss": 2.4313,
2750
- "step": 457000
2751
- },
2752
- {
2753
- "epoch": 1.32,
2754
- "learning_rate": 4.341299107733551e-05,
2755
- "loss": 2.4237,
2756
- "step": 458000
2757
- },
2758
- {
2759
- "epoch": 1.32,
2760
- "learning_rate": 4.3398608961783845e-05,
2761
- "loss": 2.4286,
2762
- "step": 459000
2763
- },
2764
- {
2765
- "epoch": 1.32,
2766
- "learning_rate": 4.338422684623217e-05,
2767
- "loss": 2.4227,
2768
- "step": 460000
2769
- },
2770
- {
2771
- "epoch": 1.33,
2772
- "learning_rate": 4.336984473068051e-05,
2773
- "loss": 2.4341,
2774
- "step": 461000
2775
- },
2776
- {
2777
- "epoch": 1.33,
2778
- "learning_rate": 4.3355462615128836e-05,
2779
- "loss": 2.4209,
2780
- "step": 462000
2781
- },
2782
- {
2783
- "epoch": 1.33,
2784
- "learning_rate": 4.334108049957717e-05,
2785
- "loss": 2.4332,
2786
- "step": 463000
2787
- },
2788
- {
2789
- "epoch": 1.33,
2790
- "learning_rate": 4.33266983840255e-05,
2791
- "loss": 2.4244,
2792
- "step": 464000
2793
- },
2794
- {
2795
- "epoch": 1.34,
2796
- "learning_rate": 4.331231626847383e-05,
2797
- "loss": 2.4186,
2798
- "step": 465000
2799
- },
2800
- {
2801
- "epoch": 1.34,
2802
- "learning_rate": 4.329793415292216e-05,
2803
- "loss": 2.4061,
2804
- "step": 466000
2805
- },
2806
- {
2807
- "epoch": 1.34,
2808
- "learning_rate": 4.328355203737049e-05,
2809
- "loss": 2.4194,
2810
- "step": 467000
2811
- },
2812
- {
2813
- "epoch": 1.35,
2814
- "learning_rate": 4.326916992181882e-05,
2815
- "loss": 2.4048,
2816
- "step": 468000
2817
- },
2818
- {
2819
- "epoch": 1.35,
2820
- "learning_rate": 4.325478780626716e-05,
2821
- "loss": 2.4199,
2822
- "step": 469000
2823
- },
2824
- {
2825
- "epoch": 1.35,
2826
- "learning_rate": 4.3240405690715484e-05,
2827
- "loss": 2.4074,
2828
- "step": 470000
2829
- },
2830
- {
2831
- "epoch": 1.35,
2832
- "learning_rate": 4.322602357516382e-05,
2833
- "loss": 2.423,
2834
- "step": 471000
2835
- },
2836
- {
2837
- "epoch": 1.36,
2838
- "learning_rate": 4.321164145961214e-05,
2839
- "loss": 2.4054,
2840
- "step": 472000
2841
- },
2842
- {
2843
- "epoch": 1.36,
2844
- "learning_rate": 4.3197259344060475e-05,
2845
- "loss": 2.4045,
2846
- "step": 473000
2847
- },
2848
- {
2849
- "epoch": 1.36,
2850
- "learning_rate": 4.318287722850881e-05,
2851
- "loss": 2.4297,
2852
- "step": 474000
2853
- },
2854
- {
2855
- "epoch": 1.37,
2856
- "learning_rate": 4.3168495112957134e-05,
2857
- "loss": 2.4086,
2858
- "step": 475000
2859
- },
2860
- {
2861
- "epoch": 1.37,
2862
- "learning_rate": 4.3154112997405467e-05,
2863
- "loss": 2.4133,
2864
- "step": 476000
2865
- },
2866
- {
2867
- "epoch": 1.37,
2868
- "learning_rate": 4.31397308818538e-05,
2869
- "loss": 2.411,
2870
- "step": 477000
2871
- },
2872
- {
2873
- "epoch": 1.37,
2874
- "learning_rate": 4.312534876630213e-05,
2875
- "loss": 2.4079,
2876
- "step": 478000
2877
- },
2878
- {
2879
- "epoch": 1.38,
2880
- "learning_rate": 4.3110966650750465e-05,
2881
- "loss": 2.4244,
2882
- "step": 479000
2883
- },
2884
- {
2885
- "epoch": 1.38,
2886
- "learning_rate": 4.309658453519879e-05,
2887
- "loss": 2.4178,
2888
- "step": 480000
2889
- },
2890
- {
2891
- "epoch": 1.38,
2892
- "learning_rate": 4.308220241964712e-05,
2893
- "loss": 2.4149,
2894
- "step": 481000
2895
- },
2896
- {
2897
- "epoch": 1.39,
2898
- "learning_rate": 4.306782030409545e-05,
2899
- "loss": 2.4193,
2900
- "step": 482000
2901
- },
2902
- {
2903
- "epoch": 1.39,
2904
- "learning_rate": 4.305343818854378e-05,
2905
- "loss": 2.3833,
2906
- "step": 483000
2907
- },
2908
- {
2909
- "epoch": 1.39,
2910
- "learning_rate": 4.3039056072992114e-05,
2911
- "loss": 2.4018,
2912
- "step": 484000
2913
- },
2914
- {
2915
- "epoch": 1.4,
2916
- "learning_rate": 4.302467395744045e-05,
2917
- "loss": 2.402,
2918
- "step": 485000
2919
- },
2920
- {
2921
- "epoch": 1.4,
2922
- "learning_rate": 4.301029184188878e-05,
2923
- "loss": 2.4102,
2924
- "step": 486000
2925
- },
2926
- {
2927
- "epoch": 1.4,
2928
- "learning_rate": 4.2995909726337106e-05,
2929
- "loss": 2.4033,
2930
- "step": 487000
2931
- },
2932
- {
2933
- "epoch": 1.4,
2934
- "learning_rate": 4.298152761078544e-05,
2935
- "loss": 2.4074,
2936
- "step": 488000
2937
- },
2938
- {
2939
- "epoch": 1.41,
2940
- "learning_rate": 4.296714549523377e-05,
2941
- "loss": 2.4216,
2942
- "step": 489000
2943
- },
2944
- {
2945
- "epoch": 1.41,
2946
- "learning_rate": 4.29527633796821e-05,
2947
- "loss": 2.4083,
2948
- "step": 490000
2949
- },
2950
- {
2951
- "epoch": 1.41,
2952
- "learning_rate": 4.293838126413043e-05,
2953
- "loss": 2.4119,
2954
- "step": 491000
2955
- },
2956
- {
2957
- "epoch": 1.42,
2958
- "learning_rate": 4.292399914857876e-05,
2959
- "loss": 2.4182,
2960
- "step": 492000
2961
- },
2962
- {
2963
- "epoch": 1.42,
2964
- "learning_rate": 4.2909617033027095e-05,
2965
- "loss": 2.4282,
2966
- "step": 493000
2967
- },
2968
- {
2969
- "epoch": 1.42,
2970
- "learning_rate": 4.289523491747542e-05,
2971
- "loss": 2.3958,
2972
- "step": 494000
2973
- },
2974
- {
2975
- "epoch": 1.42,
2976
- "learning_rate": 4.2880852801923754e-05,
2977
- "loss": 2.4014,
2978
- "step": 495000
2979
- },
2980
- {
2981
- "epoch": 1.43,
2982
- "learning_rate": 4.2866470686372086e-05,
2983
- "loss": 2.4126,
2984
- "step": 496000
2985
- },
2986
- {
2987
- "epoch": 1.43,
2988
- "learning_rate": 4.285208857082041e-05,
2989
- "loss": 2.405,
2990
- "step": 497000
2991
- },
2992
- {
2993
- "epoch": 1.43,
2994
- "learning_rate": 4.2837706455268745e-05,
2995
- "loss": 2.4054,
2996
- "step": 498000
2997
- },
2998
- {
2999
- "epoch": 1.44,
3000
- "learning_rate": 4.282332433971708e-05,
3001
- "loss": 2.3909,
3002
- "step": 499000
3003
- },
3004
- {
3005
- "epoch": 1.44,
3006
- "learning_rate": 4.280894222416541e-05,
3007
- "loss": 2.4034,
3008
- "step": 500000
3009
- },
3010
- {
3011
- "epoch": 1.44,
3012
- "learning_rate": 4.279456010861374e-05,
3013
- "loss": 2.4091,
3014
- "step": 501000
3015
- },
3016
- {
3017
- "epoch": 1.44,
3018
- "learning_rate": 4.278017799306207e-05,
3019
- "loss": 2.4143,
3020
- "step": 502000
3021
- },
3022
- {
3023
- "epoch": 1.45,
3024
- "learning_rate": 4.27657958775104e-05,
3025
- "loss": 2.3962,
3026
- "step": 503000
3027
- },
3028
- {
3029
- "epoch": 1.45,
3030
- "learning_rate": 4.275141376195873e-05,
3031
- "loss": 2.4084,
3032
- "step": 504000
3033
- },
3034
- {
3035
- "epoch": 1.45,
3036
- "learning_rate": 4.273703164640706e-05,
3037
- "loss": 2.3883,
3038
- "step": 505000
3039
- },
3040
- {
3041
- "epoch": 1.46,
3042
- "learning_rate": 4.272264953085539e-05,
3043
- "loss": 2.3972,
3044
- "step": 506000
3045
- },
3046
- {
3047
- "epoch": 1.46,
3048
- "learning_rate": 4.270826741530372e-05,
3049
- "loss": 2.4122,
3050
- "step": 507000
3051
- },
3052
- {
3053
- "epoch": 1.46,
3054
- "learning_rate": 4.269388529975206e-05,
3055
- "loss": 2.4116,
3056
- "step": 508000
3057
- },
3058
- {
3059
- "epoch": 1.46,
3060
- "learning_rate": 4.2679503184200384e-05,
3061
- "loss": 2.3773,
3062
- "step": 509000
3063
- },
3064
- {
3065
- "epoch": 1.47,
3066
- "learning_rate": 4.266512106864872e-05,
3067
- "loss": 2.3815,
3068
- "step": 510000
3069
- },
3070
- {
3071
- "epoch": 1.47,
3072
- "learning_rate": 4.265073895309705e-05,
3073
- "loss": 2.4002,
3074
- "step": 511000
3075
- },
3076
- {
3077
- "epoch": 1.47,
3078
- "learning_rate": 4.2636356837545375e-05,
3079
- "loss": 2.3904,
3080
- "step": 512000
3081
- },
3082
- {
3083
- "epoch": 1.48,
3084
- "learning_rate": 4.262197472199371e-05,
3085
- "loss": 2.3991,
3086
- "step": 513000
3087
- },
3088
- {
3089
- "epoch": 1.48,
3090
- "learning_rate": 4.2607592606442034e-05,
3091
- "loss": 2.416,
3092
- "step": 514000
3093
- },
3094
- {
3095
- "epoch": 1.48,
3096
- "learning_rate": 4.259321049089037e-05,
3097
- "loss": 2.3864,
3098
- "step": 515000
3099
- },
3100
- {
3101
- "epoch": 1.48,
3102
- "learning_rate": 4.2578828375338706e-05,
3103
- "loss": 2.3973,
3104
- "step": 516000
3105
- },
3106
- {
3107
- "epoch": 1.49,
3108
- "learning_rate": 4.256444625978703e-05,
3109
- "loss": 2.3902,
3110
- "step": 517000
3111
- },
3112
- {
3113
- "epoch": 1.49,
3114
- "learning_rate": 4.2550064144235365e-05,
3115
- "loss": 2.408,
3116
- "step": 518000
3117
- },
3118
- {
3119
- "epoch": 1.49,
3120
- "learning_rate": 4.253568202868369e-05,
3121
- "loss": 2.408,
3122
- "step": 519000
3123
- },
3124
- {
3125
- "epoch": 1.5,
3126
- "learning_rate": 4.2521299913132023e-05,
3127
- "loss": 2.3921,
3128
- "step": 520000
3129
- },
3130
- {
3131
- "epoch": 1.5,
3132
- "learning_rate": 4.2506917797580356e-05,
3133
- "loss": 2.3897,
3134
- "step": 521000
3135
- },
3136
- {
3137
- "epoch": 1.5,
3138
- "learning_rate": 4.249253568202868e-05,
3139
- "loss": 2.3883,
3140
- "step": 522000
3141
- },
3142
- {
3143
- "epoch": 1.5,
3144
- "learning_rate": 4.247815356647702e-05,
3145
- "loss": 2.4108,
3146
- "step": 523000
3147
- },
3148
- {
3149
- "epoch": 1.51,
3150
- "learning_rate": 4.246377145092535e-05,
3151
- "loss": 2.3893,
3152
- "step": 524000
3153
- },
3154
- {
3155
- "epoch": 1.51,
3156
- "learning_rate": 4.244938933537368e-05,
3157
- "loss": 2.386,
3158
- "step": 525000
3159
- },
3160
- {
3161
- "epoch": 1.51,
3162
- "learning_rate": 4.243500721982201e-05,
3163
- "loss": 2.3724,
3164
- "step": 526000
3165
- },
3166
- {
3167
- "epoch": 1.52,
3168
- "learning_rate": 4.242062510427034e-05,
3169
- "loss": 2.3829,
3170
- "step": 527000
3171
- },
3172
- {
3173
- "epoch": 1.52,
3174
- "learning_rate": 4.240624298871867e-05,
3175
- "loss": 2.4046,
3176
- "step": 528000
3177
- },
3178
- {
3179
- "epoch": 1.52,
3180
- "learning_rate": 4.2391860873167e-05,
3181
- "loss": 2.3907,
3182
- "step": 529000
3183
- },
3184
- {
3185
- "epoch": 1.52,
3186
- "learning_rate": 4.237747875761533e-05,
3187
- "loss": 2.3817,
3188
- "step": 530000
3189
- },
3190
- {
3191
- "epoch": 1.53,
3192
- "learning_rate": 4.236309664206366e-05,
3193
- "loss": 2.3762,
3194
- "step": 531000
3195
- },
3196
- {
3197
- "epoch": 1.53,
3198
- "learning_rate": 4.2348714526511995e-05,
3199
- "loss": 2.3632,
3200
- "step": 532000
3201
- },
3202
- {
3203
- "epoch": 1.53,
3204
- "learning_rate": 4.233433241096033e-05,
3205
- "loss": 2.3918,
3206
- "step": 533000
3207
- },
3208
- {
3209
- "epoch": 1.54,
3210
- "learning_rate": 4.2319950295408654e-05,
3211
- "loss": 2.3625,
3212
- "step": 534000
3213
- },
3214
- {
3215
- "epoch": 1.54,
3216
- "learning_rate": 4.2305568179856987e-05,
3217
- "loss": 2.3687,
3218
- "step": 535000
3219
- },
3220
- {
3221
- "epoch": 1.54,
3222
- "learning_rate": 4.229118606430532e-05,
3223
- "loss": 2.3758,
3224
- "step": 536000
3225
- },
3226
- {
3227
- "epoch": 1.54,
3228
- "learning_rate": 4.2276803948753645e-05,
3229
- "loss": 2.3797,
3230
- "step": 537000
3231
- },
3232
- {
3233
- "epoch": 1.55,
3234
- "learning_rate": 4.226242183320198e-05,
3235
- "loss": 2.3882,
3236
- "step": 538000
3237
- },
3238
- {
3239
- "epoch": 1.55,
3240
- "learning_rate": 4.224803971765031e-05,
3241
- "loss": 2.3859,
3242
- "step": 539000
3243
- },
3244
- {
3245
- "epoch": 1.55,
3246
- "learning_rate": 4.223365760209864e-05,
3247
- "loss": 2.3732,
3248
- "step": 540000
3249
- },
3250
- {
3251
- "epoch": 1.56,
3252
- "learning_rate": 4.221927548654697e-05,
3253
- "loss": 2.3737,
3254
- "step": 541000
3255
- },
3256
- {
3257
- "epoch": 1.56,
3258
- "learning_rate": 4.22048933709953e-05,
3259
- "loss": 2.3482,
3260
- "step": 542000
3261
- },
3262
- {
3263
- "epoch": 1.56,
3264
- "learning_rate": 4.2190511255443635e-05,
3265
- "loss": 2.3798,
3266
- "step": 543000
3267
- },
3268
- {
3269
- "epoch": 1.56,
3270
- "learning_rate": 4.217612913989196e-05,
3271
- "loss": 2.3628,
3272
- "step": 544000
3273
- },
3274
- {
3275
- "epoch": 1.57,
3276
- "learning_rate": 4.216174702434029e-05,
3277
- "loss": 2.39,
3278
- "step": 545000
3279
- },
3280
- {
3281
- "epoch": 1.57,
3282
- "learning_rate": 4.214736490878862e-05,
3283
- "loss": 2.3715,
3284
- "step": 546000
3285
- },
3286
- {
3287
- "epoch": 1.57,
3288
- "learning_rate": 4.213298279323696e-05,
3289
- "loss": 2.3636,
3290
- "step": 547000
3291
- },
3292
- {
3293
- "epoch": 1.58,
3294
- "learning_rate": 4.211860067768529e-05,
3295
- "loss": 2.3754,
3296
- "step": 548000
3297
- },
3298
- {
3299
- "epoch": 1.58,
3300
- "learning_rate": 4.210421856213362e-05,
3301
- "loss": 2.373,
3302
- "step": 549000
3303
- },
3304
- {
3305
- "epoch": 1.58,
3306
- "learning_rate": 4.208983644658195e-05,
3307
- "loss": 2.3609,
3308
- "step": 550000
3309
- },
3310
- {
3311
- "epoch": 1.58,
3312
- "learning_rate": 4.2075454331030276e-05,
3313
- "loss": 2.362,
3314
- "step": 551000
3315
- },
3316
- {
3317
- "epoch": 1.59,
3318
- "learning_rate": 4.206107221547861e-05,
3319
- "loss": 2.3691,
3320
- "step": 552000
3321
- },
3322
- {
3323
- "epoch": 1.59,
3324
- "learning_rate": 4.204669009992694e-05,
3325
- "loss": 2.3776,
3326
- "step": 553000
3327
- },
3328
- {
3329
- "epoch": 1.59,
3330
- "learning_rate": 4.2032307984375274e-05,
3331
- "loss": 2.362,
3332
- "step": 554000
3333
- },
3334
- {
3335
- "epoch": 1.6,
3336
- "learning_rate": 4.2017925868823606e-05,
3337
- "loss": 2.3763,
3338
- "step": 555000
3339
- },
3340
- {
3341
- "epoch": 1.6,
3342
- "learning_rate": 4.200354375327193e-05,
3343
- "loss": 2.3657,
3344
- "step": 556000
3345
- },
3346
- {
3347
- "epoch": 1.6,
3348
- "learning_rate": 4.1989161637720265e-05,
3349
- "loss": 2.3669,
3350
- "step": 557000
3351
- },
3352
- {
3353
- "epoch": 1.61,
3354
- "learning_rate": 4.19747795221686e-05,
3355
- "loss": 2.3735,
3356
- "step": 558000
3357
- },
3358
- {
3359
- "epoch": 1.61,
3360
- "learning_rate": 4.1960397406616924e-05,
3361
- "loss": 2.3724,
3362
- "step": 559000
3363
- },
3364
- {
3365
- "epoch": 1.61,
3366
- "learning_rate": 4.1946015291065256e-05,
3367
- "loss": 2.3772,
3368
- "step": 560000
3369
- },
3370
- {
3371
- "epoch": 1.61,
3372
- "learning_rate": 4.193163317551358e-05,
3373
- "loss": 2.3801,
3374
- "step": 561000
3375
- },
3376
- {
3377
- "epoch": 1.62,
3378
- "learning_rate": 4.191725105996192e-05,
3379
- "loss": 2.3565,
3380
- "step": 562000
3381
- },
3382
- {
3383
- "epoch": 1.62,
3384
- "learning_rate": 4.190286894441025e-05,
3385
- "loss": 2.3586,
3386
- "step": 563000
3387
- },
3388
- {
3389
- "epoch": 1.62,
3390
- "learning_rate": 4.188848682885858e-05,
3391
- "loss": 2.3583,
3392
- "step": 564000
3393
- },
3394
- {
3395
- "epoch": 1.63,
3396
- "learning_rate": 4.187410471330691e-05,
3397
- "loss": 2.3691,
3398
- "step": 565000
3399
- },
3400
- {
3401
- "epoch": 1.63,
3402
- "learning_rate": 4.185972259775524e-05,
3403
- "loss": 2.3624,
3404
- "step": 566000
3405
- },
3406
- {
3407
- "epoch": 1.63,
3408
- "learning_rate": 4.184534048220357e-05,
3409
- "loss": 2.3729,
3410
- "step": 567000
3411
- },
3412
- {
3413
- "epoch": 1.63,
3414
- "learning_rate": 4.1830958366651904e-05,
3415
- "loss": 2.3824,
3416
- "step": 568000
3417
- },
3418
- {
3419
- "epoch": 1.64,
3420
- "learning_rate": 4.181657625110023e-05,
3421
- "loss": 2.3762,
3422
- "step": 569000
3423
- },
3424
- {
3425
- "epoch": 1.64,
3426
- "learning_rate": 4.180219413554857e-05,
3427
- "loss": 2.3531,
3428
- "step": 570000
3429
- },
3430
- {
3431
- "epoch": 1.64,
3432
- "learning_rate": 4.1787812019996896e-05,
3433
- "loss": 2.3588,
3434
- "step": 571000
3435
- },
3436
- {
3437
- "epoch": 1.65,
3438
- "learning_rate": 4.177342990444523e-05,
3439
- "loss": 2.3387,
3440
- "step": 572000
3441
- },
3442
- {
3443
- "epoch": 1.65,
3444
- "learning_rate": 4.1759047788893554e-05,
3445
- "loss": 2.3401,
3446
- "step": 573000
3447
- },
3448
- {
3449
- "epoch": 1.65,
3450
- "learning_rate": 4.174466567334189e-05,
3451
- "loss": 2.3462,
3452
- "step": 574000
3453
- },
3454
- {
3455
- "epoch": 1.65,
3456
- "learning_rate": 4.173028355779022e-05,
3457
- "loss": 2.3523,
3458
- "step": 575000
3459
- },
3460
- {
3461
- "epoch": 1.66,
3462
- "learning_rate": 4.1715901442238545e-05,
3463
- "loss": 2.3447,
3464
- "step": 576000
3465
- },
3466
- {
3467
- "epoch": 1.66,
3468
- "learning_rate": 4.1701519326686885e-05,
3469
- "loss": 2.3783,
3470
- "step": 577000
3471
- },
3472
- {
3473
- "epoch": 1.66,
3474
- "learning_rate": 4.168713721113521e-05,
3475
- "loss": 2.3601,
3476
- "step": 578000
3477
- },
3478
- {
3479
- "epoch": 1.67,
3480
- "learning_rate": 4.1672755095583543e-05,
3481
- "loss": 2.3354,
3482
- "step": 579000
3483
- },
3484
- {
3485
- "epoch": 1.67,
3486
- "learning_rate": 4.1658372980031876e-05,
3487
- "loss": 2.3595,
3488
- "step": 580000
3489
- },
3490
- {
3491
- "epoch": 1.67,
3492
- "learning_rate": 4.16439908644802e-05,
3493
- "loss": 2.3532,
3494
- "step": 581000
3495
- },
3496
- {
3497
- "epoch": 1.67,
3498
- "learning_rate": 4.1629608748928535e-05,
3499
- "loss": 2.3697,
3500
- "step": 582000
3501
- },
3502
- {
3503
- "epoch": 1.68,
3504
- "learning_rate": 4.161522663337686e-05,
3505
- "loss": 2.3497,
3506
- "step": 583000
3507
- },
3508
- {
3509
- "epoch": 1.68,
3510
- "learning_rate": 4.160084451782519e-05,
3511
- "loss": 2.3429,
3512
- "step": 584000
3513
- },
3514
- {
3515
- "epoch": 1.68,
3516
- "learning_rate": 4.1586462402273526e-05,
3517
- "loss": 2.3513,
3518
- "step": 585000
3519
- },
3520
- {
3521
- "epoch": 1.69,
3522
- "learning_rate": 4.157208028672186e-05,
3523
- "loss": 2.363,
3524
- "step": 586000
3525
- },
3526
- {
3527
- "epoch": 1.69,
3528
- "learning_rate": 4.155769817117019e-05,
3529
- "loss": 2.3217,
3530
- "step": 587000
3531
- },
3532
- {
3533
- "epoch": 1.69,
3534
- "learning_rate": 4.154331605561852e-05,
3535
- "loss": 2.3594,
3536
- "step": 588000
3537
- },
3538
- {
3539
- "epoch": 1.69,
3540
- "learning_rate": 4.152893394006685e-05,
3541
- "loss": 2.3446,
3542
- "step": 589000
3543
- },
3544
- {
3545
- "epoch": 1.7,
3546
- "learning_rate": 4.151455182451518e-05,
3547
- "loss": 2.3493,
3548
- "step": 590000
3549
- },
3550
- {
3551
- "epoch": 1.7,
3552
- "learning_rate": 4.150016970896351e-05,
3553
- "loss": 2.3729,
3554
- "step": 591000
3555
- },
3556
- {
3557
- "epoch": 1.7,
3558
- "learning_rate": 4.148578759341184e-05,
3559
- "loss": 2.3466,
3560
- "step": 592000
3561
- },
3562
- {
3563
- "epoch": 1.71,
3564
- "learning_rate": 4.1471405477860174e-05,
3565
- "loss": 2.3496,
3566
- "step": 593000
3567
- },
3568
- {
3569
- "epoch": 1.71,
3570
- "learning_rate": 4.145702336230851e-05,
3571
- "loss": 2.3428,
3572
- "step": 594000
3573
- },
3574
- {
3575
- "epoch": 1.71,
3576
- "learning_rate": 4.144264124675683e-05,
3577
- "loss": 2.3488,
3578
- "step": 595000
3579
- },
3580
- {
3581
- "epoch": 1.71,
3582
- "learning_rate": 4.1428259131205165e-05,
3583
- "loss": 2.3517,
3584
- "step": 596000
3585
- },
3586
- {
3587
- "epoch": 1.72,
3588
- "learning_rate": 4.14138770156535e-05,
3589
- "loss": 2.3537,
3590
- "step": 597000
3591
- },
3592
- {
3593
- "epoch": 1.72,
3594
- "learning_rate": 4.1399494900101824e-05,
3595
- "loss": 2.3317,
3596
- "step": 598000
3597
- },
3598
- {
3599
- "epoch": 1.72,
3600
- "learning_rate": 4.1385112784550157e-05,
3601
- "loss": 2.3355,
3602
- "step": 599000
3603
- },
3604
- {
3605
- "epoch": 1.73,
3606
- "learning_rate": 4.137073066899849e-05,
3607
- "loss": 2.3676,
3608
- "step": 600000
3609
- },
3610
- {
3611
- "epoch": 1.73,
3612
- "learning_rate": 4.135634855344682e-05,
3613
- "loss": 2.3339,
3614
- "step": 601000
3615
- },
3616
- {
3617
- "epoch": 1.73,
3618
- "learning_rate": 4.1341966437895155e-05,
3619
- "loss": 2.3438,
3620
- "step": 602000
3621
- },
3622
- {
3623
- "epoch": 1.73,
3624
- "learning_rate": 4.132758432234348e-05,
3625
- "loss": 2.3417,
3626
- "step": 603000
3627
- },
3628
- {
3629
- "epoch": 1.74,
3630
- "learning_rate": 4.131320220679181e-05,
3631
- "loss": 2.3424,
3632
- "step": 604000
3633
- },
3634
- {
3635
- "epoch": 1.74,
3636
- "learning_rate": 4.129882009124014e-05,
3637
- "loss": 2.3437,
3638
- "step": 605000
3639
- },
3640
- {
3641
- "epoch": 1.74,
3642
- "learning_rate": 4.128443797568847e-05,
3643
- "loss": 2.3355,
3644
- "step": 606000
3645
- },
3646
- {
3647
- "epoch": 1.75,
3648
- "learning_rate": 4.1270055860136804e-05,
3649
- "loss": 2.3597,
3650
- "step": 607000
3651
- },
3652
- {
3653
- "epoch": 1.75,
3654
- "learning_rate": 4.125567374458514e-05,
3655
- "loss": 2.3304,
3656
- "step": 608000
3657
- },
3658
- {
3659
- "epoch": 1.75,
3660
- "learning_rate": 4.124129162903347e-05,
3661
- "loss": 2.3611,
3662
- "step": 609000
3663
- },
3664
- {
3665
- "epoch": 1.75,
3666
- "learning_rate": 4.1226909513481796e-05,
3667
- "loss": 2.3223,
3668
- "step": 610000
3669
- },
3670
- {
3671
- "epoch": 1.76,
3672
- "learning_rate": 4.121252739793013e-05,
3673
- "loss": 2.334,
3674
- "step": 611000
3675
- },
3676
- {
3677
- "epoch": 1.76,
3678
- "learning_rate": 4.119814528237846e-05,
3679
- "loss": 2.3246,
3680
- "step": 612000
3681
- },
3682
- {
3683
- "epoch": 1.76,
3684
- "learning_rate": 4.118376316682679e-05,
3685
- "loss": 2.3517,
3686
- "step": 613000
3687
- },
3688
- {
3689
- "epoch": 1.77,
3690
- "learning_rate": 4.116938105127512e-05,
3691
- "loss": 2.3431,
3692
- "step": 614000
3693
- },
3694
- {
3695
- "epoch": 1.77,
3696
- "learning_rate": 4.1154998935723446e-05,
3697
- "loss": 2.3256,
3698
- "step": 615000
3699
- },
3700
- {
3701
- "epoch": 1.77,
3702
- "learning_rate": 4.1140616820171785e-05,
3703
- "loss": 2.3464,
3704
- "step": 616000
3705
- },
3706
- {
3707
- "epoch": 1.77,
3708
- "learning_rate": 4.112623470462012e-05,
3709
- "loss": 2.3388,
3710
- "step": 617000
3711
- },
3712
- {
3713
- "epoch": 1.78,
3714
- "learning_rate": 4.1111852589068444e-05,
3715
- "loss": 2.3607,
3716
- "step": 618000
3717
- },
3718
- {
3719
- "epoch": 1.78,
3720
- "learning_rate": 4.1097470473516776e-05,
3721
- "loss": 2.3413,
3722
- "step": 619000
3723
- },
3724
- {
3725
- "epoch": 1.78,
3726
- "learning_rate": 4.10830883579651e-05,
3727
- "loss": 2.333,
3728
- "step": 620000
3729
- },
3730
- {
3731
- "epoch": 1.79,
3732
- "learning_rate": 4.1068706242413435e-05,
3733
- "loss": 2.3273,
3734
- "step": 621000
3735
- },
3736
- {
3737
- "epoch": 1.79,
3738
- "learning_rate": 4.105432412686177e-05,
3739
- "loss": 2.3172,
3740
- "step": 622000
3741
- },
3742
- {
3743
- "epoch": 1.79,
3744
- "learning_rate": 4.1039942011310094e-05,
3745
- "loss": 2.3441,
3746
- "step": 623000
3747
- },
3748
- {
3749
- "epoch": 1.79,
3750
- "learning_rate": 4.102555989575843e-05,
3751
- "loss": 2.3275,
3752
- "step": 624000
3753
- },
3754
- {
3755
- "epoch": 1.8,
3756
- "learning_rate": 4.101117778020676e-05,
3757
- "loss": 2.3418,
3758
- "step": 625000
3759
- },
3760
- {
3761
- "epoch": 1.8,
3762
- "learning_rate": 4.099679566465509e-05,
3763
- "loss": 2.3062,
3764
- "step": 626000
3765
- },
3766
- {
3767
- "epoch": 1.8,
3768
- "learning_rate": 4.0982413549103424e-05,
3769
- "loss": 2.3245,
3770
- "step": 627000
3771
- },
3772
- {
3773
- "epoch": 1.81,
3774
- "learning_rate": 4.096803143355175e-05,
3775
- "loss": 2.3102,
3776
- "step": 628000
3777
- },
3778
- {
3779
- "epoch": 1.81,
3780
- "learning_rate": 4.095364931800008e-05,
3781
- "loss": 2.3484,
3782
- "step": 629000
3783
- },
3784
- {
3785
- "epoch": 1.81,
3786
- "learning_rate": 4.093926720244841e-05,
3787
- "loss": 2.3148,
3788
- "step": 630000
3789
- },
3790
- {
3791
- "epoch": 1.82,
3792
- "learning_rate": 4.092488508689674e-05,
3793
- "loss": 2.3294,
3794
- "step": 631000
3795
- },
3796
- {
3797
- "epoch": 1.82,
3798
- "learning_rate": 4.0910502971345074e-05,
3799
- "loss": 2.3221,
3800
- "step": 632000
3801
- },
3802
- {
3803
- "epoch": 1.82,
3804
- "learning_rate": 4.089612085579341e-05,
3805
- "loss": 2.3226,
3806
- "step": 633000
3807
- },
3808
- {
3809
- "epoch": 1.82,
3810
- "learning_rate": 4.088173874024174e-05,
3811
- "loss": 2.3101,
3812
- "step": 634000
3813
- },
3814
- {
3815
- "epoch": 1.83,
3816
- "learning_rate": 4.0867356624690065e-05,
3817
- "loss": 2.3166,
3818
- "step": 635000
3819
- },
3820
- {
3821
- "epoch": 1.83,
3822
- "learning_rate": 4.08529745091384e-05,
3823
- "loss": 2.3277,
3824
- "step": 636000
3825
- },
3826
- {
3827
- "epoch": 1.83,
3828
- "learning_rate": 4.083859239358673e-05,
3829
- "loss": 2.3134,
3830
- "step": 637000
3831
- },
3832
- {
3833
- "epoch": 1.84,
3834
- "learning_rate": 4.082421027803506e-05,
3835
- "loss": 2.3346,
3836
- "step": 638000
3837
- },
3838
- {
3839
- "epoch": 1.84,
3840
- "learning_rate": 4.0809828162483396e-05,
3841
- "loss": 2.3387,
3842
- "step": 639000
3843
- },
3844
- {
3845
- "epoch": 1.84,
3846
- "learning_rate": 4.079544604693172e-05,
3847
- "loss": 2.3386,
3848
- "step": 640000
3849
- },
3850
- {
3851
- "epoch": 1.84,
3852
- "learning_rate": 4.0781063931380055e-05,
3853
- "loss": 2.3282,
3854
- "step": 641000
3855
- },
3856
- {
3857
- "epoch": 1.85,
3858
- "learning_rate": 4.076668181582838e-05,
3859
- "loss": 2.3052,
3860
- "step": 642000
3861
- },
3862
- {
3863
- "epoch": 1.85,
3864
- "learning_rate": 4.0752299700276713e-05,
3865
- "loss": 2.3323,
3866
- "step": 643000
3867
- },
3868
- {
3869
- "epoch": 1.85,
3870
- "learning_rate": 4.0737917584725046e-05,
3871
- "loss": 2.3219,
3872
- "step": 644000
3873
- },
3874
- {
3875
- "epoch": 1.86,
3876
- "learning_rate": 4.072353546917337e-05,
3877
- "loss": 2.3292,
3878
- "step": 645000
3879
- },
3880
- {
3881
- "epoch": 1.86,
3882
- "learning_rate": 4.0709153353621705e-05,
3883
- "loss": 2.3082,
3884
- "step": 646000
3885
- },
3886
- {
3887
- "epoch": 1.86,
3888
- "learning_rate": 4.069477123807004e-05,
3889
- "loss": 2.3266,
3890
- "step": 647000
3891
- },
3892
- {
3893
- "epoch": 1.86,
3894
- "learning_rate": 4.068038912251837e-05,
3895
- "loss": 2.3122,
3896
- "step": 648000
3897
- },
3898
- {
3899
- "epoch": 1.87,
3900
- "learning_rate": 4.06660070069667e-05,
3901
- "loss": 2.3103,
3902
- "step": 649000
3903
- },
3904
- {
3905
- "epoch": 1.87,
3906
- "learning_rate": 4.065162489141503e-05,
3907
- "loss": 2.3286,
3908
- "step": 650000
3909
- },
3910
- {
3911
- "epoch": 1.87,
3912
- "learning_rate": 4.063724277586336e-05,
3913
- "loss": 2.3157,
3914
- "step": 651000
3915
- },
3916
- {
3917
- "epoch": 1.88,
3918
- "learning_rate": 4.062286066031169e-05,
3919
- "loss": 2.3223,
3920
- "step": 652000
3921
- },
3922
- {
3923
- "epoch": 1.88,
3924
- "learning_rate": 4.060847854476002e-05,
3925
- "loss": 2.3045,
3926
- "step": 653000
3927
- },
3928
- {
3929
- "epoch": 1.88,
3930
- "learning_rate": 4.059409642920835e-05,
3931
- "loss": 2.3327,
3932
- "step": 654000
3933
- },
3934
- {
3935
- "epoch": 1.88,
3936
- "learning_rate": 4.0579714313656685e-05,
3937
- "loss": 2.2987,
3938
- "step": 655000
3939
- },
3940
- {
3941
- "epoch": 1.89,
3942
- "learning_rate": 4.056533219810502e-05,
3943
- "loss": 2.3388,
3944
- "step": 656000
3945
- },
3946
- {
3947
- "epoch": 1.89,
3948
- "learning_rate": 4.0550950082553344e-05,
3949
- "loss": 2.3178,
3950
- "step": 657000
3951
- },
3952
- {
3953
- "epoch": 1.89,
3954
- "learning_rate": 4.053656796700168e-05,
3955
- "loss": 2.3131,
3956
- "step": 658000
3957
- },
3958
- {
3959
- "epoch": 1.9,
3960
- "learning_rate": 4.052218585145001e-05,
3961
- "loss": 2.3014,
3962
- "step": 659000
3963
- },
3964
- {
3965
- "epoch": 1.9,
3966
- "learning_rate": 4.0507803735898335e-05,
3967
- "loss": 2.316,
3968
- "step": 660000
3969
- },
3970
- {
3971
- "epoch": 1.9,
3972
- "learning_rate": 4.049342162034667e-05,
3973
- "loss": 2.2894,
3974
- "step": 661000
3975
- },
3976
- {
3977
- "epoch": 1.9,
3978
- "learning_rate": 4.0479039504794994e-05,
3979
- "loss": 2.3215,
3980
- "step": 662000
3981
- },
3982
- {
3983
- "epoch": 1.91,
3984
- "learning_rate": 4.046465738924333e-05,
3985
- "loss": 2.3087,
3986
- "step": 663000
3987
- },
3988
- {
3989
- "epoch": 1.91,
3990
- "learning_rate": 4.045027527369166e-05,
3991
- "loss": 2.3048,
3992
- "step": 664000
3993
- },
3994
- {
3995
- "epoch": 1.91,
3996
- "learning_rate": 4.043589315813999e-05,
3997
- "loss": 2.2964,
3998
- "step": 665000
3999
- },
4000
- {
4001
- "epoch": 1.92,
4002
- "learning_rate": 4.0421511042588325e-05,
4003
- "loss": 2.3304,
4004
- "step": 666000
4005
- },
4006
- {
4007
- "epoch": 1.92,
4008
- "learning_rate": 4.040712892703665e-05,
4009
- "loss": 2.3152,
4010
- "step": 667000
4011
- },
4012
- {
4013
- "epoch": 1.92,
4014
- "learning_rate": 4.039274681148498e-05,
4015
- "loss": 2.3145,
4016
- "step": 668000
4017
- },
4018
- {
4019
- "epoch": 1.92,
4020
- "learning_rate": 4.0378364695933316e-05,
4021
- "loss": 2.3036,
4022
- "step": 669000
4023
- },
4024
- {
4025
- "epoch": 1.93,
4026
- "learning_rate": 4.036398258038165e-05,
4027
- "loss": 2.2996,
4028
- "step": 670000
4029
- },
4030
- {
4031
- "epoch": 1.93,
4032
- "learning_rate": 4.034960046482998e-05,
4033
- "loss": 2.313,
4034
- "step": 671000
4035
- },
4036
- {
4037
- "epoch": 1.93,
4038
- "learning_rate": 4.033521834927831e-05,
4039
- "loss": 2.3284,
4040
- "step": 672000
4041
- },
4042
- {
4043
- "epoch": 1.94,
4044
- "learning_rate": 4.032083623372664e-05,
4045
- "loss": 2.312,
4046
- "step": 673000
4047
- },
4048
- {
4049
- "epoch": 1.94,
4050
- "learning_rate": 4.0306454118174966e-05,
4051
- "loss": 2.3051,
4052
- "step": 674000
4053
- },
4054
- {
4055
- "epoch": 1.94,
4056
- "learning_rate": 4.02920720026233e-05,
4057
- "loss": 2.3197,
4058
- "step": 675000
4059
- },
4060
- {
4061
- "epoch": 1.94,
4062
- "learning_rate": 4.027768988707163e-05,
4063
- "loss": 2.2941,
4064
- "step": 676000
4065
- },
4066
- {
4067
- "epoch": 1.95,
4068
- "learning_rate": 4.026330777151996e-05,
4069
- "loss": 2.3032,
4070
- "step": 677000
4071
- },
4072
- {
4073
- "epoch": 1.95,
4074
- "learning_rate": 4.0248925655968296e-05,
4075
- "loss": 2.303,
4076
- "step": 678000
4077
- },
4078
- {
4079
- "epoch": 1.95,
4080
- "learning_rate": 4.023454354041662e-05,
4081
- "loss": 2.3021,
4082
- "step": 679000
4083
- },
4084
- {
4085
- "epoch": 1.96,
4086
- "learning_rate": 4.0220161424864955e-05,
4087
- "loss": 2.304,
4088
- "step": 680000
4089
- },
4090
- {
4091
- "epoch": 1.96,
4092
- "learning_rate": 4.020577930931329e-05,
4093
- "loss": 2.3208,
4094
- "step": 681000
4095
- },
4096
- {
4097
- "epoch": 1.96,
4098
- "learning_rate": 4.0191397193761614e-05,
4099
- "loss": 2.2963,
4100
- "step": 682000
4101
- },
4102
- {
4103
- "epoch": 1.96,
4104
- "learning_rate": 4.0177015078209946e-05,
4105
- "loss": 2.297,
4106
- "step": 683000
4107
- },
4108
- {
4109
- "epoch": 1.97,
4110
- "learning_rate": 4.016263296265827e-05,
4111
- "loss": 2.3003,
4112
- "step": 684000
4113
- },
4114
- {
4115
- "epoch": 1.97,
4116
- "learning_rate": 4.0148250847106605e-05,
4117
- "loss": 2.3077,
4118
- "step": 685000
4119
- },
4120
- {
4121
- "epoch": 1.97,
4122
- "learning_rate": 4.0133868731554944e-05,
4123
- "loss": 2.2932,
4124
- "step": 686000
4125
- },
4126
- {
4127
- "epoch": 1.98,
4128
- "learning_rate": 4.011948661600327e-05,
4129
- "loss": 2.3157,
4130
- "step": 687000
4131
- },
4132
- {
4133
- "epoch": 1.98,
4134
- "learning_rate": 4.01051045004516e-05,
4135
- "loss": 2.2978,
4136
- "step": 688000
4137
- },
4138
- {
4139
- "epoch": 1.98,
4140
- "learning_rate": 4.009072238489993e-05,
4141
- "loss": 2.2972,
4142
- "step": 689000
4143
- },
4144
- {
4145
- "epoch": 1.98,
4146
- "learning_rate": 4.007634026934826e-05,
4147
- "loss": 2.2958,
4148
- "step": 690000
4149
- },
4150
- {
4151
- "epoch": 1.99,
4152
- "learning_rate": 4.0061958153796594e-05,
4153
- "loss": 2.2865,
4154
- "step": 691000
4155
- },
4156
- {
4157
- "epoch": 1.99,
4158
- "learning_rate": 4.004757603824492e-05,
4159
- "loss": 2.2947,
4160
- "step": 692000
4161
- },
4162
- {
4163
- "epoch": 1.99,
4164
- "learning_rate": 4.003319392269325e-05,
4165
- "loss": 2.2913,
4166
- "step": 693000
4167
- },
4168
- {
4169
- "epoch": 2.0,
4170
- "learning_rate": 4.0018811807141586e-05,
4171
- "loss": 2.2922,
4172
- "step": 694000
4173
- },
4174
- {
4175
- "epoch": 2.0,
4176
- "learning_rate": 4.000442969158992e-05,
4177
- "loss": 2.2876,
4178
- "step": 695000
4179
- },
4180
- {
4181
- "epoch": 2.0,
4182
- "learning_rate": 3.999004757603825e-05,
4183
- "loss": 2.285,
4184
- "step": 696000
4185
- },
4186
- {
4187
- "epoch": 2.0,
4188
- "learning_rate": 3.997566546048658e-05,
4189
- "loss": 2.2848,
4190
- "step": 697000
4191
- },
4192
- {
4193
- "epoch": 2.01,
4194
- "learning_rate": 3.996128334493491e-05,
4195
- "loss": 2.2752,
4196
- "step": 698000
4197
- },
4198
- {
4199
- "epoch": 2.01,
4200
- "learning_rate": 3.9946901229383235e-05,
4201
- "loss": 2.2816,
4202
- "step": 699000
4203
- },
4204
- {
4205
- "epoch": 2.01,
4206
- "learning_rate": 3.993251911383157e-05,
4207
- "loss": 2.2844,
4208
- "step": 700000
4209
- },
4210
- {
4211
- "epoch": 2.02,
4212
- "learning_rate": 3.99181369982799e-05,
4213
- "loss": 2.3089,
4214
- "step": 701000
4215
- },
4216
- {
4217
- "epoch": 2.02,
4218
- "learning_rate": 3.9903754882728233e-05,
4219
- "loss": 2.2969,
4220
- "step": 702000
4221
- },
4222
- {
4223
- "epoch": 2.02,
4224
- "learning_rate": 3.9889372767176566e-05,
4225
- "loss": 2.2892,
4226
- "step": 703000
4227
- },
4228
- {
4229
- "epoch": 2.03,
4230
- "learning_rate": 3.987499065162489e-05,
4231
- "loss": 2.2777,
4232
- "step": 704000
4233
- },
4234
- {
4235
- "epoch": 2.03,
4236
- "learning_rate": 3.9860608536073225e-05,
4237
- "loss": 2.2933,
4238
- "step": 705000
4239
- },
4240
- {
4241
- "epoch": 2.03,
4242
- "learning_rate": 3.984622642052156e-05,
4243
- "loss": 2.2747,
4244
- "step": 706000
4245
- },
4246
- {
4247
- "epoch": 2.03,
4248
- "learning_rate": 3.983184430496988e-05,
4249
- "loss": 2.2733,
4250
- "step": 707000
4251
- },
4252
- {
4253
- "epoch": 2.04,
4254
- "learning_rate": 3.9817462189418216e-05,
4255
- "loss": 2.2944,
4256
- "step": 708000
4257
- },
4258
- {
4259
- "epoch": 2.04,
4260
- "learning_rate": 3.980308007386655e-05,
4261
- "loss": 2.2877,
4262
- "step": 709000
4263
- },
4264
- {
4265
- "epoch": 2.04,
4266
- "learning_rate": 3.978869795831488e-05,
4267
- "loss": 2.2794,
4268
- "step": 710000
4269
- },
4270
- {
4271
- "epoch": 2.05,
4272
- "learning_rate": 3.977431584276321e-05,
4273
- "loss": 2.2651,
4274
- "step": 711000
4275
- },
4276
- {
4277
- "epoch": 2.05,
4278
- "learning_rate": 3.975993372721154e-05,
4279
- "loss": 2.302,
4280
- "step": 712000
4281
- },
4282
- {
4283
- "epoch": 2.05,
4284
- "learning_rate": 3.974555161165987e-05,
4285
- "loss": 2.2904,
4286
- "step": 713000
4287
- },
4288
- {
4289
- "epoch": 2.05,
4290
- "learning_rate": 3.97311694961082e-05,
4291
- "loss": 2.2719,
4292
- "step": 714000
4293
- },
4294
- {
4295
- "epoch": 2.06,
4296
- "learning_rate": 3.971678738055653e-05,
4297
- "loss": 2.3142,
4298
- "step": 715000
4299
- },
4300
- {
4301
- "epoch": 2.06,
4302
- "learning_rate": 3.970240526500486e-05,
4303
- "loss": 2.2928,
4304
- "step": 716000
4305
- },
4306
- {
4307
- "epoch": 2.06,
4308
- "learning_rate": 3.96880231494532e-05,
4309
- "loss": 2.2849,
4310
- "step": 717000
4311
- },
4312
- {
4313
- "epoch": 2.07,
4314
- "learning_rate": 3.967364103390153e-05,
4315
- "loss": 2.2709,
4316
- "step": 718000
4317
- },
4318
- {
4319
- "epoch": 2.07,
4320
- "learning_rate": 3.9659258918349855e-05,
4321
- "loss": 2.2729,
4322
- "step": 719000
4323
- },
4324
- {
4325
- "epoch": 2.07,
4326
- "learning_rate": 3.964487680279819e-05,
4327
- "loss": 2.2696,
4328
- "step": 720000
4329
- },
4330
- {
4331
- "epoch": 2.07,
4332
- "learning_rate": 3.9630494687246514e-05,
4333
- "loss": 2.2776,
4334
- "step": 721000
4335
- },
4336
- {
4337
- "epoch": 2.08,
4338
- "learning_rate": 3.9616112571694847e-05,
4339
- "loss": 2.2878,
4340
- "step": 722000
4341
- },
4342
- {
4343
- "epoch": 2.08,
4344
- "learning_rate": 3.960173045614318e-05,
4345
- "loss": 2.2965,
4346
- "step": 723000
4347
- },
4348
- {
4349
- "epoch": 2.08,
4350
- "learning_rate": 3.9587348340591505e-05,
4351
- "loss": 2.2688,
4352
- "step": 724000
4353
- },
4354
- {
4355
- "epoch": 2.09,
4356
- "learning_rate": 3.9572966225039845e-05,
4357
- "loss": 2.279,
4358
- "step": 725000
4359
- },
4360
- {
4361
- "epoch": 2.09,
4362
- "learning_rate": 3.955858410948817e-05,
4363
- "loss": 2.2667,
4364
- "step": 726000
4365
- },
4366
- {
4367
- "epoch": 2.09,
4368
- "learning_rate": 3.95442019939365e-05,
4369
- "loss": 2.2656,
4370
- "step": 727000
4371
- },
4372
- {
4373
- "epoch": 2.09,
4374
- "learning_rate": 3.9529819878384836e-05,
4375
- "loss": 2.2948,
4376
- "step": 728000
4377
- },
4378
- {
4379
- "epoch": 2.1,
4380
- "learning_rate": 3.951543776283316e-05,
4381
- "loss": 2.2762,
4382
- "step": 729000
4383
- },
4384
- {
4385
- "epoch": 2.1,
4386
- "learning_rate": 3.9501055647281494e-05,
4387
- "loss": 2.2698,
4388
- "step": 730000
4389
- },
4390
- {
4391
- "epoch": 2.1,
4392
- "learning_rate": 3.948667353172982e-05,
4393
- "loss": 2.2756,
4394
- "step": 731000
4395
- },
4396
- {
4397
- "epoch": 2.11,
4398
- "learning_rate": 3.947229141617816e-05,
4399
- "loss": 2.2786,
4400
- "step": 732000
4401
- },
4402
- {
4403
- "epoch": 2.11,
4404
- "learning_rate": 3.9457909300626486e-05,
4405
- "loss": 2.2833,
4406
- "step": 733000
4407
- },
4408
- {
4409
- "epoch": 2.11,
4410
- "learning_rate": 3.944352718507482e-05,
4411
- "loss": 2.3012,
4412
- "step": 734000
4413
- },
4414
- {
4415
- "epoch": 2.11,
4416
- "learning_rate": 3.942914506952315e-05,
4417
- "loss": 2.2697,
4418
- "step": 735000
4419
- },
4420
- {
4421
- "epoch": 2.12,
4422
- "learning_rate": 3.941476295397148e-05,
4423
- "loss": 2.2465,
4424
- "step": 736000
4425
- },
4426
- {
4427
- "epoch": 2.12,
4428
- "learning_rate": 3.940038083841981e-05,
4429
- "loss": 2.2856,
4430
- "step": 737000
4431
- },
4432
- {
4433
- "epoch": 2.12,
4434
- "learning_rate": 3.938599872286814e-05,
4435
- "loss": 2.2749,
4436
- "step": 738000
4437
- },
4438
- {
4439
- "epoch": 2.13,
4440
- "learning_rate": 3.937161660731647e-05,
4441
- "loss": 2.2822,
4442
- "step": 739000
4443
- },
4444
- {
4445
- "epoch": 2.13,
4446
- "learning_rate": 3.935723449176481e-05,
4447
- "loss": 2.254,
4448
- "step": 740000
4449
- }
4450
- ],
4451
- "max_steps": 3476540,
4452
- "num_train_epochs": 10,
4453
- "total_flos": 5.757618140421688e+17,
4454
- "trial_name": null,
4455
- "trial_params": null
4456
- }