seamoon2333 commited on
Commit
3fb9f3e
·
1 Parent(s): b48ecad

Delete trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +0 -3028
trainer_state.json DELETED
@@ -1,3028 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 0.019230029614245605,
5
- "eval_steps": 500,
6
- "global_step": 500,
7
- "is_hyper_param_search": false,
8
- "is_local_process_zero": true,
9
- "is_world_process_zero": true,
10
- "log_history": [
11
- {
12
- "epoch": 0.0,
13
- "learning_rate": 9.98e-05,
14
- "loss": 1.0994,
15
- "step": 1
16
- },
17
- {
18
- "epoch": 0.0,
19
- "learning_rate": 9.960000000000001e-05,
20
- "loss": 0.0,
21
- "step": 2
22
- },
23
- {
24
- "epoch": 0.0,
25
- "learning_rate": 9.94e-05,
26
- "loss": 0.0,
27
- "step": 3
28
- },
29
- {
30
- "epoch": 0.0,
31
- "learning_rate": 9.92e-05,
32
- "loss": 0.0,
33
- "step": 4
34
- },
35
- {
36
- "epoch": 0.0,
37
- "learning_rate": 9.900000000000001e-05,
38
- "loss": 0.0,
39
- "step": 5
40
- },
41
- {
42
- "epoch": 0.0,
43
- "learning_rate": 9.88e-05,
44
- "loss": 0.0,
45
- "step": 6
46
- },
47
- {
48
- "epoch": 0.0,
49
- "learning_rate": 9.86e-05,
50
- "loss": 0.0,
51
- "step": 7
52
- },
53
- {
54
- "epoch": 0.0,
55
- "learning_rate": 9.84e-05,
56
- "loss": 0.0,
57
- "step": 8
58
- },
59
- {
60
- "epoch": 0.0,
61
- "learning_rate": 9.82e-05,
62
- "loss": 0.0,
63
- "step": 9
64
- },
65
- {
66
- "epoch": 0.0,
67
- "learning_rate": 9.8e-05,
68
- "loss": 0.0,
69
- "step": 10
70
- },
71
- {
72
- "epoch": 0.0,
73
- "learning_rate": 9.78e-05,
74
- "loss": 0.0,
75
- "step": 11
76
- },
77
- {
78
- "epoch": 0.0,
79
- "learning_rate": 9.76e-05,
80
- "loss": 0.0,
81
- "step": 12
82
- },
83
- {
84
- "epoch": 0.0,
85
- "learning_rate": 9.74e-05,
86
- "loss": 0.0,
87
- "step": 13
88
- },
89
- {
90
- "epoch": 0.0,
91
- "learning_rate": 9.72e-05,
92
- "loss": 0.0,
93
- "step": 14
94
- },
95
- {
96
- "epoch": 0.0,
97
- "learning_rate": 9.7e-05,
98
- "loss": 0.0,
99
- "step": 15
100
- },
101
- {
102
- "epoch": 0.0,
103
- "learning_rate": 9.680000000000001e-05,
104
- "loss": 0.0,
105
- "step": 16
106
- },
107
- {
108
- "epoch": 0.0,
109
- "learning_rate": 9.66e-05,
110
- "loss": 0.0,
111
- "step": 17
112
- },
113
- {
114
- "epoch": 0.0,
115
- "learning_rate": 9.64e-05,
116
- "loss": 0.0,
117
- "step": 18
118
- },
119
- {
120
- "epoch": 0.0,
121
- "learning_rate": 9.620000000000001e-05,
122
- "loss": 0.0,
123
- "step": 19
124
- },
125
- {
126
- "epoch": 0.0,
127
- "learning_rate": 9.6e-05,
128
- "loss": 0.0,
129
- "step": 20
130
- },
131
- {
132
- "epoch": 0.0,
133
- "learning_rate": 9.58e-05,
134
- "loss": 0.0,
135
- "step": 21
136
- },
137
- {
138
- "epoch": 0.0,
139
- "learning_rate": 9.56e-05,
140
- "loss": 0.0,
141
- "step": 22
142
- },
143
- {
144
- "epoch": 0.0,
145
- "learning_rate": 9.54e-05,
146
- "loss": 0.0,
147
- "step": 23
148
- },
149
- {
150
- "epoch": 0.0,
151
- "learning_rate": 9.52e-05,
152
- "loss": 0.0,
153
- "step": 24
154
- },
155
- {
156
- "epoch": 0.0,
157
- "learning_rate": 9.5e-05,
158
- "loss": 0.0,
159
- "step": 25
160
- },
161
- {
162
- "epoch": 0.0,
163
- "learning_rate": 9.48e-05,
164
- "loss": 0.0,
165
- "step": 26
166
- },
167
- {
168
- "epoch": 0.0,
169
- "learning_rate": 9.46e-05,
170
- "loss": 0.0,
171
- "step": 27
172
- },
173
- {
174
- "epoch": 0.0,
175
- "learning_rate": 9.44e-05,
176
- "loss": 0.0,
177
- "step": 28
178
- },
179
- {
180
- "epoch": 0.0,
181
- "learning_rate": 9.42e-05,
182
- "loss": 0.0,
183
- "step": 29
184
- },
185
- {
186
- "epoch": 0.0,
187
- "learning_rate": 9.4e-05,
188
- "loss": 0.0,
189
- "step": 30
190
- },
191
- {
192
- "epoch": 0.0,
193
- "learning_rate": 9.38e-05,
194
- "loss": 0.0,
195
- "step": 31
196
- },
197
- {
198
- "epoch": 0.0,
199
- "learning_rate": 9.360000000000001e-05,
200
- "loss": 0.0,
201
- "step": 32
202
- },
203
- {
204
- "epoch": 0.0,
205
- "learning_rate": 9.340000000000001e-05,
206
- "loss": 0.0,
207
- "step": 33
208
- },
209
- {
210
- "epoch": 0.0,
211
- "learning_rate": 9.320000000000002e-05,
212
- "loss": 0.0,
213
- "step": 34
214
- },
215
- {
216
- "epoch": 0.0,
217
- "learning_rate": 9.300000000000001e-05,
218
- "loss": 0.0,
219
- "step": 35
220
- },
221
- {
222
- "epoch": 0.0,
223
- "learning_rate": 9.28e-05,
224
- "loss": 0.0,
225
- "step": 36
226
- },
227
- {
228
- "epoch": 0.0,
229
- "learning_rate": 9.260000000000001e-05,
230
- "loss": 0.0,
231
- "step": 37
232
- },
233
- {
234
- "epoch": 0.0,
235
- "learning_rate": 9.240000000000001e-05,
236
- "loss": 0.0,
237
- "step": 38
238
- },
239
- {
240
- "epoch": 0.0,
241
- "learning_rate": 9.22e-05,
242
- "loss": 0.0,
243
- "step": 39
244
- },
245
- {
246
- "epoch": 0.0,
247
- "learning_rate": 9.200000000000001e-05,
248
- "loss": 0.0,
249
- "step": 40
250
- },
251
- {
252
- "epoch": 0.0,
253
- "learning_rate": 9.180000000000001e-05,
254
- "loss": 0.0,
255
- "step": 41
256
- },
257
- {
258
- "epoch": 0.0,
259
- "learning_rate": 9.16e-05,
260
- "loss": 0.0,
261
- "step": 42
262
- },
263
- {
264
- "epoch": 0.0,
265
- "learning_rate": 9.140000000000001e-05,
266
- "loss": 0.0,
267
- "step": 43
268
- },
269
- {
270
- "epoch": 0.0,
271
- "learning_rate": 9.120000000000001e-05,
272
- "loss": 0.0,
273
- "step": 44
274
- },
275
- {
276
- "epoch": 0.0,
277
- "learning_rate": 9.1e-05,
278
- "loss": 0.0,
279
- "step": 45
280
- },
281
- {
282
- "epoch": 0.0,
283
- "learning_rate": 9.080000000000001e-05,
284
- "loss": 0.0,
285
- "step": 46
286
- },
287
- {
288
- "epoch": 0.0,
289
- "learning_rate": 9.06e-05,
290
- "loss": 0.0,
291
- "step": 47
292
- },
293
- {
294
- "epoch": 0.0,
295
- "learning_rate": 9.04e-05,
296
- "loss": 0.0,
297
- "step": 48
298
- },
299
- {
300
- "epoch": 0.0,
301
- "learning_rate": 9.020000000000001e-05,
302
- "loss": 0.0,
303
- "step": 49
304
- },
305
- {
306
- "epoch": 0.0,
307
- "learning_rate": 9e-05,
308
- "loss": 0.0,
309
- "step": 50
310
- },
311
- {
312
- "epoch": 0.0,
313
- "learning_rate": 8.98e-05,
314
- "loss": 0.0,
315
- "step": 51
316
- },
317
- {
318
- "epoch": 0.0,
319
- "learning_rate": 8.960000000000001e-05,
320
- "loss": 0.0,
321
- "step": 52
322
- },
323
- {
324
- "epoch": 0.0,
325
- "learning_rate": 8.94e-05,
326
- "loss": 0.0,
327
- "step": 53
328
- },
329
- {
330
- "epoch": 0.0,
331
- "learning_rate": 8.92e-05,
332
- "loss": 0.0,
333
- "step": 54
334
- },
335
- {
336
- "epoch": 0.0,
337
- "learning_rate": 8.900000000000001e-05,
338
- "loss": 0.0,
339
- "step": 55
340
- },
341
- {
342
- "epoch": 0.0,
343
- "learning_rate": 8.88e-05,
344
- "loss": 0.0,
345
- "step": 56
346
- },
347
- {
348
- "epoch": 0.0,
349
- "learning_rate": 8.86e-05,
350
- "loss": 0.0,
351
- "step": 57
352
- },
353
- {
354
- "epoch": 0.0,
355
- "learning_rate": 8.840000000000001e-05,
356
- "loss": 0.0,
357
- "step": 58
358
- },
359
- {
360
- "epoch": 0.0,
361
- "learning_rate": 8.82e-05,
362
- "loss": 0.0,
363
- "step": 59
364
- },
365
- {
366
- "epoch": 0.0,
367
- "learning_rate": 8.800000000000001e-05,
368
- "loss": 0.0,
369
- "step": 60
370
- },
371
- {
372
- "epoch": 0.0,
373
- "learning_rate": 8.78e-05,
374
- "loss": 0.0,
375
- "step": 61
376
- },
377
- {
378
- "epoch": 0.0,
379
- "learning_rate": 8.76e-05,
380
- "loss": 0.0,
381
- "step": 62
382
- },
383
- {
384
- "epoch": 0.0,
385
- "learning_rate": 8.740000000000001e-05,
386
- "loss": 0.0,
387
- "step": 63
388
- },
389
- {
390
- "epoch": 0.0,
391
- "learning_rate": 8.72e-05,
392
- "loss": 0.0,
393
- "step": 64
394
- },
395
- {
396
- "epoch": 0.0,
397
- "learning_rate": 8.7e-05,
398
- "loss": 0.0,
399
- "step": 65
400
- },
401
- {
402
- "epoch": 0.0,
403
- "learning_rate": 8.680000000000001e-05,
404
- "loss": 0.0,
405
- "step": 66
406
- },
407
- {
408
- "epoch": 0.0,
409
- "learning_rate": 8.66e-05,
410
- "loss": 0.0,
411
- "step": 67
412
- },
413
- {
414
- "epoch": 0.0,
415
- "learning_rate": 8.64e-05,
416
- "loss": 0.0,
417
- "step": 68
418
- },
419
- {
420
- "epoch": 0.0,
421
- "learning_rate": 8.620000000000001e-05,
422
- "loss": 0.0,
423
- "step": 69
424
- },
425
- {
426
- "epoch": 0.0,
427
- "learning_rate": 8.6e-05,
428
- "loss": 0.0,
429
- "step": 70
430
- },
431
- {
432
- "epoch": 0.0,
433
- "learning_rate": 8.58e-05,
434
- "loss": 0.0,
435
- "step": 71
436
- },
437
- {
438
- "epoch": 0.0,
439
- "learning_rate": 8.560000000000001e-05,
440
- "loss": 0.0,
441
- "step": 72
442
- },
443
- {
444
- "epoch": 0.0,
445
- "learning_rate": 8.54e-05,
446
- "loss": 0.0,
447
- "step": 73
448
- },
449
- {
450
- "epoch": 0.0,
451
- "learning_rate": 8.52e-05,
452
- "loss": 0.0,
453
- "step": 74
454
- },
455
- {
456
- "epoch": 0.0,
457
- "learning_rate": 8.5e-05,
458
- "loss": 0.0,
459
- "step": 75
460
- },
461
- {
462
- "epoch": 0.0,
463
- "learning_rate": 8.48e-05,
464
- "loss": 0.0,
465
- "step": 76
466
- },
467
- {
468
- "epoch": 0.0,
469
- "learning_rate": 8.46e-05,
470
- "loss": 0.0,
471
- "step": 77
472
- },
473
- {
474
- "epoch": 0.0,
475
- "learning_rate": 8.44e-05,
476
- "loss": 0.0,
477
- "step": 78
478
- },
479
- {
480
- "epoch": 0.0,
481
- "learning_rate": 8.42e-05,
482
- "loss": 0.0,
483
- "step": 79
484
- },
485
- {
486
- "epoch": 0.0,
487
- "learning_rate": 8.4e-05,
488
- "loss": 0.0,
489
- "step": 80
490
- },
491
- {
492
- "epoch": 0.0,
493
- "learning_rate": 8.38e-05,
494
- "loss": 0.0,
495
- "step": 81
496
- },
497
- {
498
- "epoch": 0.0,
499
- "learning_rate": 8.36e-05,
500
- "loss": 0.0,
501
- "step": 82
502
- },
503
- {
504
- "epoch": 0.0,
505
- "learning_rate": 8.34e-05,
506
- "loss": 0.0,
507
- "step": 83
508
- },
509
- {
510
- "epoch": 0.0,
511
- "learning_rate": 8.32e-05,
512
- "loss": 0.0,
513
- "step": 84
514
- },
515
- {
516
- "epoch": 0.0,
517
- "learning_rate": 8.3e-05,
518
- "loss": 0.0,
519
- "step": 85
520
- },
521
- {
522
- "epoch": 0.0,
523
- "learning_rate": 8.28e-05,
524
- "loss": 0.0,
525
- "step": 86
526
- },
527
- {
528
- "epoch": 0.0,
529
- "learning_rate": 8.26e-05,
530
- "loss": 0.0,
531
- "step": 87
532
- },
533
- {
534
- "epoch": 0.0,
535
- "learning_rate": 8.24e-05,
536
- "loss": 0.0,
537
- "step": 88
538
- },
539
- {
540
- "epoch": 0.0,
541
- "learning_rate": 8.22e-05,
542
- "loss": 0.0,
543
- "step": 89
544
- },
545
- {
546
- "epoch": 0.0,
547
- "learning_rate": 8.2e-05,
548
- "loss": 0.0,
549
- "step": 90
550
- },
551
- {
552
- "epoch": 0.0,
553
- "learning_rate": 8.18e-05,
554
- "loss": 0.0,
555
- "step": 91
556
- },
557
- {
558
- "epoch": 0.0,
559
- "learning_rate": 8.16e-05,
560
- "loss": 0.0,
561
- "step": 92
562
- },
563
- {
564
- "epoch": 0.0,
565
- "learning_rate": 8.14e-05,
566
- "loss": 0.0,
567
- "step": 93
568
- },
569
- {
570
- "epoch": 0.0,
571
- "learning_rate": 8.120000000000001e-05,
572
- "loss": 0.0,
573
- "step": 94
574
- },
575
- {
576
- "epoch": 0.0,
577
- "learning_rate": 8.1e-05,
578
- "loss": 0.0,
579
- "step": 95
580
- },
581
- {
582
- "epoch": 0.0,
583
- "learning_rate": 8.080000000000001e-05,
584
- "loss": 0.0,
585
- "step": 96
586
- },
587
- {
588
- "epoch": 0.0,
589
- "learning_rate": 8.060000000000001e-05,
590
- "loss": 0.0,
591
- "step": 97
592
- },
593
- {
594
- "epoch": 0.0,
595
- "learning_rate": 8.04e-05,
596
- "loss": 0.0,
597
- "step": 98
598
- },
599
- {
600
- "epoch": 0.0,
601
- "learning_rate": 8.020000000000001e-05,
602
- "loss": 0.0,
603
- "step": 99
604
- },
605
- {
606
- "epoch": 0.0,
607
- "learning_rate": 8e-05,
608
- "loss": 0.0,
609
- "step": 100
610
- },
611
- {
612
- "epoch": 0.0,
613
- "learning_rate": 7.98e-05,
614
- "loss": 0.0,
615
- "step": 101
616
- },
617
- {
618
- "epoch": 0.0,
619
- "learning_rate": 7.960000000000001e-05,
620
- "loss": 0.0,
621
- "step": 102
622
- },
623
- {
624
- "epoch": 0.0,
625
- "learning_rate": 7.94e-05,
626
- "loss": 0.0,
627
- "step": 103
628
- },
629
- {
630
- "epoch": 0.0,
631
- "learning_rate": 7.920000000000001e-05,
632
- "loss": 0.0,
633
- "step": 104
634
- },
635
- {
636
- "epoch": 0.0,
637
- "learning_rate": 7.900000000000001e-05,
638
- "loss": 0.0,
639
- "step": 105
640
- },
641
- {
642
- "epoch": 0.0,
643
- "learning_rate": 7.88e-05,
644
- "loss": 0.0,
645
- "step": 106
646
- },
647
- {
648
- "epoch": 0.0,
649
- "learning_rate": 7.860000000000001e-05,
650
- "loss": 0.0,
651
- "step": 107
652
- },
653
- {
654
- "epoch": 0.0,
655
- "learning_rate": 7.840000000000001e-05,
656
- "loss": 0.0,
657
- "step": 108
658
- },
659
- {
660
- "epoch": 0.0,
661
- "learning_rate": 7.82e-05,
662
- "loss": 0.0,
663
- "step": 109
664
- },
665
- {
666
- "epoch": 0.0,
667
- "learning_rate": 7.800000000000001e-05,
668
- "loss": 0.0,
669
- "step": 110
670
- },
671
- {
672
- "epoch": 0.0,
673
- "learning_rate": 7.780000000000001e-05,
674
- "loss": 0.0,
675
- "step": 111
676
- },
677
- {
678
- "epoch": 0.0,
679
- "learning_rate": 7.76e-05,
680
- "loss": 0.0,
681
- "step": 112
682
- },
683
- {
684
- "epoch": 0.0,
685
- "learning_rate": 7.740000000000001e-05,
686
- "loss": 0.0,
687
- "step": 113
688
- },
689
- {
690
- "epoch": 0.0,
691
- "learning_rate": 7.72e-05,
692
- "loss": 0.0,
693
- "step": 114
694
- },
695
- {
696
- "epoch": 0.0,
697
- "learning_rate": 7.7e-05,
698
- "loss": 0.0,
699
- "step": 115
700
- },
701
- {
702
- "epoch": 0.0,
703
- "learning_rate": 7.680000000000001e-05,
704
- "loss": 0.0,
705
- "step": 116
706
- },
707
- {
708
- "epoch": 0.0,
709
- "learning_rate": 7.66e-05,
710
- "loss": 0.0,
711
- "step": 117
712
- },
713
- {
714
- "epoch": 0.0,
715
- "learning_rate": 7.64e-05,
716
- "loss": 0.0,
717
- "step": 118
718
- },
719
- {
720
- "epoch": 0.0,
721
- "learning_rate": 7.620000000000001e-05,
722
- "loss": 0.0,
723
- "step": 119
724
- },
725
- {
726
- "epoch": 0.0,
727
- "learning_rate": 7.6e-05,
728
- "loss": 0.0,
729
- "step": 120
730
- },
731
- {
732
- "epoch": 0.0,
733
- "learning_rate": 7.58e-05,
734
- "loss": 0.0,
735
- "step": 121
736
- },
737
- {
738
- "epoch": 0.0,
739
- "learning_rate": 7.560000000000001e-05,
740
- "loss": 0.0,
741
- "step": 122
742
- },
743
- {
744
- "epoch": 0.0,
745
- "learning_rate": 7.54e-05,
746
- "loss": 0.0,
747
- "step": 123
748
- },
749
- {
750
- "epoch": 0.0,
751
- "learning_rate": 7.52e-05,
752
- "loss": 0.0,
753
- "step": 124
754
- },
755
- {
756
- "epoch": 0.0,
757
- "learning_rate": 7.500000000000001e-05,
758
- "loss": 0.0,
759
- "step": 125
760
- },
761
- {
762
- "epoch": 0.0,
763
- "learning_rate": 7.48e-05,
764
- "loss": 0.0,
765
- "step": 126
766
- },
767
- {
768
- "epoch": 0.0,
769
- "learning_rate": 7.46e-05,
770
- "loss": 0.0,
771
- "step": 127
772
- },
773
- {
774
- "epoch": 0.0,
775
- "learning_rate": 7.44e-05,
776
- "loss": 0.0,
777
- "step": 128
778
- },
779
- {
780
- "epoch": 0.0,
781
- "learning_rate": 7.42e-05,
782
- "loss": 0.0,
783
- "step": 129
784
- },
785
- {
786
- "epoch": 0.0,
787
- "learning_rate": 7.4e-05,
788
- "loss": 0.0,
789
- "step": 130
790
- },
791
- {
792
- "epoch": 0.01,
793
- "learning_rate": 7.38e-05,
794
- "loss": 0.0,
795
- "step": 131
796
- },
797
- {
798
- "epoch": 0.01,
799
- "learning_rate": 7.36e-05,
800
- "loss": 0.0,
801
- "step": 132
802
- },
803
- {
804
- "epoch": 0.01,
805
- "learning_rate": 7.340000000000001e-05,
806
- "loss": 0.0,
807
- "step": 133
808
- },
809
- {
810
- "epoch": 0.01,
811
- "learning_rate": 7.32e-05,
812
- "loss": 0.0,
813
- "step": 134
814
- },
815
- {
816
- "epoch": 0.01,
817
- "learning_rate": 7.3e-05,
818
- "loss": 0.0,
819
- "step": 135
820
- },
821
- {
822
- "epoch": 0.01,
823
- "learning_rate": 7.280000000000001e-05,
824
- "loss": 0.0,
825
- "step": 136
826
- },
827
- {
828
- "epoch": 0.01,
829
- "learning_rate": 7.26e-05,
830
- "loss": 0.0,
831
- "step": 137
832
- },
833
- {
834
- "epoch": 0.01,
835
- "learning_rate": 7.24e-05,
836
- "loss": 0.0,
837
- "step": 138
838
- },
839
- {
840
- "epoch": 0.01,
841
- "learning_rate": 7.22e-05,
842
- "loss": 0.0,
843
- "step": 139
844
- },
845
- {
846
- "epoch": 0.01,
847
- "learning_rate": 7.2e-05,
848
- "loss": 0.0,
849
- "step": 140
850
- },
851
- {
852
- "epoch": 0.01,
853
- "learning_rate": 7.18e-05,
854
- "loss": 0.0,
855
- "step": 141
856
- },
857
- {
858
- "epoch": 0.01,
859
- "learning_rate": 7.16e-05,
860
- "loss": 0.0,
861
- "step": 142
862
- },
863
- {
864
- "epoch": 0.01,
865
- "learning_rate": 7.14e-05,
866
- "loss": 0.0,
867
- "step": 143
868
- },
869
- {
870
- "epoch": 0.01,
871
- "learning_rate": 7.12e-05,
872
- "loss": 0.0,
873
- "step": 144
874
- },
875
- {
876
- "epoch": 0.01,
877
- "learning_rate": 7.1e-05,
878
- "loss": 0.0,
879
- "step": 145
880
- },
881
- {
882
- "epoch": 0.01,
883
- "learning_rate": 7.08e-05,
884
- "loss": 0.0,
885
- "step": 146
886
- },
887
- {
888
- "epoch": 0.01,
889
- "learning_rate": 7.06e-05,
890
- "loss": 0.0,
891
- "step": 147
892
- },
893
- {
894
- "epoch": 0.01,
895
- "learning_rate": 7.04e-05,
896
- "loss": 0.0,
897
- "step": 148
898
- },
899
- {
900
- "epoch": 0.01,
901
- "learning_rate": 7.02e-05,
902
- "loss": 0.0,
903
- "step": 149
904
- },
905
- {
906
- "epoch": 0.01,
907
- "learning_rate": 7e-05,
908
- "loss": 0.0,
909
- "step": 150
910
- },
911
- {
912
- "epoch": 0.01,
913
- "learning_rate": 6.98e-05,
914
- "loss": 0.0,
915
- "step": 151
916
- },
917
- {
918
- "epoch": 0.01,
919
- "learning_rate": 6.96e-05,
920
- "loss": 0.0,
921
- "step": 152
922
- },
923
- {
924
- "epoch": 0.01,
925
- "learning_rate": 6.939999999999999e-05,
926
- "loss": 0.0,
927
- "step": 153
928
- },
929
- {
930
- "epoch": 0.01,
931
- "learning_rate": 6.92e-05,
932
- "loss": 0.0,
933
- "step": 154
934
- },
935
- {
936
- "epoch": 0.01,
937
- "learning_rate": 6.9e-05,
938
- "loss": 0.0,
939
- "step": 155
940
- },
941
- {
942
- "epoch": 0.01,
943
- "learning_rate": 6.879999999999999e-05,
944
- "loss": 0.0,
945
- "step": 156
946
- },
947
- {
948
- "epoch": 0.01,
949
- "learning_rate": 6.860000000000001e-05,
950
- "loss": 0.0,
951
- "step": 157
952
- },
953
- {
954
- "epoch": 0.01,
955
- "learning_rate": 6.840000000000001e-05,
956
- "loss": 0.0,
957
- "step": 158
958
- },
959
- {
960
- "epoch": 0.01,
961
- "learning_rate": 6.82e-05,
962
- "loss": 0.0,
963
- "step": 159
964
- },
965
- {
966
- "epoch": 0.01,
967
- "learning_rate": 6.800000000000001e-05,
968
- "loss": 0.0,
969
- "step": 160
970
- },
971
- {
972
- "epoch": 0.01,
973
- "learning_rate": 6.780000000000001e-05,
974
- "loss": 0.0,
975
- "step": 161
976
- },
977
- {
978
- "epoch": 0.01,
979
- "learning_rate": 6.76e-05,
980
- "loss": 0.0,
981
- "step": 162
982
- },
983
- {
984
- "epoch": 0.01,
985
- "learning_rate": 6.740000000000001e-05,
986
- "loss": 0.0,
987
- "step": 163
988
- },
989
- {
990
- "epoch": 0.01,
991
- "learning_rate": 6.720000000000001e-05,
992
- "loss": 0.0,
993
- "step": 164
994
- },
995
- {
996
- "epoch": 0.01,
997
- "learning_rate": 6.7e-05,
998
- "loss": 0.0,
999
- "step": 165
1000
- },
1001
- {
1002
- "epoch": 0.01,
1003
- "learning_rate": 6.680000000000001e-05,
1004
- "loss": 0.0,
1005
- "step": 166
1006
- },
1007
- {
1008
- "epoch": 0.01,
1009
- "learning_rate": 6.66e-05,
1010
- "loss": 0.0,
1011
- "step": 167
1012
- },
1013
- {
1014
- "epoch": 0.01,
1015
- "learning_rate": 6.64e-05,
1016
- "loss": 0.0,
1017
- "step": 168
1018
- },
1019
- {
1020
- "epoch": 0.01,
1021
- "learning_rate": 6.620000000000001e-05,
1022
- "loss": 0.0,
1023
- "step": 169
1024
- },
1025
- {
1026
- "epoch": 0.01,
1027
- "learning_rate": 6.6e-05,
1028
- "loss": 0.0,
1029
- "step": 170
1030
- },
1031
- {
1032
- "epoch": 0.01,
1033
- "learning_rate": 6.58e-05,
1034
- "loss": 0.0,
1035
- "step": 171
1036
- },
1037
- {
1038
- "epoch": 0.01,
1039
- "learning_rate": 6.560000000000001e-05,
1040
- "loss": 0.0,
1041
- "step": 172
1042
- },
1043
- {
1044
- "epoch": 0.01,
1045
- "learning_rate": 6.54e-05,
1046
- "loss": 0.0,
1047
- "step": 173
1048
- },
1049
- {
1050
- "epoch": 0.01,
1051
- "learning_rate": 6.52e-05,
1052
- "loss": 0.0,
1053
- "step": 174
1054
- },
1055
- {
1056
- "epoch": 0.01,
1057
- "learning_rate": 6.500000000000001e-05,
1058
- "loss": 0.0,
1059
- "step": 175
1060
- },
1061
- {
1062
- "epoch": 0.01,
1063
- "learning_rate": 6.48e-05,
1064
- "loss": 0.0,
1065
- "step": 176
1066
- },
1067
- {
1068
- "epoch": 0.01,
1069
- "learning_rate": 6.460000000000001e-05,
1070
- "loss": 0.0,
1071
- "step": 177
1072
- },
1073
- {
1074
- "epoch": 0.01,
1075
- "learning_rate": 6.440000000000001e-05,
1076
- "loss": 0.0,
1077
- "step": 178
1078
- },
1079
- {
1080
- "epoch": 0.01,
1081
- "learning_rate": 6.42e-05,
1082
- "loss": 0.0,
1083
- "step": 179
1084
- },
1085
- {
1086
- "epoch": 0.01,
1087
- "learning_rate": 6.400000000000001e-05,
1088
- "loss": 0.0,
1089
- "step": 180
1090
- },
1091
- {
1092
- "epoch": 0.01,
1093
- "learning_rate": 6.38e-05,
1094
- "loss": 0.0,
1095
- "step": 181
1096
- },
1097
- {
1098
- "epoch": 0.01,
1099
- "learning_rate": 6.36e-05,
1100
- "loss": 0.0,
1101
- "step": 182
1102
- },
1103
- {
1104
- "epoch": 0.01,
1105
- "learning_rate": 6.340000000000001e-05,
1106
- "loss": 0.0,
1107
- "step": 183
1108
- },
1109
- {
1110
- "epoch": 0.01,
1111
- "learning_rate": 6.32e-05,
1112
- "loss": 0.0,
1113
- "step": 184
1114
- },
1115
- {
1116
- "epoch": 0.01,
1117
- "learning_rate": 6.3e-05,
1118
- "loss": 0.0,
1119
- "step": 185
1120
- },
1121
- {
1122
- "epoch": 0.01,
1123
- "learning_rate": 6.280000000000001e-05,
1124
- "loss": 0.0,
1125
- "step": 186
1126
- },
1127
- {
1128
- "epoch": 0.01,
1129
- "learning_rate": 6.26e-05,
1130
- "loss": 0.0,
1131
- "step": 187
1132
- },
1133
- {
1134
- "epoch": 0.01,
1135
- "learning_rate": 6.24e-05,
1136
- "loss": 0.0,
1137
- "step": 188
1138
- },
1139
- {
1140
- "epoch": 0.01,
1141
- "learning_rate": 6.220000000000001e-05,
1142
- "loss": 0.0,
1143
- "step": 189
1144
- },
1145
- {
1146
- "epoch": 0.01,
1147
- "learning_rate": 6.2e-05,
1148
- "loss": 0.0,
1149
- "step": 190
1150
- },
1151
- {
1152
- "epoch": 0.01,
1153
- "learning_rate": 6.18e-05,
1154
- "loss": 0.0,
1155
- "step": 191
1156
- },
1157
- {
1158
- "epoch": 0.01,
1159
- "learning_rate": 6.16e-05,
1160
- "loss": 0.0,
1161
- "step": 192
1162
- },
1163
- {
1164
- "epoch": 0.01,
1165
- "learning_rate": 6.14e-05,
1166
- "loss": 0.0,
1167
- "step": 193
1168
- },
1169
- {
1170
- "epoch": 0.01,
1171
- "learning_rate": 6.12e-05,
1172
- "loss": 0.0,
1173
- "step": 194
1174
- },
1175
- {
1176
- "epoch": 0.01,
1177
- "learning_rate": 6.1e-05,
1178
- "loss": 0.0,
1179
- "step": 195
1180
- },
1181
- {
1182
- "epoch": 0.01,
1183
- "learning_rate": 6.08e-05,
1184
- "loss": 0.0,
1185
- "step": 196
1186
- },
1187
- {
1188
- "epoch": 0.01,
1189
- "learning_rate": 6.06e-05,
1190
- "loss": 0.0,
1191
- "step": 197
1192
- },
1193
- {
1194
- "epoch": 0.01,
1195
- "learning_rate": 6.04e-05,
1196
- "loss": 0.0,
1197
- "step": 198
1198
- },
1199
- {
1200
- "epoch": 0.01,
1201
- "learning_rate": 6.02e-05,
1202
- "loss": 0.0,
1203
- "step": 199
1204
- },
1205
- {
1206
- "epoch": 0.01,
1207
- "learning_rate": 6e-05,
1208
- "loss": 0.0,
1209
- "step": 200
1210
- },
1211
- {
1212
- "epoch": 0.01,
1213
- "learning_rate": 5.9800000000000003e-05,
1214
- "loss": 0.0,
1215
- "step": 201
1216
- },
1217
- {
1218
- "epoch": 0.01,
1219
- "learning_rate": 5.96e-05,
1220
- "loss": 0.0,
1221
- "step": 202
1222
- },
1223
- {
1224
- "epoch": 0.01,
1225
- "learning_rate": 5.94e-05,
1226
- "loss": 0.0,
1227
- "step": 203
1228
- },
1229
- {
1230
- "epoch": 0.01,
1231
- "learning_rate": 5.92e-05,
1232
- "loss": 0.0,
1233
- "step": 204
1234
- },
1235
- {
1236
- "epoch": 0.01,
1237
- "learning_rate": 5.9e-05,
1238
- "loss": 0.0,
1239
- "step": 205
1240
- },
1241
- {
1242
- "epoch": 0.01,
1243
- "learning_rate": 5.88e-05,
1244
- "loss": 0.0,
1245
- "step": 206
1246
- },
1247
- {
1248
- "epoch": 0.01,
1249
- "learning_rate": 5.86e-05,
1250
- "loss": 0.0,
1251
- "step": 207
1252
- },
1253
- {
1254
- "epoch": 0.01,
1255
- "learning_rate": 5.8399999999999997e-05,
1256
- "loss": 0.0,
1257
- "step": 208
1258
- },
1259
- {
1260
- "epoch": 0.01,
1261
- "learning_rate": 5.82e-05,
1262
- "loss": 0.0,
1263
- "step": 209
1264
- },
1265
- {
1266
- "epoch": 0.01,
1267
- "learning_rate": 5.8e-05,
1268
- "loss": 0.0,
1269
- "step": 210
1270
- },
1271
- {
1272
- "epoch": 0.01,
1273
- "learning_rate": 5.7799999999999995e-05,
1274
- "loss": 0.0,
1275
- "step": 211
1276
- },
1277
- {
1278
- "epoch": 0.01,
1279
- "learning_rate": 5.76e-05,
1280
- "loss": 0.0,
1281
- "step": 212
1282
- },
1283
- {
1284
- "epoch": 0.01,
1285
- "learning_rate": 5.74e-05,
1286
- "loss": 0.0,
1287
- "step": 213
1288
- },
1289
- {
1290
- "epoch": 0.01,
1291
- "learning_rate": 5.72e-05,
1292
- "loss": 0.0,
1293
- "step": 214
1294
- },
1295
- {
1296
- "epoch": 0.01,
1297
- "learning_rate": 5.6999999999999996e-05,
1298
- "loss": 0.0,
1299
- "step": 215
1300
- },
1301
- {
1302
- "epoch": 0.01,
1303
- "learning_rate": 5.68e-05,
1304
- "loss": 0.0,
1305
- "step": 216
1306
- },
1307
- {
1308
- "epoch": 0.01,
1309
- "learning_rate": 5.66e-05,
1310
- "loss": 0.0,
1311
- "step": 217
1312
- },
1313
- {
1314
- "epoch": 0.01,
1315
- "learning_rate": 5.6399999999999995e-05,
1316
- "loss": 0.0,
1317
- "step": 218
1318
- },
1319
- {
1320
- "epoch": 0.01,
1321
- "learning_rate": 5.620000000000001e-05,
1322
- "loss": 0.0,
1323
- "step": 219
1324
- },
1325
- {
1326
- "epoch": 0.01,
1327
- "learning_rate": 5.6000000000000006e-05,
1328
- "loss": 0.0,
1329
- "step": 220
1330
- },
1331
- {
1332
- "epoch": 0.01,
1333
- "learning_rate": 5.580000000000001e-05,
1334
- "loss": 0.0,
1335
- "step": 221
1336
- },
1337
- {
1338
- "epoch": 0.01,
1339
- "learning_rate": 5.560000000000001e-05,
1340
- "loss": 0.0,
1341
- "step": 222
1342
- },
1343
- {
1344
- "epoch": 0.01,
1345
- "learning_rate": 5.5400000000000005e-05,
1346
- "loss": 0.0,
1347
- "step": 223
1348
- },
1349
- {
1350
- "epoch": 0.01,
1351
- "learning_rate": 5.520000000000001e-05,
1352
- "loss": 0.0,
1353
- "step": 224
1354
- },
1355
- {
1356
- "epoch": 0.01,
1357
- "learning_rate": 5.500000000000001e-05,
1358
- "loss": 0.0,
1359
- "step": 225
1360
- },
1361
- {
1362
- "epoch": 0.01,
1363
- "learning_rate": 5.4800000000000004e-05,
1364
- "loss": 0.0,
1365
- "step": 226
1366
- },
1367
- {
1368
- "epoch": 0.01,
1369
- "learning_rate": 5.4600000000000006e-05,
1370
- "loss": 0.0,
1371
- "step": 227
1372
- },
1373
- {
1374
- "epoch": 0.01,
1375
- "learning_rate": 5.440000000000001e-05,
1376
- "loss": 0.0,
1377
- "step": 228
1378
- },
1379
- {
1380
- "epoch": 0.01,
1381
- "learning_rate": 5.420000000000001e-05,
1382
- "loss": 0.0,
1383
- "step": 229
1384
- },
1385
- {
1386
- "epoch": 0.01,
1387
- "learning_rate": 5.4000000000000005e-05,
1388
- "loss": 0.0,
1389
- "step": 230
1390
- },
1391
- {
1392
- "epoch": 0.01,
1393
- "learning_rate": 5.380000000000001e-05,
1394
- "loss": 0.0,
1395
- "step": 231
1396
- },
1397
- {
1398
- "epoch": 0.01,
1399
- "learning_rate": 5.360000000000001e-05,
1400
- "loss": 0.0,
1401
- "step": 232
1402
- },
1403
- {
1404
- "epoch": 0.01,
1405
- "learning_rate": 5.3400000000000004e-05,
1406
- "loss": 0.0,
1407
- "step": 233
1408
- },
1409
- {
1410
- "epoch": 0.01,
1411
- "learning_rate": 5.3200000000000006e-05,
1412
- "loss": 0.0,
1413
- "step": 234
1414
- },
1415
- {
1416
- "epoch": 0.01,
1417
- "learning_rate": 5.300000000000001e-05,
1418
- "loss": 0.0,
1419
- "step": 235
1420
- },
1421
- {
1422
- "epoch": 0.01,
1423
- "learning_rate": 5.28e-05,
1424
- "loss": 0.0,
1425
- "step": 236
1426
- },
1427
- {
1428
- "epoch": 0.01,
1429
- "learning_rate": 5.2600000000000005e-05,
1430
- "loss": 0.0,
1431
- "step": 237
1432
- },
1433
- {
1434
- "epoch": 0.01,
1435
- "learning_rate": 5.2400000000000007e-05,
1436
- "loss": 0.0,
1437
- "step": 238
1438
- },
1439
- {
1440
- "epoch": 0.01,
1441
- "learning_rate": 5.22e-05,
1442
- "loss": 0.0,
1443
- "step": 239
1444
- },
1445
- {
1446
- "epoch": 0.01,
1447
- "learning_rate": 5.2000000000000004e-05,
1448
- "loss": 0.0,
1449
- "step": 240
1450
- },
1451
- {
1452
- "epoch": 0.01,
1453
- "learning_rate": 5.1800000000000005e-05,
1454
- "loss": 0.0,
1455
- "step": 241
1456
- },
1457
- {
1458
- "epoch": 0.01,
1459
- "learning_rate": 5.16e-05,
1460
- "loss": 0.0,
1461
- "step": 242
1462
- },
1463
- {
1464
- "epoch": 0.01,
1465
- "learning_rate": 5.14e-05,
1466
- "loss": 0.0,
1467
- "step": 243
1468
- },
1469
- {
1470
- "epoch": 0.01,
1471
- "learning_rate": 5.1200000000000004e-05,
1472
- "loss": 0.0,
1473
- "step": 244
1474
- },
1475
- {
1476
- "epoch": 0.01,
1477
- "learning_rate": 5.1000000000000006e-05,
1478
- "loss": 0.0,
1479
- "step": 245
1480
- },
1481
- {
1482
- "epoch": 0.01,
1483
- "learning_rate": 5.08e-05,
1484
- "loss": 0.0,
1485
- "step": 246
1486
- },
1487
- {
1488
- "epoch": 0.01,
1489
- "learning_rate": 5.0600000000000003e-05,
1490
- "loss": 0.0,
1491
- "step": 247
1492
- },
1493
- {
1494
- "epoch": 0.01,
1495
- "learning_rate": 5.0400000000000005e-05,
1496
- "loss": 0.0,
1497
- "step": 248
1498
- },
1499
- {
1500
- "epoch": 0.01,
1501
- "learning_rate": 5.02e-05,
1502
- "loss": 0.0,
1503
- "step": 249
1504
- },
1505
- {
1506
- "epoch": 0.01,
1507
- "learning_rate": 5e-05,
1508
- "loss": 0.0,
1509
- "step": 250
1510
- },
1511
- {
1512
- "epoch": 0.01,
1513
- "learning_rate": 4.9800000000000004e-05,
1514
- "loss": 0.0,
1515
- "step": 251
1516
- },
1517
- {
1518
- "epoch": 0.01,
1519
- "learning_rate": 4.96e-05,
1520
- "loss": 0.0,
1521
- "step": 252
1522
- },
1523
- {
1524
- "epoch": 0.01,
1525
- "learning_rate": 4.94e-05,
1526
- "loss": 0.0,
1527
- "step": 253
1528
- },
1529
- {
1530
- "epoch": 0.01,
1531
- "learning_rate": 4.92e-05,
1532
- "loss": 0.0,
1533
- "step": 254
1534
- },
1535
- {
1536
- "epoch": 0.01,
1537
- "learning_rate": 4.9e-05,
1538
- "loss": 0.0,
1539
- "step": 255
1540
- },
1541
- {
1542
- "epoch": 0.01,
1543
- "learning_rate": 4.88e-05,
1544
- "loss": 0.0,
1545
- "step": 256
1546
- },
1547
- {
1548
- "epoch": 0.01,
1549
- "learning_rate": 4.86e-05,
1550
- "loss": 0.0,
1551
- "step": 257
1552
- },
1553
- {
1554
- "epoch": 0.01,
1555
- "learning_rate": 4.8400000000000004e-05,
1556
- "loss": 0.0,
1557
- "step": 258
1558
- },
1559
- {
1560
- "epoch": 0.01,
1561
- "learning_rate": 4.82e-05,
1562
- "loss": 0.0,
1563
- "step": 259
1564
- },
1565
- {
1566
- "epoch": 0.01,
1567
- "learning_rate": 4.8e-05,
1568
- "loss": 0.0,
1569
- "step": 260
1570
- },
1571
- {
1572
- "epoch": 0.01,
1573
- "learning_rate": 4.78e-05,
1574
- "loss": 0.0,
1575
- "step": 261
1576
- },
1577
- {
1578
- "epoch": 0.01,
1579
- "learning_rate": 4.76e-05,
1580
- "loss": 0.0,
1581
- "step": 262
1582
- },
1583
- {
1584
- "epoch": 0.01,
1585
- "learning_rate": 4.74e-05,
1586
- "loss": 0.0,
1587
- "step": 263
1588
- },
1589
- {
1590
- "epoch": 0.01,
1591
- "learning_rate": 4.72e-05,
1592
- "loss": 0.0,
1593
- "step": 264
1594
- },
1595
- {
1596
- "epoch": 0.01,
1597
- "learning_rate": 4.7e-05,
1598
- "loss": 0.0,
1599
- "step": 265
1600
- },
1601
- {
1602
- "epoch": 0.01,
1603
- "learning_rate": 4.6800000000000006e-05,
1604
- "loss": 0.0,
1605
- "step": 266
1606
- },
1607
- {
1608
- "epoch": 0.01,
1609
- "learning_rate": 4.660000000000001e-05,
1610
- "loss": 0.0,
1611
- "step": 267
1612
- },
1613
- {
1614
- "epoch": 0.01,
1615
- "learning_rate": 4.64e-05,
1616
- "loss": 0.0,
1617
- "step": 268
1618
- },
1619
- {
1620
- "epoch": 0.01,
1621
- "learning_rate": 4.6200000000000005e-05,
1622
- "loss": 0.0,
1623
- "step": 269
1624
- },
1625
- {
1626
- "epoch": 0.01,
1627
- "learning_rate": 4.600000000000001e-05,
1628
- "loss": 0.0,
1629
- "step": 270
1630
- },
1631
- {
1632
- "epoch": 0.01,
1633
- "learning_rate": 4.58e-05,
1634
- "loss": 0.0,
1635
- "step": 271
1636
- },
1637
- {
1638
- "epoch": 0.01,
1639
- "learning_rate": 4.5600000000000004e-05,
1640
- "loss": 0.0,
1641
- "step": 272
1642
- },
1643
- {
1644
- "epoch": 0.01,
1645
- "learning_rate": 4.5400000000000006e-05,
1646
- "loss": 0.0,
1647
- "step": 273
1648
- },
1649
- {
1650
- "epoch": 0.01,
1651
- "learning_rate": 4.52e-05,
1652
- "loss": 0.0,
1653
- "step": 274
1654
- },
1655
- {
1656
- "epoch": 0.01,
1657
- "learning_rate": 4.5e-05,
1658
- "loss": 0.0,
1659
- "step": 275
1660
- },
1661
- {
1662
- "epoch": 0.01,
1663
- "learning_rate": 4.4800000000000005e-05,
1664
- "loss": 0.0,
1665
- "step": 276
1666
- },
1667
- {
1668
- "epoch": 0.01,
1669
- "learning_rate": 4.46e-05,
1670
- "loss": 0.0,
1671
- "step": 277
1672
- },
1673
- {
1674
- "epoch": 0.01,
1675
- "learning_rate": 4.44e-05,
1676
- "loss": 0.0,
1677
- "step": 278
1678
- },
1679
- {
1680
- "epoch": 0.01,
1681
- "learning_rate": 4.4200000000000004e-05,
1682
- "loss": 0.0,
1683
- "step": 279
1684
- },
1685
- {
1686
- "epoch": 0.01,
1687
- "learning_rate": 4.4000000000000006e-05,
1688
- "loss": 0.0,
1689
- "step": 280
1690
- },
1691
- {
1692
- "epoch": 0.01,
1693
- "learning_rate": 4.38e-05,
1694
- "loss": 0.0,
1695
- "step": 281
1696
- },
1697
- {
1698
- "epoch": 0.01,
1699
- "learning_rate": 4.36e-05,
1700
- "loss": 0.0,
1701
- "step": 282
1702
- },
1703
- {
1704
- "epoch": 0.01,
1705
- "learning_rate": 4.3400000000000005e-05,
1706
- "loss": 0.0,
1707
- "step": 283
1708
- },
1709
- {
1710
- "epoch": 0.01,
1711
- "learning_rate": 4.32e-05,
1712
- "loss": 0.0,
1713
- "step": 284
1714
- },
1715
- {
1716
- "epoch": 0.01,
1717
- "learning_rate": 4.3e-05,
1718
- "loss": 0.0,
1719
- "step": 285
1720
- },
1721
- {
1722
- "epoch": 0.01,
1723
- "learning_rate": 4.2800000000000004e-05,
1724
- "loss": 0.0,
1725
- "step": 286
1726
- },
1727
- {
1728
- "epoch": 0.01,
1729
- "learning_rate": 4.26e-05,
1730
- "loss": 0.0,
1731
- "step": 287
1732
- },
1733
- {
1734
- "epoch": 0.01,
1735
- "learning_rate": 4.24e-05,
1736
- "loss": 0.0,
1737
- "step": 288
1738
- },
1739
- {
1740
- "epoch": 0.01,
1741
- "learning_rate": 4.22e-05,
1742
- "loss": 0.0,
1743
- "step": 289
1744
- },
1745
- {
1746
- "epoch": 0.01,
1747
- "learning_rate": 4.2e-05,
1748
- "loss": 0.0,
1749
- "step": 290
1750
- },
1751
- {
1752
- "epoch": 0.01,
1753
- "learning_rate": 4.18e-05,
1754
- "loss": 0.0,
1755
- "step": 291
1756
- },
1757
- {
1758
- "epoch": 0.01,
1759
- "learning_rate": 4.16e-05,
1760
- "loss": 0.0,
1761
- "step": 292
1762
- },
1763
- {
1764
- "epoch": 0.01,
1765
- "learning_rate": 4.14e-05,
1766
- "loss": 0.0,
1767
- "step": 293
1768
- },
1769
- {
1770
- "epoch": 0.01,
1771
- "learning_rate": 4.12e-05,
1772
- "loss": 0.0,
1773
- "step": 294
1774
- },
1775
- {
1776
- "epoch": 0.01,
1777
- "learning_rate": 4.1e-05,
1778
- "loss": 0.0,
1779
- "step": 295
1780
- },
1781
- {
1782
- "epoch": 0.01,
1783
- "learning_rate": 4.08e-05,
1784
- "loss": 0.0,
1785
- "step": 296
1786
- },
1787
- {
1788
- "epoch": 0.01,
1789
- "learning_rate": 4.0600000000000004e-05,
1790
- "loss": 0.0,
1791
- "step": 297
1792
- },
1793
- {
1794
- "epoch": 0.01,
1795
- "learning_rate": 4.0400000000000006e-05,
1796
- "loss": 0.0,
1797
- "step": 298
1798
- },
1799
- {
1800
- "epoch": 0.01,
1801
- "learning_rate": 4.02e-05,
1802
- "loss": 0.0,
1803
- "step": 299
1804
- },
1805
- {
1806
- "epoch": 0.01,
1807
- "learning_rate": 4e-05,
1808
- "loss": 0.0,
1809
- "step": 300
1810
- },
1811
- {
1812
- "epoch": 0.01,
1813
- "learning_rate": 3.9800000000000005e-05,
1814
- "loss": 0.0,
1815
- "step": 301
1816
- },
1817
- {
1818
- "epoch": 0.01,
1819
- "learning_rate": 3.960000000000001e-05,
1820
- "loss": 0.0,
1821
- "step": 302
1822
- },
1823
- {
1824
- "epoch": 0.01,
1825
- "learning_rate": 3.94e-05,
1826
- "loss": 0.0,
1827
- "step": 303
1828
- },
1829
- {
1830
- "epoch": 0.01,
1831
- "learning_rate": 3.9200000000000004e-05,
1832
- "loss": 0.0,
1833
- "step": 304
1834
- },
1835
- {
1836
- "epoch": 0.01,
1837
- "learning_rate": 3.9000000000000006e-05,
1838
- "loss": 0.0,
1839
- "step": 305
1840
- },
1841
- {
1842
- "epoch": 0.01,
1843
- "learning_rate": 3.88e-05,
1844
- "loss": 0.0,
1845
- "step": 306
1846
- },
1847
- {
1848
- "epoch": 0.01,
1849
- "learning_rate": 3.86e-05,
1850
- "loss": 0.0,
1851
- "step": 307
1852
- },
1853
- {
1854
- "epoch": 0.01,
1855
- "learning_rate": 3.8400000000000005e-05,
1856
- "loss": 0.0,
1857
- "step": 308
1858
- },
1859
- {
1860
- "epoch": 0.01,
1861
- "learning_rate": 3.82e-05,
1862
- "loss": 0.0,
1863
- "step": 309
1864
- },
1865
- {
1866
- "epoch": 0.01,
1867
- "learning_rate": 3.8e-05,
1868
- "loss": 0.0,
1869
- "step": 310
1870
- },
1871
- {
1872
- "epoch": 0.01,
1873
- "learning_rate": 3.7800000000000004e-05,
1874
- "loss": 0.0,
1875
- "step": 311
1876
- },
1877
- {
1878
- "epoch": 0.01,
1879
- "learning_rate": 3.76e-05,
1880
- "loss": 0.0,
1881
- "step": 312
1882
- },
1883
- {
1884
- "epoch": 0.01,
1885
- "learning_rate": 3.74e-05,
1886
- "loss": 0.0,
1887
- "step": 313
1888
- },
1889
- {
1890
- "epoch": 0.01,
1891
- "learning_rate": 3.72e-05,
1892
- "loss": 0.0,
1893
- "step": 314
1894
- },
1895
- {
1896
- "epoch": 0.01,
1897
- "learning_rate": 3.7e-05,
1898
- "loss": 0.0,
1899
- "step": 315
1900
- },
1901
- {
1902
- "epoch": 0.01,
1903
- "learning_rate": 3.68e-05,
1904
- "loss": 0.0,
1905
- "step": 316
1906
- },
1907
- {
1908
- "epoch": 0.01,
1909
- "learning_rate": 3.66e-05,
1910
- "loss": 0.0,
1911
- "step": 317
1912
- },
1913
- {
1914
- "epoch": 0.01,
1915
- "learning_rate": 3.6400000000000004e-05,
1916
- "loss": 0.0,
1917
- "step": 318
1918
- },
1919
- {
1920
- "epoch": 0.01,
1921
- "learning_rate": 3.62e-05,
1922
- "loss": 0.0,
1923
- "step": 319
1924
- },
1925
- {
1926
- "epoch": 0.01,
1927
- "learning_rate": 3.6e-05,
1928
- "loss": 0.0,
1929
- "step": 320
1930
- },
1931
- {
1932
- "epoch": 0.01,
1933
- "learning_rate": 3.58e-05,
1934
- "loss": 0.0,
1935
- "step": 321
1936
- },
1937
- {
1938
- "epoch": 0.01,
1939
- "learning_rate": 3.56e-05,
1940
- "loss": 0.0,
1941
- "step": 322
1942
- },
1943
- {
1944
- "epoch": 0.01,
1945
- "learning_rate": 3.54e-05,
1946
- "loss": 0.0,
1947
- "step": 323
1948
- },
1949
- {
1950
- "epoch": 0.01,
1951
- "learning_rate": 3.52e-05,
1952
- "loss": 0.0,
1953
- "step": 324
1954
- },
1955
- {
1956
- "epoch": 0.01,
1957
- "learning_rate": 3.5e-05,
1958
- "loss": 0.0,
1959
- "step": 325
1960
- },
1961
- {
1962
- "epoch": 0.01,
1963
- "learning_rate": 3.48e-05,
1964
- "loss": 0.0,
1965
- "step": 326
1966
- },
1967
- {
1968
- "epoch": 0.01,
1969
- "learning_rate": 3.46e-05,
1970
- "loss": 0.0,
1971
- "step": 327
1972
- },
1973
- {
1974
- "epoch": 0.01,
1975
- "learning_rate": 3.4399999999999996e-05,
1976
- "loss": 0.0,
1977
- "step": 328
1978
- },
1979
- {
1980
- "epoch": 0.01,
1981
- "learning_rate": 3.4200000000000005e-05,
1982
- "loss": 0.0,
1983
- "step": 329
1984
- },
1985
- {
1986
- "epoch": 0.01,
1987
- "learning_rate": 3.4000000000000007e-05,
1988
- "loss": 0.0,
1989
- "step": 330
1990
- },
1991
- {
1992
- "epoch": 0.01,
1993
- "learning_rate": 3.38e-05,
1994
- "loss": 0.0,
1995
- "step": 331
1996
- },
1997
- {
1998
- "epoch": 0.01,
1999
- "learning_rate": 3.3600000000000004e-05,
2000
- "loss": 0.0,
2001
- "step": 332
2002
- },
2003
- {
2004
- "epoch": 0.01,
2005
- "learning_rate": 3.3400000000000005e-05,
2006
- "loss": 0.0,
2007
- "step": 333
2008
- },
2009
- {
2010
- "epoch": 0.01,
2011
- "learning_rate": 3.32e-05,
2012
- "loss": 0.0,
2013
- "step": 334
2014
- },
2015
- {
2016
- "epoch": 0.01,
2017
- "learning_rate": 3.3e-05,
2018
- "loss": 0.0,
2019
- "step": 335
2020
- },
2021
- {
2022
- "epoch": 0.01,
2023
- "learning_rate": 3.2800000000000004e-05,
2024
- "loss": 0.0,
2025
- "step": 336
2026
- },
2027
- {
2028
- "epoch": 0.01,
2029
- "learning_rate": 3.26e-05,
2030
- "loss": 0.0,
2031
- "step": 337
2032
- },
2033
- {
2034
- "epoch": 0.01,
2035
- "learning_rate": 3.24e-05,
2036
- "loss": 0.0,
2037
- "step": 338
2038
- },
2039
- {
2040
- "epoch": 0.01,
2041
- "learning_rate": 3.2200000000000003e-05,
2042
- "loss": 0.0,
2043
- "step": 339
2044
- },
2045
- {
2046
- "epoch": 0.01,
2047
- "learning_rate": 3.2000000000000005e-05,
2048
- "loss": 0.0,
2049
- "step": 340
2050
- },
2051
- {
2052
- "epoch": 0.01,
2053
- "learning_rate": 3.18e-05,
2054
- "loss": 0.0,
2055
- "step": 341
2056
- },
2057
- {
2058
- "epoch": 0.01,
2059
- "learning_rate": 3.16e-05,
2060
- "loss": 0.0,
2061
- "step": 342
2062
- },
2063
- {
2064
- "epoch": 0.01,
2065
- "learning_rate": 3.1400000000000004e-05,
2066
- "loss": 0.0,
2067
- "step": 343
2068
- },
2069
- {
2070
- "epoch": 0.01,
2071
- "learning_rate": 3.12e-05,
2072
- "loss": 0.0,
2073
- "step": 344
2074
- },
2075
- {
2076
- "epoch": 0.01,
2077
- "learning_rate": 3.1e-05,
2078
- "loss": 0.0,
2079
- "step": 345
2080
- },
2081
- {
2082
- "epoch": 0.01,
2083
- "learning_rate": 3.08e-05,
2084
- "loss": 0.0,
2085
- "step": 346
2086
- },
2087
- {
2088
- "epoch": 0.01,
2089
- "learning_rate": 3.06e-05,
2090
- "loss": 0.0,
2091
- "step": 347
2092
- },
2093
- {
2094
- "epoch": 0.01,
2095
- "learning_rate": 3.04e-05,
2096
- "loss": 0.0,
2097
- "step": 348
2098
- },
2099
- {
2100
- "epoch": 0.01,
2101
- "learning_rate": 3.02e-05,
2102
- "loss": 0.0,
2103
- "step": 349
2104
- },
2105
- {
2106
- "epoch": 0.01,
2107
- "learning_rate": 3e-05,
2108
- "loss": 0.0,
2109
- "step": 350
2110
- },
2111
- {
2112
- "epoch": 0.01,
2113
- "learning_rate": 2.98e-05,
2114
- "loss": 0.0,
2115
- "step": 351
2116
- },
2117
- {
2118
- "epoch": 0.01,
2119
- "learning_rate": 2.96e-05,
2120
- "loss": 0.0,
2121
- "step": 352
2122
- },
2123
- {
2124
- "epoch": 0.01,
2125
- "learning_rate": 2.94e-05,
2126
- "loss": 0.0,
2127
- "step": 353
2128
- },
2129
- {
2130
- "epoch": 0.01,
2131
- "learning_rate": 2.9199999999999998e-05,
2132
- "loss": 0.0,
2133
- "step": 354
2134
- },
2135
- {
2136
- "epoch": 0.01,
2137
- "learning_rate": 2.9e-05,
2138
- "loss": 0.0,
2139
- "step": 355
2140
- },
2141
- {
2142
- "epoch": 0.01,
2143
- "learning_rate": 2.88e-05,
2144
- "loss": 0.0,
2145
- "step": 356
2146
- },
2147
- {
2148
- "epoch": 0.01,
2149
- "learning_rate": 2.86e-05,
2150
- "loss": 0.0,
2151
- "step": 357
2152
- },
2153
- {
2154
- "epoch": 0.01,
2155
- "learning_rate": 2.84e-05,
2156
- "loss": 0.0,
2157
- "step": 358
2158
- },
2159
- {
2160
- "epoch": 0.01,
2161
- "learning_rate": 2.8199999999999998e-05,
2162
- "loss": 0.0,
2163
- "step": 359
2164
- },
2165
- {
2166
- "epoch": 0.01,
2167
- "learning_rate": 2.8000000000000003e-05,
2168
- "loss": 0.0,
2169
- "step": 360
2170
- },
2171
- {
2172
- "epoch": 0.01,
2173
- "learning_rate": 2.7800000000000005e-05,
2174
- "loss": 0.0,
2175
- "step": 361
2176
- },
2177
- {
2178
- "epoch": 0.01,
2179
- "learning_rate": 2.7600000000000003e-05,
2180
- "loss": 0.0,
2181
- "step": 362
2182
- },
2183
- {
2184
- "epoch": 0.01,
2185
- "learning_rate": 2.7400000000000002e-05,
2186
- "loss": 0.0,
2187
- "step": 363
2188
- },
2189
- {
2190
- "epoch": 0.01,
2191
- "learning_rate": 2.7200000000000004e-05,
2192
- "loss": 0.0,
2193
- "step": 364
2194
- },
2195
- {
2196
- "epoch": 0.01,
2197
- "learning_rate": 2.7000000000000002e-05,
2198
- "loss": 0.0,
2199
- "step": 365
2200
- },
2201
- {
2202
- "epoch": 0.01,
2203
- "learning_rate": 2.6800000000000004e-05,
2204
- "loss": 0.0,
2205
- "step": 366
2206
- },
2207
- {
2208
- "epoch": 0.01,
2209
- "learning_rate": 2.6600000000000003e-05,
2210
- "loss": 0.0,
2211
- "step": 367
2212
- },
2213
- {
2214
- "epoch": 0.01,
2215
- "learning_rate": 2.64e-05,
2216
- "loss": 0.0,
2217
- "step": 368
2218
- },
2219
- {
2220
- "epoch": 0.01,
2221
- "learning_rate": 2.6200000000000003e-05,
2222
- "loss": 0.0,
2223
- "step": 369
2224
- },
2225
- {
2226
- "epoch": 0.01,
2227
- "learning_rate": 2.6000000000000002e-05,
2228
- "loss": 0.0,
2229
- "step": 370
2230
- },
2231
- {
2232
- "epoch": 0.01,
2233
- "learning_rate": 2.58e-05,
2234
- "loss": 0.0,
2235
- "step": 371
2236
- },
2237
- {
2238
- "epoch": 0.01,
2239
- "learning_rate": 2.5600000000000002e-05,
2240
- "loss": 0.0,
2241
- "step": 372
2242
- },
2243
- {
2244
- "epoch": 0.01,
2245
- "learning_rate": 2.54e-05,
2246
- "loss": 0.0,
2247
- "step": 373
2248
- },
2249
- {
2250
- "epoch": 0.01,
2251
- "learning_rate": 2.5200000000000003e-05,
2252
- "loss": 0.0,
2253
- "step": 374
2254
- },
2255
- {
2256
- "epoch": 0.01,
2257
- "learning_rate": 2.5e-05,
2258
- "loss": 0.0,
2259
- "step": 375
2260
- },
2261
- {
2262
- "epoch": 0.01,
2263
- "learning_rate": 2.48e-05,
2264
- "loss": 0.0,
2265
- "step": 376
2266
- },
2267
- {
2268
- "epoch": 0.01,
2269
- "learning_rate": 2.46e-05,
2270
- "loss": 0.0,
2271
- "step": 377
2272
- },
2273
- {
2274
- "epoch": 0.01,
2275
- "learning_rate": 2.44e-05,
2276
- "loss": 0.0,
2277
- "step": 378
2278
- },
2279
- {
2280
- "epoch": 0.01,
2281
- "learning_rate": 2.4200000000000002e-05,
2282
- "loss": 0.0,
2283
- "step": 379
2284
- },
2285
- {
2286
- "epoch": 0.01,
2287
- "learning_rate": 2.4e-05,
2288
- "loss": 0.0,
2289
- "step": 380
2290
- },
2291
- {
2292
- "epoch": 0.01,
2293
- "learning_rate": 2.38e-05,
2294
- "loss": 0.0,
2295
- "step": 381
2296
- },
2297
- {
2298
- "epoch": 0.01,
2299
- "learning_rate": 2.36e-05,
2300
- "loss": 0.0,
2301
- "step": 382
2302
- },
2303
- {
2304
- "epoch": 0.01,
2305
- "learning_rate": 2.3400000000000003e-05,
2306
- "loss": 0.0,
2307
- "step": 383
2308
- },
2309
- {
2310
- "epoch": 0.01,
2311
- "learning_rate": 2.32e-05,
2312
- "loss": 0.0,
2313
- "step": 384
2314
- },
2315
- {
2316
- "epoch": 0.01,
2317
- "learning_rate": 2.3000000000000003e-05,
2318
- "loss": 0.0,
2319
- "step": 385
2320
- },
2321
- {
2322
- "epoch": 0.01,
2323
- "learning_rate": 2.2800000000000002e-05,
2324
- "loss": 0.0,
2325
- "step": 386
2326
- },
2327
- {
2328
- "epoch": 0.01,
2329
- "learning_rate": 2.26e-05,
2330
- "loss": 0.0,
2331
- "step": 387
2332
- },
2333
- {
2334
- "epoch": 0.01,
2335
- "learning_rate": 2.2400000000000002e-05,
2336
- "loss": 0.0,
2337
- "step": 388
2338
- },
2339
- {
2340
- "epoch": 0.01,
2341
- "learning_rate": 2.22e-05,
2342
- "loss": 0.0,
2343
- "step": 389
2344
- },
2345
- {
2346
- "epoch": 0.01,
2347
- "learning_rate": 2.2000000000000003e-05,
2348
- "loss": 0.0,
2349
- "step": 390
2350
- },
2351
- {
2352
- "epoch": 0.02,
2353
- "learning_rate": 2.18e-05,
2354
- "loss": 0.0,
2355
- "step": 391
2356
- },
2357
- {
2358
- "epoch": 0.02,
2359
- "learning_rate": 2.16e-05,
2360
- "loss": 0.0,
2361
- "step": 392
2362
- },
2363
- {
2364
- "epoch": 0.02,
2365
- "learning_rate": 2.1400000000000002e-05,
2366
- "loss": 0.0,
2367
- "step": 393
2368
- },
2369
- {
2370
- "epoch": 0.02,
2371
- "learning_rate": 2.12e-05,
2372
- "loss": 0.0,
2373
- "step": 394
2374
- },
2375
- {
2376
- "epoch": 0.02,
2377
- "learning_rate": 2.1e-05,
2378
- "loss": 0.0,
2379
- "step": 395
2380
- },
2381
- {
2382
- "epoch": 0.02,
2383
- "learning_rate": 2.08e-05,
2384
- "loss": 0.0,
2385
- "step": 396
2386
- },
2387
- {
2388
- "epoch": 0.02,
2389
- "learning_rate": 2.06e-05,
2390
- "loss": 0.0,
2391
- "step": 397
2392
- },
2393
- {
2394
- "epoch": 0.02,
2395
- "learning_rate": 2.04e-05,
2396
- "loss": 0.0,
2397
- "step": 398
2398
- },
2399
- {
2400
- "epoch": 0.02,
2401
- "learning_rate": 2.0200000000000003e-05,
2402
- "loss": 0.0,
2403
- "step": 399
2404
- },
2405
- {
2406
- "epoch": 0.02,
2407
- "learning_rate": 2e-05,
2408
- "loss": 0.0,
2409
- "step": 400
2410
- },
2411
- {
2412
- "epoch": 0.02,
2413
- "learning_rate": 1.9800000000000004e-05,
2414
- "loss": 0.0,
2415
- "step": 401
2416
- },
2417
- {
2418
- "epoch": 0.02,
2419
- "learning_rate": 1.9600000000000002e-05,
2420
- "loss": 0.0,
2421
- "step": 402
2422
- },
2423
- {
2424
- "epoch": 0.02,
2425
- "learning_rate": 1.94e-05,
2426
- "loss": 0.0,
2427
- "step": 403
2428
- },
2429
- {
2430
- "epoch": 0.02,
2431
- "learning_rate": 1.9200000000000003e-05,
2432
- "loss": 0.0,
2433
- "step": 404
2434
- },
2435
- {
2436
- "epoch": 0.02,
2437
- "learning_rate": 1.9e-05,
2438
- "loss": 0.0,
2439
- "step": 405
2440
- },
2441
- {
2442
- "epoch": 0.02,
2443
- "learning_rate": 1.88e-05,
2444
- "loss": 0.0,
2445
- "step": 406
2446
- },
2447
- {
2448
- "epoch": 0.02,
2449
- "learning_rate": 1.86e-05,
2450
- "loss": 0.0,
2451
- "step": 407
2452
- },
2453
- {
2454
- "epoch": 0.02,
2455
- "learning_rate": 1.84e-05,
2456
- "loss": 0.0,
2457
- "step": 408
2458
- },
2459
- {
2460
- "epoch": 0.02,
2461
- "learning_rate": 1.8200000000000002e-05,
2462
- "loss": 0.0,
2463
- "step": 409
2464
- },
2465
- {
2466
- "epoch": 0.02,
2467
- "learning_rate": 1.8e-05,
2468
- "loss": 0.0,
2469
- "step": 410
2470
- },
2471
- {
2472
- "epoch": 0.02,
2473
- "learning_rate": 1.78e-05,
2474
- "loss": 0.0,
2475
- "step": 411
2476
- },
2477
- {
2478
- "epoch": 0.02,
2479
- "learning_rate": 1.76e-05,
2480
- "loss": 0.0,
2481
- "step": 412
2482
- },
2483
- {
2484
- "epoch": 0.02,
2485
- "learning_rate": 1.74e-05,
2486
- "loss": 0.0,
2487
- "step": 413
2488
- },
2489
- {
2490
- "epoch": 0.02,
2491
- "learning_rate": 1.7199999999999998e-05,
2492
- "loss": 0.0,
2493
- "step": 414
2494
- },
2495
- {
2496
- "epoch": 0.02,
2497
- "learning_rate": 1.7000000000000003e-05,
2498
- "loss": 0.0,
2499
- "step": 415
2500
- },
2501
- {
2502
- "epoch": 0.02,
2503
- "learning_rate": 1.6800000000000002e-05,
2504
- "loss": 0.0,
2505
- "step": 416
2506
- },
2507
- {
2508
- "epoch": 0.02,
2509
- "learning_rate": 1.66e-05,
2510
- "loss": 0.0,
2511
- "step": 417
2512
- },
2513
- {
2514
- "epoch": 0.02,
2515
- "learning_rate": 1.6400000000000002e-05,
2516
- "loss": 0.0,
2517
- "step": 418
2518
- },
2519
- {
2520
- "epoch": 0.02,
2521
- "learning_rate": 1.62e-05,
2522
- "loss": 0.0,
2523
- "step": 419
2524
- },
2525
- {
2526
- "epoch": 0.02,
2527
- "learning_rate": 1.6000000000000003e-05,
2528
- "loss": 0.0,
2529
- "step": 420
2530
- },
2531
- {
2532
- "epoch": 0.02,
2533
- "learning_rate": 1.58e-05,
2534
- "loss": 0.0,
2535
- "step": 421
2536
- },
2537
- {
2538
- "epoch": 0.02,
2539
- "learning_rate": 1.56e-05,
2540
- "loss": 0.0,
2541
- "step": 422
2542
- },
2543
- {
2544
- "epoch": 0.02,
2545
- "learning_rate": 1.54e-05,
2546
- "loss": 0.0,
2547
- "step": 423
2548
- },
2549
- {
2550
- "epoch": 0.02,
2551
- "learning_rate": 1.52e-05,
2552
- "loss": 0.0,
2553
- "step": 424
2554
- },
2555
- {
2556
- "epoch": 0.02,
2557
- "learning_rate": 1.5e-05,
2558
- "loss": 0.0,
2559
- "step": 425
2560
- },
2561
- {
2562
- "epoch": 0.02,
2563
- "learning_rate": 1.48e-05,
2564
- "loss": 0.0,
2565
- "step": 426
2566
- },
2567
- {
2568
- "epoch": 0.02,
2569
- "learning_rate": 1.4599999999999999e-05,
2570
- "loss": 0.0,
2571
- "step": 427
2572
- },
2573
- {
2574
- "epoch": 0.02,
2575
- "learning_rate": 1.44e-05,
2576
- "loss": 0.0,
2577
- "step": 428
2578
- },
2579
- {
2580
- "epoch": 0.02,
2581
- "learning_rate": 1.42e-05,
2582
- "loss": 0.0,
2583
- "step": 429
2584
- },
2585
- {
2586
- "epoch": 0.02,
2587
- "learning_rate": 1.4000000000000001e-05,
2588
- "loss": 0.0,
2589
- "step": 430
2590
- },
2591
- {
2592
- "epoch": 0.02,
2593
- "learning_rate": 1.3800000000000002e-05,
2594
- "loss": 0.0,
2595
- "step": 431
2596
- },
2597
- {
2598
- "epoch": 0.02,
2599
- "learning_rate": 1.3600000000000002e-05,
2600
- "loss": 0.0,
2601
- "step": 432
2602
- },
2603
- {
2604
- "epoch": 0.02,
2605
- "learning_rate": 1.3400000000000002e-05,
2606
- "loss": 0.0,
2607
- "step": 433
2608
- },
2609
- {
2610
- "epoch": 0.02,
2611
- "learning_rate": 1.32e-05,
2612
- "loss": 0.0,
2613
- "step": 434
2614
- },
2615
- {
2616
- "epoch": 0.02,
2617
- "learning_rate": 1.3000000000000001e-05,
2618
- "loss": 0.0,
2619
- "step": 435
2620
- },
2621
- {
2622
- "epoch": 0.02,
2623
- "learning_rate": 1.2800000000000001e-05,
2624
- "loss": 0.0,
2625
- "step": 436
2626
- },
2627
- {
2628
- "epoch": 0.02,
2629
- "learning_rate": 1.2600000000000001e-05,
2630
- "loss": 0.0,
2631
- "step": 437
2632
- },
2633
- {
2634
- "epoch": 0.02,
2635
- "learning_rate": 1.24e-05,
2636
- "loss": 0.0,
2637
- "step": 438
2638
- },
2639
- {
2640
- "epoch": 0.02,
2641
- "learning_rate": 1.22e-05,
2642
- "loss": 0.0,
2643
- "step": 439
2644
- },
2645
- {
2646
- "epoch": 0.02,
2647
- "learning_rate": 1.2e-05,
2648
- "loss": 0.0,
2649
- "step": 440
2650
- },
2651
- {
2652
- "epoch": 0.02,
2653
- "learning_rate": 1.18e-05,
2654
- "loss": 0.0,
2655
- "step": 441
2656
- },
2657
- {
2658
- "epoch": 0.02,
2659
- "learning_rate": 1.16e-05,
2660
- "loss": 0.0,
2661
- "step": 442
2662
- },
2663
- {
2664
- "epoch": 0.02,
2665
- "learning_rate": 1.1400000000000001e-05,
2666
- "loss": 0.0,
2667
- "step": 443
2668
- },
2669
- {
2670
- "epoch": 0.02,
2671
- "learning_rate": 1.1200000000000001e-05,
2672
- "loss": 0.0,
2673
- "step": 444
2674
- },
2675
- {
2676
- "epoch": 0.02,
2677
- "learning_rate": 1.1000000000000001e-05,
2678
- "loss": 0.0,
2679
- "step": 445
2680
- },
2681
- {
2682
- "epoch": 0.02,
2683
- "learning_rate": 1.08e-05,
2684
- "loss": 0.0,
2685
- "step": 446
2686
- },
2687
- {
2688
- "epoch": 0.02,
2689
- "learning_rate": 1.06e-05,
2690
- "loss": 0.0,
2691
- "step": 447
2692
- },
2693
- {
2694
- "epoch": 0.02,
2695
- "learning_rate": 1.04e-05,
2696
- "loss": 0.0,
2697
- "step": 448
2698
- },
2699
- {
2700
- "epoch": 0.02,
2701
- "learning_rate": 1.02e-05,
2702
- "loss": 0.0,
2703
- "step": 449
2704
- },
2705
- {
2706
- "epoch": 0.02,
2707
- "learning_rate": 1e-05,
2708
- "loss": 0.0,
2709
- "step": 450
2710
- },
2711
- {
2712
- "epoch": 0.02,
2713
- "learning_rate": 9.800000000000001e-06,
2714
- "loss": 0.0,
2715
- "step": 451
2716
- },
2717
- {
2718
- "epoch": 0.02,
2719
- "learning_rate": 9.600000000000001e-06,
2720
- "loss": 0.0,
2721
- "step": 452
2722
- },
2723
- {
2724
- "epoch": 0.02,
2725
- "learning_rate": 9.4e-06,
2726
- "loss": 0.0,
2727
- "step": 453
2728
- },
2729
- {
2730
- "epoch": 0.02,
2731
- "learning_rate": 9.2e-06,
2732
- "loss": 0.0,
2733
- "step": 454
2734
- },
2735
- {
2736
- "epoch": 0.02,
2737
- "learning_rate": 9e-06,
2738
- "loss": 0.0,
2739
- "step": 455
2740
- },
2741
- {
2742
- "epoch": 0.02,
2743
- "learning_rate": 8.8e-06,
2744
- "loss": 0.0,
2745
- "step": 456
2746
- },
2747
- {
2748
- "epoch": 0.02,
2749
- "learning_rate": 8.599999999999999e-06,
2750
- "loss": 0.0,
2751
- "step": 457
2752
- },
2753
- {
2754
- "epoch": 0.02,
2755
- "learning_rate": 8.400000000000001e-06,
2756
- "loss": 0.0,
2757
- "step": 458
2758
- },
2759
- {
2760
- "epoch": 0.02,
2761
- "learning_rate": 8.200000000000001e-06,
2762
- "loss": 0.0,
2763
- "step": 459
2764
- },
2765
- {
2766
- "epoch": 0.02,
2767
- "learning_rate": 8.000000000000001e-06,
2768
- "loss": 0.0,
2769
- "step": 460
2770
- },
2771
- {
2772
- "epoch": 0.02,
2773
- "learning_rate": 7.8e-06,
2774
- "loss": 0.0,
2775
- "step": 461
2776
- },
2777
- {
2778
- "epoch": 0.02,
2779
- "learning_rate": 7.6e-06,
2780
- "loss": 0.0,
2781
- "step": 462
2782
- },
2783
- {
2784
- "epoch": 0.02,
2785
- "learning_rate": 7.4e-06,
2786
- "loss": 0.0,
2787
- "step": 463
2788
- },
2789
- {
2790
- "epoch": 0.02,
2791
- "learning_rate": 7.2e-06,
2792
- "loss": 0.0,
2793
- "step": 464
2794
- },
2795
- {
2796
- "epoch": 0.02,
2797
- "learning_rate": 7.000000000000001e-06,
2798
- "loss": 0.0,
2799
- "step": 465
2800
- },
2801
- {
2802
- "epoch": 0.02,
2803
- "learning_rate": 6.800000000000001e-06,
2804
- "loss": 0.0,
2805
- "step": 466
2806
- },
2807
- {
2808
- "epoch": 0.02,
2809
- "learning_rate": 6.6e-06,
2810
- "loss": 0.0,
2811
- "step": 467
2812
- },
2813
- {
2814
- "epoch": 0.02,
2815
- "learning_rate": 6.4000000000000006e-06,
2816
- "loss": 0.0,
2817
- "step": 468
2818
- },
2819
- {
2820
- "epoch": 0.02,
2821
- "learning_rate": 6.2e-06,
2822
- "loss": 0.0,
2823
- "step": 469
2824
- },
2825
- {
2826
- "epoch": 0.02,
2827
- "learning_rate": 6e-06,
2828
- "loss": 0.0,
2829
- "step": 470
2830
- },
2831
- {
2832
- "epoch": 0.02,
2833
- "learning_rate": 5.8e-06,
2834
- "loss": 0.0,
2835
- "step": 471
2836
- },
2837
- {
2838
- "epoch": 0.02,
2839
- "learning_rate": 5.600000000000001e-06,
2840
- "loss": 0.0,
2841
- "step": 472
2842
- },
2843
- {
2844
- "epoch": 0.02,
2845
- "learning_rate": 5.4e-06,
2846
- "loss": 0.0,
2847
- "step": 473
2848
- },
2849
- {
2850
- "epoch": 0.02,
2851
- "learning_rate": 5.2e-06,
2852
- "loss": 0.0,
2853
- "step": 474
2854
- },
2855
- {
2856
- "epoch": 0.02,
2857
- "learning_rate": 5e-06,
2858
- "loss": 0.0,
2859
- "step": 475
2860
- },
2861
- {
2862
- "epoch": 0.02,
2863
- "learning_rate": 4.800000000000001e-06,
2864
- "loss": 0.0,
2865
- "step": 476
2866
- },
2867
- {
2868
- "epoch": 0.02,
2869
- "learning_rate": 4.6e-06,
2870
- "loss": 0.0,
2871
- "step": 477
2872
- },
2873
- {
2874
- "epoch": 0.02,
2875
- "learning_rate": 4.4e-06,
2876
- "loss": 0.0,
2877
- "step": 478
2878
- },
2879
- {
2880
- "epoch": 0.02,
2881
- "learning_rate": 4.2000000000000004e-06,
2882
- "loss": 0.0,
2883
- "step": 479
2884
- },
2885
- {
2886
- "epoch": 0.02,
2887
- "learning_rate": 4.000000000000001e-06,
2888
- "loss": 0.0,
2889
- "step": 480
2890
- },
2891
- {
2892
- "epoch": 0.02,
2893
- "learning_rate": 3.8e-06,
2894
- "loss": 0.0,
2895
- "step": 481
2896
- },
2897
- {
2898
- "epoch": 0.02,
2899
- "learning_rate": 3.6e-06,
2900
- "loss": 0.0,
2901
- "step": 482
2902
- },
2903
- {
2904
- "epoch": 0.02,
2905
- "learning_rate": 3.4000000000000005e-06,
2906
- "loss": 0.0,
2907
- "step": 483
2908
- },
2909
- {
2910
- "epoch": 0.02,
2911
- "learning_rate": 3.2000000000000003e-06,
2912
- "loss": 0.0,
2913
- "step": 484
2914
- },
2915
- {
2916
- "epoch": 0.02,
2917
- "learning_rate": 3e-06,
2918
- "loss": 0.0,
2919
- "step": 485
2920
- },
2921
- {
2922
- "epoch": 0.02,
2923
- "learning_rate": 2.8000000000000003e-06,
2924
- "loss": 0.0,
2925
- "step": 486
2926
- },
2927
- {
2928
- "epoch": 0.02,
2929
- "learning_rate": 2.6e-06,
2930
- "loss": 0.0,
2931
- "step": 487
2932
- },
2933
- {
2934
- "epoch": 0.02,
2935
- "learning_rate": 2.4000000000000003e-06,
2936
- "loss": 0.0,
2937
- "step": 488
2938
- },
2939
- {
2940
- "epoch": 0.02,
2941
- "learning_rate": 2.2e-06,
2942
- "loss": 0.0,
2943
- "step": 489
2944
- },
2945
- {
2946
- "epoch": 0.02,
2947
- "learning_rate": 2.0000000000000003e-06,
2948
- "loss": 0.0,
2949
- "step": 490
2950
- },
2951
- {
2952
- "epoch": 0.02,
2953
- "learning_rate": 1.8e-06,
2954
- "loss": 0.0,
2955
- "step": 491
2956
- },
2957
- {
2958
- "epoch": 0.02,
2959
- "learning_rate": 1.6000000000000001e-06,
2960
- "loss": 0.0,
2961
- "step": 492
2962
- },
2963
- {
2964
- "epoch": 0.02,
2965
- "learning_rate": 1.4000000000000001e-06,
2966
- "loss": 0.0,
2967
- "step": 493
2968
- },
2969
- {
2970
- "epoch": 0.02,
2971
- "learning_rate": 1.2000000000000002e-06,
2972
- "loss": 0.0,
2973
- "step": 494
2974
- },
2975
- {
2976
- "epoch": 0.02,
2977
- "learning_rate": 1.0000000000000002e-06,
2978
- "loss": 0.0,
2979
- "step": 495
2980
- },
2981
- {
2982
- "epoch": 0.02,
2983
- "learning_rate": 8.000000000000001e-07,
2984
- "loss": 0.0,
2985
- "step": 496
2986
- },
2987
- {
2988
- "epoch": 0.02,
2989
- "learning_rate": 6.000000000000001e-07,
2990
- "loss": 0.0,
2991
- "step": 497
2992
- },
2993
- {
2994
- "epoch": 0.02,
2995
- "learning_rate": 4.0000000000000003e-07,
2996
- "loss": 0.0,
2997
- "step": 498
2998
- },
2999
- {
3000
- "epoch": 0.02,
3001
- "learning_rate": 2.0000000000000002e-07,
3002
- "loss": 0.0,
3003
- "step": 499
3004
- },
3005
- {
3006
- "epoch": 0.02,
3007
- "learning_rate": 0.0,
3008
- "loss": 0.0,
3009
- "step": 500
3010
- },
3011
- {
3012
- "epoch": 0.02,
3013
- "step": 500,
3014
- "total_flos": 4.1364075053056e+16,
3015
- "train_loss": 0.00219873046875,
3016
- "train_runtime": 130.5539,
3017
- "train_samples_per_second": 7.66,
3018
- "train_steps_per_second": 3.83
3019
- }
3020
- ],
3021
- "logging_steps": 1.0,
3022
- "max_steps": 500,
3023
- "num_train_epochs": 1,
3024
- "save_steps": 50,
3025
- "total_flos": 4.1364075053056e+16,
3026
- "trial_name": null,
3027
- "trial_params": null
3028
- }