sollamon commited on
Commit
9d4c937
·
verified ·
1 Parent(s): b659372

Delete trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +0 -3025
trainer_state.json DELETED
@@ -1,3025 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 18.383685286886188,
5
- "global_step": 250000,
6
- "is_hyper_param_search": false,
7
- "is_local_process_zero": true,
8
- "is_world_process_zero": true,
9
- "log_history": [
10
- {
11
- "epoch": 0.04,
12
- "learning_rate": 1e-05,
13
- "loss": 7.5478,
14
- "step": 500
15
- },
16
- {
17
- "epoch": 0.07,
18
- "learning_rate": 2e-05,
19
- "loss": 5.8238,
20
- "step": 1000
21
- },
22
- {
23
- "epoch": 0.11,
24
- "learning_rate": 3e-05,
25
- "loss": 5.0177,
26
- "step": 1500
27
- },
28
- {
29
- "epoch": 0.15,
30
- "learning_rate": 4e-05,
31
- "loss": 4.481,
32
- "step": 2000
33
- },
34
- {
35
- "epoch": 0.18,
36
- "learning_rate": 5e-05,
37
- "loss": 4.1159,
38
- "step": 2500
39
- },
40
- {
41
- "epoch": 0.22,
42
- "learning_rate": 6e-05,
43
- "loss": 3.8654,
44
- "step": 3000
45
- },
46
- {
47
- "epoch": 0.26,
48
- "learning_rate": 7.000000000000001e-05,
49
- "loss": 3.6867,
50
- "step": 3500
51
- },
52
- {
53
- "epoch": 0.29,
54
- "learning_rate": 8e-05,
55
- "loss": 3.5543,
56
- "step": 4000
57
- },
58
- {
59
- "epoch": 0.33,
60
- "learning_rate": 8.999999999999999e-05,
61
- "loss": 3.4529,
62
- "step": 4500
63
- },
64
- {
65
- "epoch": 0.37,
66
- "learning_rate": 0.0001,
67
- "loss": 3.3706,
68
- "step": 5000
69
- },
70
- {
71
- "epoch": 0.4,
72
- "learning_rate": 0.00011,
73
- "loss": 3.3042,
74
- "step": 5500
75
- },
76
- {
77
- "epoch": 0.44,
78
- "learning_rate": 0.00012,
79
- "loss": 3.2473,
80
- "step": 6000
81
- },
82
- {
83
- "epoch": 0.48,
84
- "learning_rate": 0.00013000000000000002,
85
- "loss": 3.2044,
86
- "step": 6500
87
- },
88
- {
89
- "epoch": 0.51,
90
- "learning_rate": 0.00014000000000000001,
91
- "loss": 3.1616,
92
- "step": 7000
93
- },
94
- {
95
- "epoch": 0.55,
96
- "learning_rate": 0.00015,
97
- "loss": 3.1317,
98
- "step": 7500
99
- },
100
- {
101
- "epoch": 0.59,
102
- "learning_rate": 0.00016,
103
- "loss": 3.1003,
104
- "step": 8000
105
- },
106
- {
107
- "epoch": 0.63,
108
- "learning_rate": 0.00017,
109
- "loss": 3.0747,
110
- "step": 8500
111
- },
112
- {
113
- "epoch": 0.66,
114
- "learning_rate": 0.00017999999999999998,
115
- "loss": 3.0469,
116
- "step": 9000
117
- },
118
- {
119
- "epoch": 0.7,
120
- "learning_rate": 0.00019,
121
- "loss": 3.0317,
122
- "step": 9500
123
- },
124
- {
125
- "epoch": 0.74,
126
- "learning_rate": 0.0002,
127
- "loss": 3.0075,
128
- "step": 10000
129
- },
130
- {
131
- "epoch": 0.77,
132
- "learning_rate": 0.00021,
133
- "loss": 2.9949,
134
- "step": 10500
135
- },
136
- {
137
- "epoch": 0.81,
138
- "learning_rate": 0.00022,
139
- "loss": 2.9786,
140
- "step": 11000
141
- },
142
- {
143
- "epoch": 0.85,
144
- "learning_rate": 0.00023,
145
- "loss": 2.9666,
146
- "step": 11500
147
- },
148
- {
149
- "epoch": 0.88,
150
- "learning_rate": 0.00024,
151
- "loss": 2.9489,
152
- "step": 12000
153
- },
154
- {
155
- "epoch": 0.92,
156
- "learning_rate": 0.00025,
157
- "loss": 2.9409,
158
- "step": 12500
159
- },
160
- {
161
- "epoch": 0.96,
162
- "learning_rate": 0.00026000000000000003,
163
- "loss": 2.9302,
164
- "step": 13000
165
- },
166
- {
167
- "epoch": 0.99,
168
- "learning_rate": 0.00027,
169
- "loss": 2.9231,
170
- "step": 13500
171
- },
172
- {
173
- "epoch": 1.03,
174
- "learning_rate": 0.00028000000000000003,
175
- "loss": 2.8963,
176
- "step": 14000
177
- },
178
- {
179
- "epoch": 1.07,
180
- "learning_rate": 0.00029,
181
- "loss": 2.8837,
182
- "step": 14500
183
- },
184
- {
185
- "epoch": 1.1,
186
- "learning_rate": 0.0003,
187
- "loss": 2.8798,
188
- "step": 15000
189
- },
190
- {
191
- "epoch": 1.14,
192
- "learning_rate": 0.00031,
193
- "loss": 2.8784,
194
- "step": 15500
195
- },
196
- {
197
- "epoch": 1.18,
198
- "learning_rate": 0.00032,
199
- "loss": 2.8685,
200
- "step": 16000
201
- },
202
- {
203
- "epoch": 1.21,
204
- "learning_rate": 0.00033,
205
- "loss": 2.8641,
206
- "step": 16500
207
- },
208
- {
209
- "epoch": 1.25,
210
- "learning_rate": 0.00034,
211
- "loss": 2.8616,
212
- "step": 17000
213
- },
214
- {
215
- "epoch": 1.29,
216
- "learning_rate": 0.00035,
217
- "loss": 2.8567,
218
- "step": 17500
219
- },
220
- {
221
- "epoch": 1.32,
222
- "learning_rate": 0.00035999999999999997,
223
- "loss": 2.8544,
224
- "step": 18000
225
- },
226
- {
227
- "epoch": 1.36,
228
- "learning_rate": 0.00037,
229
- "loss": 2.846,
230
- "step": 18500
231
- },
232
- {
233
- "epoch": 1.4,
234
- "learning_rate": 0.00038,
235
- "loss": 2.8457,
236
- "step": 19000
237
- },
238
- {
239
- "epoch": 1.43,
240
- "learning_rate": 0.00039000000000000005,
241
- "loss": 2.8446,
242
- "step": 19500
243
- },
244
- {
245
- "epoch": 1.47,
246
- "learning_rate": 0.0004,
247
- "loss": 2.8356,
248
- "step": 20000
249
- },
250
- {
251
- "epoch": 1.51,
252
- "learning_rate": 0.00041,
253
- "loss": 2.8338,
254
- "step": 20500
255
- },
256
- {
257
- "epoch": 1.54,
258
- "learning_rate": 0.00042,
259
- "loss": 2.8304,
260
- "step": 21000
261
- },
262
- {
263
- "epoch": 1.58,
264
- "learning_rate": 0.00043,
265
- "loss": 2.8269,
266
- "step": 21500
267
- },
268
- {
269
- "epoch": 1.62,
270
- "learning_rate": 0.00044,
271
- "loss": 2.8268,
272
- "step": 22000
273
- },
274
- {
275
- "epoch": 1.65,
276
- "learning_rate": 0.00045000000000000004,
277
- "loss": 2.8224,
278
- "step": 22500
279
- },
280
- {
281
- "epoch": 1.69,
282
- "learning_rate": 0.00046,
283
- "loss": 2.8261,
284
- "step": 23000
285
- },
286
- {
287
- "epoch": 1.73,
288
- "learning_rate": 0.00047,
289
- "loss": 2.8189,
290
- "step": 23500
291
- },
292
- {
293
- "epoch": 1.76,
294
- "learning_rate": 0.00048,
295
- "loss": 2.8113,
296
- "step": 24000
297
- },
298
- {
299
- "epoch": 1.8,
300
- "learning_rate": 0.00049,
301
- "loss": 2.807,
302
- "step": 24500
303
- },
304
- {
305
- "epoch": 1.84,
306
- "learning_rate": 0.0005,
307
- "loss": 2.8181,
308
- "step": 25000
309
- },
310
- {
311
- "epoch": 1.88,
312
- "learning_rate": 0.0004988888888888889,
313
- "loss": 2.7993,
314
- "step": 25500
315
- },
316
- {
317
- "epoch": 1.91,
318
- "learning_rate": 0.0004977777777777778,
319
- "loss": 2.794,
320
- "step": 26000
321
- },
322
- {
323
- "epoch": 1.95,
324
- "learning_rate": 0.0004966666666666666,
325
- "loss": 2.795,
326
- "step": 26500
327
- },
328
- {
329
- "epoch": 1.99,
330
- "learning_rate": 0.0004955555555555556,
331
- "loss": 2.7797,
332
- "step": 27000
333
- },
334
- {
335
- "epoch": 2.02,
336
- "learning_rate": 0.0004944444444444445,
337
- "loss": 2.7595,
338
- "step": 27500
339
- },
340
- {
341
- "epoch": 2.06,
342
- "learning_rate": 0.0004933333333333334,
343
- "loss": 2.7412,
344
- "step": 28000
345
- },
346
- {
347
- "epoch": 2.1,
348
- "learning_rate": 0.0004922222222222222,
349
- "loss": 2.7442,
350
- "step": 28500
351
- },
352
- {
353
- "epoch": 2.13,
354
- "learning_rate": 0.0004911111111111111,
355
- "loss": 2.7406,
356
- "step": 29000
357
- },
358
- {
359
- "epoch": 2.17,
360
- "learning_rate": 0.00049,
361
- "loss": 2.7369,
362
- "step": 29500
363
- },
364
- {
365
- "epoch": 2.21,
366
- "learning_rate": 0.0004888888888888889,
367
- "loss": 2.7356,
368
- "step": 30000
369
- },
370
- {
371
- "epoch": 2.24,
372
- "learning_rate": 0.0004877777777777778,
373
- "loss": 2.7308,
374
- "step": 30500
375
- },
376
- {
377
- "epoch": 2.28,
378
- "learning_rate": 0.0004866666666666667,
379
- "loss": 2.7344,
380
- "step": 31000
381
- },
382
- {
383
- "epoch": 2.32,
384
- "learning_rate": 0.0004855555555555556,
385
- "loss": 2.7292,
386
- "step": 31500
387
- },
388
- {
389
- "epoch": 2.35,
390
- "learning_rate": 0.00048444444444444446,
391
- "loss": 2.724,
392
- "step": 32000
393
- },
394
- {
395
- "epoch": 2.39,
396
- "learning_rate": 0.00048333333333333334,
397
- "loss": 2.7212,
398
- "step": 32500
399
- },
400
- {
401
- "epoch": 2.43,
402
- "learning_rate": 0.0004822222222222222,
403
- "loss": 2.7227,
404
- "step": 33000
405
- },
406
- {
407
- "epoch": 2.46,
408
- "learning_rate": 0.0004811111111111111,
409
- "loss": 2.716,
410
- "step": 33500
411
- },
412
- {
413
- "epoch": 2.5,
414
- "learning_rate": 0.00048,
415
- "loss": 2.7139,
416
- "step": 34000
417
- },
418
- {
419
- "epoch": 2.54,
420
- "learning_rate": 0.0004788888888888889,
421
- "loss": 2.7083,
422
- "step": 34500
423
- },
424
- {
425
- "epoch": 2.57,
426
- "learning_rate": 0.0004777777777777778,
427
- "loss": 2.7062,
428
- "step": 35000
429
- },
430
- {
431
- "epoch": 2.61,
432
- "learning_rate": 0.0004766666666666667,
433
- "loss": 2.7066,
434
- "step": 35500
435
- },
436
- {
437
- "epoch": 2.65,
438
- "learning_rate": 0.00047555555555555556,
439
- "loss": 2.7028,
440
- "step": 36000
441
- },
442
- {
443
- "epoch": 2.68,
444
- "learning_rate": 0.00047444444444444444,
445
- "loss": 2.6972,
446
- "step": 36500
447
- },
448
- {
449
- "epoch": 2.72,
450
- "learning_rate": 0.00047333333333333336,
451
- "loss": 2.6951,
452
- "step": 37000
453
- },
454
- {
455
- "epoch": 2.76,
456
- "learning_rate": 0.00047222222222222224,
457
- "loss": 2.6941,
458
- "step": 37500
459
- },
460
- {
461
- "epoch": 2.79,
462
- "learning_rate": 0.0004711111111111111,
463
- "loss": 2.6943,
464
- "step": 38000
465
- },
466
- {
467
- "epoch": 2.83,
468
- "learning_rate": 0.00047,
469
- "loss": 2.6884,
470
- "step": 38500
471
- },
472
- {
473
- "epoch": 2.87,
474
- "learning_rate": 0.0004688888888888889,
475
- "loss": 2.6858,
476
- "step": 39000
477
- },
478
- {
479
- "epoch": 2.9,
480
- "learning_rate": 0.0004677777777777778,
481
- "loss": 2.685,
482
- "step": 39500
483
- },
484
- {
485
- "epoch": 2.94,
486
- "learning_rate": 0.00046666666666666666,
487
- "loss": 2.6823,
488
- "step": 40000
489
- },
490
- {
491
- "epoch": 2.98,
492
- "learning_rate": 0.0004655555555555556,
493
- "loss": 2.6765,
494
- "step": 40500
495
- },
496
- {
497
- "epoch": 3.01,
498
- "learning_rate": 0.00046444444444444446,
499
- "loss": 2.6626,
500
- "step": 41000
501
- },
502
- {
503
- "epoch": 3.05,
504
- "learning_rate": 0.00046333333333333334,
505
- "loss": 2.6395,
506
- "step": 41500
507
- },
508
- {
509
- "epoch": 3.09,
510
- "learning_rate": 0.0004622222222222222,
511
- "loss": 2.6378,
512
- "step": 42000
513
- },
514
- {
515
- "epoch": 3.13,
516
- "learning_rate": 0.00046111111111111114,
517
- "loss": 2.6419,
518
- "step": 42500
519
- },
520
- {
521
- "epoch": 3.16,
522
- "learning_rate": 0.00046,
523
- "loss": 2.6451,
524
- "step": 43000
525
- },
526
- {
527
- "epoch": 3.2,
528
- "learning_rate": 0.0004588888888888889,
529
- "loss": 2.6396,
530
- "step": 43500
531
- },
532
- {
533
- "epoch": 3.24,
534
- "learning_rate": 0.0004577777777777778,
535
- "loss": 2.6415,
536
- "step": 44000
537
- },
538
- {
539
- "epoch": 3.27,
540
- "learning_rate": 0.0004566666666666667,
541
- "loss": 2.6467,
542
- "step": 44500
543
- },
544
- {
545
- "epoch": 3.31,
546
- "learning_rate": 0.00045555555555555556,
547
- "loss": 2.64,
548
- "step": 45000
549
- },
550
- {
551
- "epoch": 3.35,
552
- "learning_rate": 0.00045444444444444444,
553
- "loss": 2.6391,
554
- "step": 45500
555
- },
556
- {
557
- "epoch": 3.38,
558
- "learning_rate": 0.0004533333333333333,
559
- "loss": 2.6377,
560
- "step": 46000
561
- },
562
- {
563
- "epoch": 3.42,
564
- "learning_rate": 0.00045222222222222224,
565
- "loss": 2.639,
566
- "step": 46500
567
- },
568
- {
569
- "epoch": 3.46,
570
- "learning_rate": 0.0004511111111111111,
571
- "loss": 2.6368,
572
- "step": 47000
573
- },
574
- {
575
- "epoch": 3.49,
576
- "learning_rate": 0.00045000000000000004,
577
- "loss": 2.6374,
578
- "step": 47500
579
- },
580
- {
581
- "epoch": 3.53,
582
- "learning_rate": 0.0004488888888888889,
583
- "loss": 2.6349,
584
- "step": 48000
585
- },
586
- {
587
- "epoch": 3.57,
588
- "learning_rate": 0.0004477777777777778,
589
- "loss": 2.6321,
590
- "step": 48500
591
- },
592
- {
593
- "epoch": 3.6,
594
- "learning_rate": 0.00044666666666666666,
595
- "loss": 2.6286,
596
- "step": 49000
597
- },
598
- {
599
- "epoch": 3.64,
600
- "learning_rate": 0.00044555555555555554,
601
- "loss": 2.6316,
602
- "step": 49500
603
- },
604
- {
605
- "epoch": 3.68,
606
- "learning_rate": 0.0004444444444444444,
607
- "loss": 2.629,
608
- "step": 50000
609
- },
610
- {
611
- "epoch": 3.71,
612
- "learning_rate": 0.00044333333333333334,
613
- "loss": 2.6283,
614
- "step": 50500
615
- },
616
- {
617
- "epoch": 3.75,
618
- "learning_rate": 0.00044222222222222227,
619
- "loss": 2.6266,
620
- "step": 51000
621
- },
622
- {
623
- "epoch": 3.79,
624
- "learning_rate": 0.00044111111111111114,
625
- "loss": 2.6265,
626
- "step": 51500
627
- },
628
- {
629
- "epoch": 3.82,
630
- "learning_rate": 0.00044,
631
- "loss": 2.6228,
632
- "step": 52000
633
- },
634
- {
635
- "epoch": 3.86,
636
- "learning_rate": 0.0004388888888888889,
637
- "loss": 2.6223,
638
- "step": 52500
639
- },
640
- {
641
- "epoch": 3.9,
642
- "learning_rate": 0.00043777777777777776,
643
- "loss": 2.6245,
644
- "step": 53000
645
- },
646
- {
647
- "epoch": 3.93,
648
- "learning_rate": 0.00043666666666666664,
649
- "loss": 2.6204,
650
- "step": 53500
651
- },
652
- {
653
- "epoch": 3.97,
654
- "learning_rate": 0.0004355555555555555,
655
- "loss": 2.6199,
656
- "step": 54000
657
- },
658
- {
659
- "epoch": 4.01,
660
- "learning_rate": 0.0004344444444444445,
661
- "loss": 2.6167,
662
- "step": 54500
663
- },
664
- {
665
- "epoch": 4.04,
666
- "learning_rate": 0.00043333333333333337,
667
- "loss": 2.5796,
668
- "step": 55000
669
- },
670
- {
671
- "epoch": 4.08,
672
- "learning_rate": 0.00043222222222222224,
673
- "loss": 2.5805,
674
- "step": 55500
675
- },
676
- {
677
- "epoch": 4.12,
678
- "learning_rate": 0.0004311111111111111,
679
- "loss": 2.5849,
680
- "step": 56000
681
- },
682
- {
683
- "epoch": 4.15,
684
- "learning_rate": 0.00043,
685
- "loss": 2.5859,
686
- "step": 56500
687
- },
688
- {
689
- "epoch": 4.19,
690
- "learning_rate": 0.00042888888888888886,
691
- "loss": 2.5873,
692
- "step": 57000
693
- },
694
- {
695
- "epoch": 4.23,
696
- "learning_rate": 0.0004277777777777778,
697
- "loss": 2.5868,
698
- "step": 57500
699
- },
700
- {
701
- "epoch": 4.27,
702
- "learning_rate": 0.0004266666666666667,
703
- "loss": 2.5884,
704
- "step": 58000
705
- },
706
- {
707
- "epoch": 4.3,
708
- "learning_rate": 0.0004255555555555556,
709
- "loss": 2.5852,
710
- "step": 58500
711
- },
712
- {
713
- "epoch": 4.34,
714
- "learning_rate": 0.00042444444444444447,
715
- "loss": 2.5867,
716
- "step": 59000
717
- },
718
- {
719
- "epoch": 4.38,
720
- "learning_rate": 0.00042333333333333334,
721
- "loss": 2.5881,
722
- "step": 59500
723
- },
724
- {
725
- "epoch": 4.41,
726
- "learning_rate": 0.0004222222222222222,
727
- "loss": 2.5879,
728
- "step": 60000
729
- },
730
- {
731
- "epoch": 4.45,
732
- "learning_rate": 0.0004211111111111111,
733
- "loss": 2.586,
734
- "step": 60500
735
- },
736
- {
737
- "epoch": 4.49,
738
- "learning_rate": 0.00042,
739
- "loss": 2.5848,
740
- "step": 61000
741
- },
742
- {
743
- "epoch": 4.52,
744
- "learning_rate": 0.0004188888888888889,
745
- "loss": 2.5854,
746
- "step": 61500
747
- },
748
- {
749
- "epoch": 4.56,
750
- "learning_rate": 0.0004177777777777778,
751
- "loss": 2.587,
752
- "step": 62000
753
- },
754
- {
755
- "epoch": 4.6,
756
- "learning_rate": 0.0004166666666666667,
757
- "loss": 2.5863,
758
- "step": 62500
759
- },
760
- {
761
- "epoch": 4.63,
762
- "learning_rate": 0.00041555555555555557,
763
- "loss": 2.5831,
764
- "step": 63000
765
- },
766
- {
767
- "epoch": 4.67,
768
- "learning_rate": 0.00041444444444444444,
769
- "loss": 2.5808,
770
- "step": 63500
771
- },
772
- {
773
- "epoch": 4.71,
774
- "learning_rate": 0.0004133333333333333,
775
- "loss": 2.5834,
776
- "step": 64000
777
- },
778
- {
779
- "epoch": 4.74,
780
- "learning_rate": 0.00041222222222222224,
781
- "loss": 2.5814,
782
- "step": 64500
783
- },
784
- {
785
- "epoch": 4.78,
786
- "learning_rate": 0.0004111111111111111,
787
- "loss": 2.579,
788
- "step": 65000
789
- },
790
- {
791
- "epoch": 4.82,
792
- "learning_rate": 0.00041,
793
- "loss": 2.5842,
794
- "step": 65500
795
- },
796
- {
797
- "epoch": 4.85,
798
- "learning_rate": 0.0004088888888888889,
799
- "loss": 2.5786,
800
- "step": 66000
801
- },
802
- {
803
- "epoch": 4.89,
804
- "learning_rate": 0.0004077777777777778,
805
- "loss": 2.5814,
806
- "step": 66500
807
- },
808
- {
809
- "epoch": 4.93,
810
- "learning_rate": 0.00040666666666666667,
811
- "loss": 2.5786,
812
- "step": 67000
813
- },
814
- {
815
- "epoch": 4.96,
816
- "learning_rate": 0.00040555555555555554,
817
- "loss": 2.58,
818
- "step": 67500
819
- },
820
- {
821
- "epoch": 5.0,
822
- "learning_rate": 0.00040444444444444447,
823
- "loss": 2.5795,
824
- "step": 68000
825
- },
826
- {
827
- "epoch": 5.04,
828
- "learning_rate": 0.00040333333333333334,
829
- "loss": 2.5391,
830
- "step": 68500
831
- },
832
- {
833
- "epoch": 5.07,
834
- "learning_rate": 0.0004022222222222222,
835
- "loss": 2.5398,
836
- "step": 69000
837
- },
838
- {
839
- "epoch": 5.11,
840
- "learning_rate": 0.0004011111111111111,
841
- "loss": 2.5421,
842
- "step": 69500
843
- },
844
- {
845
- "epoch": 5.15,
846
- "learning_rate": 0.0004,
847
- "loss": 2.545,
848
- "step": 70000
849
- },
850
- {
851
- "epoch": 5.18,
852
- "learning_rate": 0.0003988888888888889,
853
- "loss": 2.5493,
854
- "step": 70500
855
- },
856
- {
857
- "epoch": 5.22,
858
- "learning_rate": 0.00039777777777777777,
859
- "loss": 2.5484,
860
- "step": 71000
861
- },
862
- {
863
- "epoch": 5.26,
864
- "learning_rate": 0.0003966666666666667,
865
- "loss": 2.5483,
866
- "step": 71500
867
- },
868
- {
869
- "epoch": 5.29,
870
- "learning_rate": 0.00039555555555555557,
871
- "loss": 2.5511,
872
- "step": 72000
873
- },
874
- {
875
- "epoch": 5.33,
876
- "learning_rate": 0.00039444444444444444,
877
- "loss": 2.5497,
878
- "step": 72500
879
- },
880
- {
881
- "epoch": 5.37,
882
- "learning_rate": 0.0003933333333333333,
883
- "loss": 2.5495,
884
- "step": 73000
885
- },
886
- {
887
- "epoch": 5.4,
888
- "learning_rate": 0.00039222222222222225,
889
- "loss": 2.5508,
890
- "step": 73500
891
- },
892
- {
893
- "epoch": 5.44,
894
- "learning_rate": 0.0003911111111111111,
895
- "loss": 2.5474,
896
- "step": 74000
897
- },
898
- {
899
- "epoch": 5.48,
900
- "learning_rate": 0.00039000000000000005,
901
- "loss": 2.5508,
902
- "step": 74500
903
- },
904
- {
905
- "epoch": 5.52,
906
- "learning_rate": 0.0003888888888888889,
907
- "loss": 2.5512,
908
- "step": 75000
909
- },
910
- {
911
- "epoch": 5.55,
912
- "learning_rate": 0.0003877777777777778,
913
- "loss": 2.5488,
914
- "step": 75500
915
- },
916
- {
917
- "epoch": 5.59,
918
- "learning_rate": 0.00038666666666666667,
919
- "loss": 2.5511,
920
- "step": 76000
921
- },
922
- {
923
- "epoch": 5.63,
924
- "learning_rate": 0.00038555555555555554,
925
- "loss": 2.5507,
926
- "step": 76500
927
- },
928
- {
929
- "epoch": 5.66,
930
- "learning_rate": 0.0003844444444444444,
931
- "loss": 2.5507,
932
- "step": 77000
933
- },
934
- {
935
- "epoch": 5.7,
936
- "learning_rate": 0.00038333333333333334,
937
- "loss": 2.5502,
938
- "step": 77500
939
- },
940
- {
941
- "epoch": 5.74,
942
- "learning_rate": 0.0003822222222222223,
943
- "loss": 2.5464,
944
- "step": 78000
945
- },
946
- {
947
- "epoch": 5.77,
948
- "learning_rate": 0.00038111111111111115,
949
- "loss": 2.5483,
950
- "step": 78500
951
- },
952
- {
953
- "epoch": 5.81,
954
- "learning_rate": 0.00038,
955
- "loss": 2.5463,
956
- "step": 79000
957
- },
958
- {
959
- "epoch": 5.85,
960
- "learning_rate": 0.0003788888888888889,
961
- "loss": 2.5474,
962
- "step": 79500
963
- },
964
- {
965
- "epoch": 5.88,
966
- "learning_rate": 0.00037777777777777777,
967
- "loss": 2.5482,
968
- "step": 80000
969
- },
970
- {
971
- "epoch": 5.92,
972
- "learning_rate": 0.00037666666666666664,
973
- "loss": 2.548,
974
- "step": 80500
975
- },
976
- {
977
- "epoch": 5.96,
978
- "learning_rate": 0.0003755555555555555,
979
- "loss": 2.5435,
980
- "step": 81000
981
- },
982
- {
983
- "epoch": 5.99,
984
- "learning_rate": 0.0003744444444444445,
985
- "loss": 2.5452,
986
- "step": 81500
987
- },
988
- {
989
- "epoch": 6.03,
990
- "learning_rate": 0.0003733333333333334,
991
- "loss": 2.5167,
992
- "step": 82000
993
- },
994
- {
995
- "epoch": 6.07,
996
- "learning_rate": 0.00037222222222222225,
997
- "loss": 2.5106,
998
- "step": 82500
999
- },
1000
- {
1001
- "epoch": 6.1,
1002
- "learning_rate": 0.0003711111111111111,
1003
- "loss": 2.5125,
1004
- "step": 83000
1005
- },
1006
- {
1007
- "epoch": 6.14,
1008
- "learning_rate": 0.00037,
1009
- "loss": 2.5132,
1010
- "step": 83500
1011
- },
1012
- {
1013
- "epoch": 6.18,
1014
- "learning_rate": 0.00036888888888888887,
1015
- "loss": 2.5186,
1016
- "step": 84000
1017
- },
1018
- {
1019
- "epoch": 6.21,
1020
- "learning_rate": 0.00036777777777777774,
1021
- "loss": 2.5191,
1022
- "step": 84500
1023
- },
1024
- {
1025
- "epoch": 6.25,
1026
- "learning_rate": 0.00036666666666666667,
1027
- "loss": 2.5202,
1028
- "step": 85000
1029
- },
1030
- {
1031
- "epoch": 6.29,
1032
- "learning_rate": 0.0003655555555555556,
1033
- "loss": 2.5169,
1034
- "step": 85500
1035
- },
1036
- {
1037
- "epoch": 6.32,
1038
- "learning_rate": 0.00036444444444444447,
1039
- "loss": 2.5171,
1040
- "step": 86000
1041
- },
1042
- {
1043
- "epoch": 6.36,
1044
- "learning_rate": 0.00036333333333333335,
1045
- "loss": 2.5197,
1046
- "step": 86500
1047
- },
1048
- {
1049
- "epoch": 6.4,
1050
- "learning_rate": 0.0003622222222222222,
1051
- "loss": 2.5188,
1052
- "step": 87000
1053
- },
1054
- {
1055
- "epoch": 6.43,
1056
- "learning_rate": 0.0003611111111111111,
1057
- "loss": 2.5198,
1058
- "step": 87500
1059
- },
1060
- {
1061
- "epoch": 6.47,
1062
- "learning_rate": 0.00035999999999999997,
1063
- "loss": 2.5223,
1064
- "step": 88000
1065
- },
1066
- {
1067
- "epoch": 6.51,
1068
- "learning_rate": 0.0003588888888888889,
1069
- "loss": 2.522,
1070
- "step": 88500
1071
- },
1072
- {
1073
- "epoch": 6.54,
1074
- "learning_rate": 0.00035777777777777777,
1075
- "loss": 2.5226,
1076
- "step": 89000
1077
- },
1078
- {
1079
- "epoch": 6.58,
1080
- "learning_rate": 0.0003566666666666667,
1081
- "loss": 2.5218,
1082
- "step": 89500
1083
- },
1084
- {
1085
- "epoch": 6.62,
1086
- "learning_rate": 0.00035555555555555557,
1087
- "loss": 2.5222,
1088
- "step": 90000
1089
- },
1090
- {
1091
- "epoch": 6.65,
1092
- "learning_rate": 0.00035444444444444445,
1093
- "loss": 2.5184,
1094
- "step": 90500
1095
- },
1096
- {
1097
- "epoch": 6.69,
1098
- "learning_rate": 0.0003533333333333333,
1099
- "loss": 2.5206,
1100
- "step": 91000
1101
- },
1102
- {
1103
- "epoch": 6.73,
1104
- "learning_rate": 0.00035222222222222225,
1105
- "loss": 2.5247,
1106
- "step": 91500
1107
- },
1108
- {
1109
- "epoch": 6.77,
1110
- "learning_rate": 0.0003511111111111111,
1111
- "loss": 2.5244,
1112
- "step": 92000
1113
- },
1114
- {
1115
- "epoch": 6.8,
1116
- "learning_rate": 0.00035,
1117
- "loss": 2.5228,
1118
- "step": 92500
1119
- },
1120
- {
1121
- "epoch": 6.84,
1122
- "learning_rate": 0.0003488888888888889,
1123
- "loss": 2.5208,
1124
- "step": 93000
1125
- },
1126
- {
1127
- "epoch": 6.88,
1128
- "learning_rate": 0.0003477777777777778,
1129
- "loss": 2.5208,
1130
- "step": 93500
1131
- },
1132
- {
1133
- "epoch": 6.91,
1134
- "learning_rate": 0.00034666666666666667,
1135
- "loss": 2.5202,
1136
- "step": 94000
1137
- },
1138
- {
1139
- "epoch": 6.95,
1140
- "learning_rate": 0.00034555555555555555,
1141
- "loss": 2.5219,
1142
- "step": 94500
1143
- },
1144
- {
1145
- "epoch": 6.99,
1146
- "learning_rate": 0.0003444444444444445,
1147
- "loss": 2.5204,
1148
- "step": 95000
1149
- },
1150
- {
1151
- "epoch": 7.02,
1152
- "learning_rate": 0.00034333333333333335,
1153
- "loss": 2.4973,
1154
- "step": 95500
1155
- },
1156
- {
1157
- "epoch": 7.06,
1158
- "learning_rate": 0.0003422222222222222,
1159
- "loss": 2.4822,
1160
- "step": 96000
1161
- },
1162
- {
1163
- "epoch": 7.1,
1164
- "learning_rate": 0.0003411111111111111,
1165
- "loss": 2.4842,
1166
- "step": 96500
1167
- },
1168
- {
1169
- "epoch": 7.13,
1170
- "learning_rate": 0.00034,
1171
- "loss": 2.4887,
1172
- "step": 97000
1173
- },
1174
- {
1175
- "epoch": 7.17,
1176
- "learning_rate": 0.0003388888888888889,
1177
- "loss": 2.4893,
1178
- "step": 97500
1179
- },
1180
- {
1181
- "epoch": 7.21,
1182
- "learning_rate": 0.00033777777777777777,
1183
- "loss": 2.4901,
1184
- "step": 98000
1185
- },
1186
- {
1187
- "epoch": 7.24,
1188
- "learning_rate": 0.0003366666666666667,
1189
- "loss": 2.4918,
1190
- "step": 98500
1191
- },
1192
- {
1193
- "epoch": 7.28,
1194
- "learning_rate": 0.0003355555555555556,
1195
- "loss": 2.4945,
1196
- "step": 99000
1197
- },
1198
- {
1199
- "epoch": 7.32,
1200
- "learning_rate": 0.00033444444444444445,
1201
- "loss": 2.4978,
1202
- "step": 99500
1203
- },
1204
- {
1205
- "epoch": 7.35,
1206
- "learning_rate": 0.0003333333333333333,
1207
- "loss": 2.498,
1208
- "step": 100000
1209
- },
1210
- {
1211
- "epoch": 7.39,
1212
- "learning_rate": 0.0003322222222222222,
1213
- "loss": 2.4966,
1214
- "step": 100500
1215
- },
1216
- {
1217
- "epoch": 7.43,
1218
- "learning_rate": 0.0003311111111111111,
1219
- "loss": 2.4983,
1220
- "step": 101000
1221
- },
1222
- {
1223
- "epoch": 7.46,
1224
- "learning_rate": 0.00033,
1225
- "loss": 2.4984,
1226
- "step": 101500
1227
- },
1228
- {
1229
- "epoch": 7.5,
1230
- "learning_rate": 0.0003288888888888889,
1231
- "loss": 2.4961,
1232
- "step": 102000
1233
- },
1234
- {
1235
- "epoch": 7.54,
1236
- "learning_rate": 0.0003277777777777778,
1237
- "loss": 2.4996,
1238
- "step": 102500
1239
- },
1240
- {
1241
- "epoch": 7.57,
1242
- "learning_rate": 0.0003266666666666667,
1243
- "loss": 2.4984,
1244
- "step": 103000
1245
- },
1246
- {
1247
- "epoch": 7.61,
1248
- "learning_rate": 0.00032555555555555555,
1249
- "loss": 2.4992,
1250
- "step": 103500
1251
- },
1252
- {
1253
- "epoch": 7.65,
1254
- "learning_rate": 0.0003244444444444444,
1255
- "loss": 2.4985,
1256
- "step": 104000
1257
- },
1258
- {
1259
- "epoch": 7.68,
1260
- "learning_rate": 0.0003233333333333333,
1261
- "loss": 2.5009,
1262
- "step": 104500
1263
- },
1264
- {
1265
- "epoch": 7.72,
1266
- "learning_rate": 0.0003222222222222222,
1267
- "loss": 2.4971,
1268
- "step": 105000
1269
- },
1270
- {
1271
- "epoch": 7.76,
1272
- "learning_rate": 0.00032111111111111115,
1273
- "loss": 2.4994,
1274
- "step": 105500
1275
- },
1276
- {
1277
- "epoch": 7.79,
1278
- "learning_rate": 0.00032,
1279
- "loss": 2.4963,
1280
- "step": 106000
1281
- },
1282
- {
1283
- "epoch": 7.83,
1284
- "learning_rate": 0.0003188888888888889,
1285
- "loss": 2.4976,
1286
- "step": 106500
1287
- },
1288
- {
1289
- "epoch": 7.87,
1290
- "learning_rate": 0.0003177777777777778,
1291
- "loss": 2.4977,
1292
- "step": 107000
1293
- },
1294
- {
1295
- "epoch": 7.9,
1296
- "learning_rate": 0.00031666666666666665,
1297
- "loss": 2.5017,
1298
- "step": 107500
1299
- },
1300
- {
1301
- "epoch": 7.94,
1302
- "learning_rate": 0.0003155555555555555,
1303
- "loss": 2.4991,
1304
- "step": 108000
1305
- },
1306
- {
1307
- "epoch": 7.98,
1308
- "learning_rate": 0.0003144444444444445,
1309
- "loss": 2.5011,
1310
- "step": 108500
1311
- },
1312
- {
1313
- "epoch": 8.02,
1314
- "learning_rate": 0.0003133333333333334,
1315
- "loss": 2.4842,
1316
- "step": 109000
1317
- },
1318
- {
1319
- "epoch": 8.05,
1320
- "learning_rate": 0.00031222222222222225,
1321
- "loss": 2.4579,
1322
- "step": 109500
1323
- },
1324
- {
1325
- "epoch": 8.09,
1326
- "learning_rate": 0.0003111111111111111,
1327
- "loss": 2.4613,
1328
- "step": 110000
1329
- },
1330
- {
1331
- "epoch": 8.13,
1332
- "learning_rate": 0.00031,
1333
- "loss": 2.4681,
1334
- "step": 110500
1335
- },
1336
- {
1337
- "epoch": 8.16,
1338
- "learning_rate": 0.0003088888888888889,
1339
- "loss": 2.4665,
1340
- "step": 111000
1341
- },
1342
- {
1343
- "epoch": 8.2,
1344
- "learning_rate": 0.00030777777777777775,
1345
- "loss": 2.4709,
1346
- "step": 111500
1347
- },
1348
- {
1349
- "epoch": 8.24,
1350
- "learning_rate": 0.0003066666666666667,
1351
- "loss": 2.47,
1352
- "step": 112000
1353
- },
1354
- {
1355
- "epoch": 8.27,
1356
- "learning_rate": 0.0003055555555555556,
1357
- "loss": 2.4714,
1358
- "step": 112500
1359
- },
1360
- {
1361
- "epoch": 8.31,
1362
- "learning_rate": 0.0003044444444444445,
1363
- "loss": 2.473,
1364
- "step": 113000
1365
- },
1366
- {
1367
- "epoch": 8.35,
1368
- "learning_rate": 0.00030333333333333335,
1369
- "loss": 2.4769,
1370
- "step": 113500
1371
- },
1372
- {
1373
- "epoch": 8.38,
1374
- "learning_rate": 0.0003022222222222222,
1375
- "loss": 2.4753,
1376
- "step": 114000
1377
- },
1378
- {
1379
- "epoch": 8.42,
1380
- "learning_rate": 0.0003011111111111111,
1381
- "loss": 2.4772,
1382
- "step": 114500
1383
- },
1384
- {
1385
- "epoch": 8.46,
1386
- "learning_rate": 0.0003,
1387
- "loss": 2.4789,
1388
- "step": 115000
1389
- },
1390
- {
1391
- "epoch": 8.49,
1392
- "learning_rate": 0.0002988888888888889,
1393
- "loss": 2.4787,
1394
- "step": 115500
1395
- },
1396
- {
1397
- "epoch": 8.53,
1398
- "learning_rate": 0.0002977777777777778,
1399
- "loss": 2.4789,
1400
- "step": 116000
1401
- },
1402
- {
1403
- "epoch": 8.57,
1404
- "learning_rate": 0.0002966666666666667,
1405
- "loss": 2.4771,
1406
- "step": 116500
1407
- },
1408
- {
1409
- "epoch": 8.6,
1410
- "learning_rate": 0.0002955555555555556,
1411
- "loss": 2.48,
1412
- "step": 117000
1413
- },
1414
- {
1415
- "epoch": 8.64,
1416
- "learning_rate": 0.00029444444444444445,
1417
- "loss": 2.477,
1418
- "step": 117500
1419
- },
1420
- {
1421
- "epoch": 8.68,
1422
- "learning_rate": 0.0002933333333333333,
1423
- "loss": 2.4761,
1424
- "step": 118000
1425
- },
1426
- {
1427
- "epoch": 8.71,
1428
- "learning_rate": 0.0002922222222222222,
1429
- "loss": 2.4798,
1430
- "step": 118500
1431
- },
1432
- {
1433
- "epoch": 8.75,
1434
- "learning_rate": 0.00029111111111111113,
1435
- "loss": 2.4795,
1436
- "step": 119000
1437
- },
1438
- {
1439
- "epoch": 8.79,
1440
- "learning_rate": 0.00029,
1441
- "loss": 2.4845,
1442
- "step": 119500
1443
- },
1444
- {
1445
- "epoch": 8.82,
1446
- "learning_rate": 0.0002888888888888889,
1447
- "loss": 2.48,
1448
- "step": 120000
1449
- },
1450
- {
1451
- "epoch": 8.86,
1452
- "learning_rate": 0.0002877777777777778,
1453
- "loss": 2.4808,
1454
- "step": 120500
1455
- },
1456
- {
1457
- "epoch": 8.9,
1458
- "learning_rate": 0.0002866666666666667,
1459
- "loss": 2.4786,
1460
- "step": 121000
1461
- },
1462
- {
1463
- "epoch": 8.93,
1464
- "learning_rate": 0.00028555555555555555,
1465
- "loss": 2.4771,
1466
- "step": 121500
1467
- },
1468
- {
1469
- "epoch": 8.97,
1470
- "learning_rate": 0.0002844444444444444,
1471
- "loss": 2.4775,
1472
- "step": 122000
1473
- },
1474
- {
1475
- "epoch": 9.01,
1476
- "learning_rate": 0.00028333333333333335,
1477
- "loss": 2.4704,
1478
- "step": 122500
1479
- },
1480
- {
1481
- "epoch": 9.04,
1482
- "learning_rate": 0.00028222222222222223,
1483
- "loss": 2.4411,
1484
- "step": 123000
1485
- },
1486
- {
1487
- "epoch": 9.08,
1488
- "learning_rate": 0.0002811111111111111,
1489
- "loss": 2.4426,
1490
- "step": 123500
1491
- },
1492
- {
1493
- "epoch": 9.12,
1494
- "learning_rate": 0.00028000000000000003,
1495
- "loss": 2.449,
1496
- "step": 124000
1497
- },
1498
- {
1499
- "epoch": 9.16,
1500
- "learning_rate": 0.0002788888888888889,
1501
- "loss": 2.4474,
1502
- "step": 124500
1503
- },
1504
- {
1505
- "epoch": 9.19,
1506
- "learning_rate": 0.0002777777777777778,
1507
- "loss": 2.4491,
1508
- "step": 125000
1509
- },
1510
- {
1511
- "epoch": 9.23,
1512
- "learning_rate": 0.00027666666666666665,
1513
- "loss": 2.4512,
1514
- "step": 125500
1515
- },
1516
- {
1517
- "epoch": 9.27,
1518
- "learning_rate": 0.0002755555555555556,
1519
- "loss": 2.4531,
1520
- "step": 126000
1521
- },
1522
- {
1523
- "epoch": 9.3,
1524
- "learning_rate": 0.00027444444444444445,
1525
- "loss": 2.4534,
1526
- "step": 126500
1527
- },
1528
- {
1529
- "epoch": 9.34,
1530
- "learning_rate": 0.00027333333333333333,
1531
- "loss": 2.4576,
1532
- "step": 127000
1533
- },
1534
- {
1535
- "epoch": 9.38,
1536
- "learning_rate": 0.0002722222222222222,
1537
- "loss": 2.4584,
1538
- "step": 127500
1539
- },
1540
- {
1541
- "epoch": 9.41,
1542
- "learning_rate": 0.00027111111111111113,
1543
- "loss": 2.4542,
1544
- "step": 128000
1545
- },
1546
- {
1547
- "epoch": 9.45,
1548
- "learning_rate": 0.00027,
1549
- "loss": 2.4572,
1550
- "step": 128500
1551
- },
1552
- {
1553
- "epoch": 9.49,
1554
- "learning_rate": 0.00026888888888888893,
1555
- "loss": 2.4619,
1556
- "step": 129000
1557
- },
1558
- {
1559
- "epoch": 9.52,
1560
- "learning_rate": 0.0002677777777777778,
1561
- "loss": 2.4578,
1562
- "step": 129500
1563
- },
1564
- {
1565
- "epoch": 9.56,
1566
- "learning_rate": 0.0002666666666666667,
1567
- "loss": 2.4596,
1568
- "step": 130000
1569
- },
1570
- {
1571
- "epoch": 9.6,
1572
- "learning_rate": 0.00026555555555555555,
1573
- "loss": 2.4569,
1574
- "step": 130500
1575
- },
1576
- {
1577
- "epoch": 9.63,
1578
- "learning_rate": 0.00026444444444444443,
1579
- "loss": 2.4611,
1580
- "step": 131000
1581
- },
1582
- {
1583
- "epoch": 9.67,
1584
- "learning_rate": 0.0002633333333333333,
1585
- "loss": 2.4605,
1586
- "step": 131500
1587
- },
1588
- {
1589
- "epoch": 9.71,
1590
- "learning_rate": 0.00026222222222222223,
1591
- "loss": 2.4632,
1592
- "step": 132000
1593
- },
1594
- {
1595
- "epoch": 9.74,
1596
- "learning_rate": 0.00026111111111111116,
1597
- "loss": 2.4603,
1598
- "step": 132500
1599
- },
1600
- {
1601
- "epoch": 9.78,
1602
- "learning_rate": 0.00026000000000000003,
1603
- "loss": 2.4609,
1604
- "step": 133000
1605
- },
1606
- {
1607
- "epoch": 9.82,
1608
- "learning_rate": 0.0002588888888888889,
1609
- "loss": 2.4605,
1610
- "step": 133500
1611
- },
1612
- {
1613
- "epoch": 9.85,
1614
- "learning_rate": 0.0002577777777777778,
1615
- "loss": 2.4623,
1616
- "step": 134000
1617
- },
1618
- {
1619
- "epoch": 9.89,
1620
- "learning_rate": 0.00025666666666666665,
1621
- "loss": 2.4634,
1622
- "step": 134500
1623
- },
1624
- {
1625
- "epoch": 9.93,
1626
- "learning_rate": 0.00025555555555555553,
1627
- "loss": 2.4601,
1628
- "step": 135000
1629
- },
1630
- {
1631
- "epoch": 9.96,
1632
- "learning_rate": 0.0002544444444444444,
1633
- "loss": 2.4608,
1634
- "step": 135500
1635
- },
1636
- {
1637
- "epoch": 10.0,
1638
- "learning_rate": 0.0002533333333333334,
1639
- "loss": 2.4625,
1640
- "step": 136000
1641
- },
1642
- {
1643
- "epoch": 10.04,
1644
- "learning_rate": 0.00025222222222222226,
1645
- "loss": 2.423,
1646
- "step": 136500
1647
- },
1648
- {
1649
- "epoch": 10.07,
1650
- "learning_rate": 0.00025111111111111113,
1651
- "loss": 2.4245,
1652
- "step": 137000
1653
- },
1654
- {
1655
- "epoch": 10.11,
1656
- "learning_rate": 0.00025,
1657
- "loss": 2.4306,
1658
- "step": 137500
1659
- },
1660
- {
1661
- "epoch": 10.15,
1662
- "learning_rate": 0.0002488888888888889,
1663
- "loss": 2.4346,
1664
- "step": 138000
1665
- },
1666
- {
1667
- "epoch": 10.18,
1668
- "learning_rate": 0.0002477777777777778,
1669
- "loss": 2.436,
1670
- "step": 138500
1671
- },
1672
- {
1673
- "epoch": 10.22,
1674
- "learning_rate": 0.0002466666666666667,
1675
- "loss": 2.4336,
1676
- "step": 139000
1677
- },
1678
- {
1679
- "epoch": 10.26,
1680
- "learning_rate": 0.00024555555555555556,
1681
- "loss": 2.4358,
1682
- "step": 139500
1683
- },
1684
- {
1685
- "epoch": 10.29,
1686
- "learning_rate": 0.00024444444444444443,
1687
- "loss": 2.4363,
1688
- "step": 140000
1689
- },
1690
- {
1691
- "epoch": 10.33,
1692
- "learning_rate": 0.00024333333333333336,
1693
- "loss": 2.4386,
1694
- "step": 140500
1695
- },
1696
- {
1697
- "epoch": 10.37,
1698
- "learning_rate": 0.00024222222222222223,
1699
- "loss": 2.4396,
1700
- "step": 141000
1701
- },
1702
- {
1703
- "epoch": 10.41,
1704
- "learning_rate": 0.0002411111111111111,
1705
- "loss": 2.4395,
1706
- "step": 141500
1707
- },
1708
- {
1709
- "epoch": 10.44,
1710
- "learning_rate": 0.00024,
1711
- "loss": 2.4379,
1712
- "step": 142000
1713
- },
1714
- {
1715
- "epoch": 10.48,
1716
- "learning_rate": 0.0002388888888888889,
1717
- "loss": 2.4396,
1718
- "step": 142500
1719
- },
1720
- {
1721
- "epoch": 10.52,
1722
- "learning_rate": 0.00023777777777777778,
1723
- "loss": 2.4418,
1724
- "step": 143000
1725
- },
1726
- {
1727
- "epoch": 10.55,
1728
- "learning_rate": 0.00023666666666666668,
1729
- "loss": 2.4415,
1730
- "step": 143500
1731
- },
1732
- {
1733
- "epoch": 10.59,
1734
- "learning_rate": 0.00023555555555555556,
1735
- "loss": 2.442,
1736
- "step": 144000
1737
- },
1738
- {
1739
- "epoch": 10.63,
1740
- "learning_rate": 0.00023444444444444446,
1741
- "loss": 2.4421,
1742
- "step": 144500
1743
- },
1744
- {
1745
- "epoch": 10.66,
1746
- "learning_rate": 0.00023333333333333333,
1747
- "loss": 2.4426,
1748
- "step": 145000
1749
- },
1750
- {
1751
- "epoch": 10.7,
1752
- "learning_rate": 0.00023222222222222223,
1753
- "loss": 2.443,
1754
- "step": 145500
1755
- },
1756
- {
1757
- "epoch": 10.74,
1758
- "learning_rate": 0.0002311111111111111,
1759
- "loss": 2.4423,
1760
- "step": 146000
1761
- },
1762
- {
1763
- "epoch": 10.77,
1764
- "learning_rate": 0.00023,
1765
- "loss": 2.4454,
1766
- "step": 146500
1767
- },
1768
- {
1769
- "epoch": 10.81,
1770
- "learning_rate": 0.0002288888888888889,
1771
- "loss": 2.4419,
1772
- "step": 147000
1773
- },
1774
- {
1775
- "epoch": 10.85,
1776
- "learning_rate": 0.00022777777777777778,
1777
- "loss": 2.4415,
1778
- "step": 147500
1779
- },
1780
- {
1781
- "epoch": 10.88,
1782
- "learning_rate": 0.00022666666666666666,
1783
- "loss": 2.4434,
1784
- "step": 148000
1785
- },
1786
- {
1787
- "epoch": 10.92,
1788
- "learning_rate": 0.00022555555555555556,
1789
- "loss": 2.4447,
1790
- "step": 148500
1791
- },
1792
- {
1793
- "epoch": 10.96,
1794
- "learning_rate": 0.00022444444444444446,
1795
- "loss": 2.4437,
1796
- "step": 149000
1797
- },
1798
- {
1799
- "epoch": 10.99,
1800
- "learning_rate": 0.00022333333333333333,
1801
- "loss": 2.4475,
1802
- "step": 149500
1803
- },
1804
- {
1805
- "epoch": 11.03,
1806
- "learning_rate": 0.0002222222222222222,
1807
- "loss": 2.4159,
1808
- "step": 150000
1809
- },
1810
- {
1811
- "epoch": 11.07,
1812
- "learning_rate": 0.00022111111111111113,
1813
- "loss": 2.4095,
1814
- "step": 150500
1815
- },
1816
- {
1817
- "epoch": 11.1,
1818
- "learning_rate": 0.00022,
1819
- "loss": 2.4117,
1820
- "step": 151000
1821
- },
1822
- {
1823
- "epoch": 11.14,
1824
- "learning_rate": 0.00021888888888888888,
1825
- "loss": 2.4166,
1826
- "step": 151500
1827
- },
1828
- {
1829
- "epoch": 11.18,
1830
- "learning_rate": 0.00021777777777777776,
1831
- "loss": 2.4161,
1832
- "step": 152000
1833
- },
1834
- {
1835
- "epoch": 11.21,
1836
- "learning_rate": 0.00021666666666666668,
1837
- "loss": 2.416,
1838
- "step": 152500
1839
- },
1840
- {
1841
- "epoch": 11.25,
1842
- "learning_rate": 0.00021555555555555556,
1843
- "loss": 2.4175,
1844
- "step": 153000
1845
- },
1846
- {
1847
- "epoch": 11.29,
1848
- "learning_rate": 0.00021444444444444443,
1849
- "loss": 2.4184,
1850
- "step": 153500
1851
- },
1852
- {
1853
- "epoch": 11.32,
1854
- "learning_rate": 0.00021333333333333336,
1855
- "loss": 2.4206,
1856
- "step": 154000
1857
- },
1858
- {
1859
- "epoch": 11.36,
1860
- "learning_rate": 0.00021222222222222223,
1861
- "loss": 2.4262,
1862
- "step": 154500
1863
- },
1864
- {
1865
- "epoch": 11.4,
1866
- "learning_rate": 0.0002111111111111111,
1867
- "loss": 2.4236,
1868
- "step": 155000
1869
- },
1870
- {
1871
- "epoch": 11.43,
1872
- "learning_rate": 0.00021,
1873
- "loss": 2.4249,
1874
- "step": 155500
1875
- },
1876
- {
1877
- "epoch": 11.47,
1878
- "learning_rate": 0.0002088888888888889,
1879
- "loss": 2.4244,
1880
- "step": 156000
1881
- },
1882
- {
1883
- "epoch": 11.51,
1884
- "learning_rate": 0.00020777777777777778,
1885
- "loss": 2.4235,
1886
- "step": 156500
1887
- },
1888
- {
1889
- "epoch": 11.54,
1890
- "learning_rate": 0.00020666666666666666,
1891
- "loss": 2.4211,
1892
- "step": 157000
1893
- },
1894
- {
1895
- "epoch": 11.58,
1896
- "learning_rate": 0.00020555555555555556,
1897
- "loss": 2.429,
1898
- "step": 157500
1899
- },
1900
- {
1901
- "epoch": 11.62,
1902
- "learning_rate": 0.00020444444444444446,
1903
- "loss": 2.4279,
1904
- "step": 158000
1905
- },
1906
- {
1907
- "epoch": 11.66,
1908
- "learning_rate": 0.00020333333333333333,
1909
- "loss": 2.4269,
1910
- "step": 158500
1911
- },
1912
- {
1913
- "epoch": 11.69,
1914
- "learning_rate": 0.00020222222222222223,
1915
- "loss": 2.4294,
1916
- "step": 159000
1917
- },
1918
- {
1919
- "epoch": 11.73,
1920
- "learning_rate": 0.0002011111111111111,
1921
- "loss": 2.4267,
1922
- "step": 159500
1923
- },
1924
- {
1925
- "epoch": 11.77,
1926
- "learning_rate": 0.0002,
1927
- "loss": 2.4241,
1928
- "step": 160000
1929
- },
1930
- {
1931
- "epoch": 11.8,
1932
- "learning_rate": 0.00019888888888888888,
1933
- "loss": 2.4278,
1934
- "step": 160500
1935
- },
1936
- {
1937
- "epoch": 11.84,
1938
- "learning_rate": 0.00019777777777777778,
1939
- "loss": 2.429,
1940
- "step": 161000
1941
- },
1942
- {
1943
- "epoch": 11.88,
1944
- "learning_rate": 0.00019666666666666666,
1945
- "loss": 2.4277,
1946
- "step": 161500
1947
- },
1948
- {
1949
- "epoch": 11.91,
1950
- "learning_rate": 0.00019555555555555556,
1951
- "loss": 2.4281,
1952
- "step": 162000
1953
- },
1954
- {
1955
- "epoch": 11.95,
1956
- "learning_rate": 0.00019444444444444446,
1957
- "loss": 2.4282,
1958
- "step": 162500
1959
- },
1960
- {
1961
- "epoch": 11.99,
1962
- "learning_rate": 0.00019333333333333333,
1963
- "loss": 2.4299,
1964
- "step": 163000
1965
- },
1966
- {
1967
- "epoch": 12.02,
1968
- "learning_rate": 0.0001922222222222222,
1969
- "loss": 2.4077,
1970
- "step": 163500
1971
- },
1972
- {
1973
- "epoch": 12.06,
1974
- "learning_rate": 0.00019111111111111114,
1975
- "loss": 2.3924,
1976
- "step": 164000
1977
- },
1978
- {
1979
- "epoch": 12.1,
1980
- "learning_rate": 0.00019,
1981
- "loss": 2.3967,
1982
- "step": 164500
1983
- },
1984
- {
1985
- "epoch": 12.13,
1986
- "learning_rate": 0.00018888888888888888,
1987
- "loss": 2.3987,
1988
- "step": 165000
1989
- },
1990
- {
1991
- "epoch": 12.17,
1992
- "learning_rate": 0.00018777777777777776,
1993
- "loss": 2.4014,
1994
- "step": 165500
1995
- },
1996
- {
1997
- "epoch": 12.21,
1998
- "learning_rate": 0.0001866666666666667,
1999
- "loss": 2.4003,
2000
- "step": 166000
2001
- },
2002
- {
2003
- "epoch": 12.24,
2004
- "learning_rate": 0.00018555555555555556,
2005
- "loss": 2.4062,
2006
- "step": 166500
2007
- },
2008
- {
2009
- "epoch": 12.28,
2010
- "learning_rate": 0.00018444444444444443,
2011
- "loss": 2.4038,
2012
- "step": 167000
2013
- },
2014
- {
2015
- "epoch": 12.32,
2016
- "learning_rate": 0.00018333333333333334,
2017
- "loss": 2.406,
2018
- "step": 167500
2019
- },
2020
- {
2021
- "epoch": 12.35,
2022
- "learning_rate": 0.00018222222222222224,
2023
- "loss": 2.4058,
2024
- "step": 168000
2025
- },
2026
- {
2027
- "epoch": 12.39,
2028
- "learning_rate": 0.0001811111111111111,
2029
- "loss": 2.406,
2030
- "step": 168500
2031
- },
2032
- {
2033
- "epoch": 12.43,
2034
- "learning_rate": 0.00017999999999999998,
2035
- "loss": 2.4058,
2036
- "step": 169000
2037
- },
2038
- {
2039
- "epoch": 12.46,
2040
- "learning_rate": 0.00017888888888888889,
2041
- "loss": 2.4086,
2042
- "step": 169500
2043
- },
2044
- {
2045
- "epoch": 12.5,
2046
- "learning_rate": 0.00017777777777777779,
2047
- "loss": 2.41,
2048
- "step": 170000
2049
- },
2050
- {
2051
- "epoch": 12.54,
2052
- "learning_rate": 0.00017666666666666666,
2053
- "loss": 2.4093,
2054
- "step": 170500
2055
- },
2056
- {
2057
- "epoch": 12.57,
2058
- "learning_rate": 0.00017555555555555556,
2059
- "loss": 2.411,
2060
- "step": 171000
2061
- },
2062
- {
2063
- "epoch": 12.61,
2064
- "learning_rate": 0.00017444444444444446,
2065
- "loss": 2.4099,
2066
- "step": 171500
2067
- },
2068
- {
2069
- "epoch": 12.65,
2070
- "learning_rate": 0.00017333333333333334,
2071
- "loss": 2.4121,
2072
- "step": 172000
2073
- },
2074
- {
2075
- "epoch": 12.68,
2076
- "learning_rate": 0.00017222222222222224,
2077
- "loss": 2.4129,
2078
- "step": 172500
2079
- },
2080
- {
2081
- "epoch": 12.72,
2082
- "learning_rate": 0.0001711111111111111,
2083
- "loss": 2.4139,
2084
- "step": 173000
2085
- },
2086
- {
2087
- "epoch": 12.76,
2088
- "learning_rate": 0.00017,
2089
- "loss": 2.4121,
2090
- "step": 173500
2091
- },
2092
- {
2093
- "epoch": 12.8,
2094
- "learning_rate": 0.00016888888888888889,
2095
- "loss": 2.412,
2096
- "step": 174000
2097
- },
2098
- {
2099
- "epoch": 12.83,
2100
- "learning_rate": 0.0001677777777777778,
2101
- "loss": 2.408,
2102
- "step": 174500
2103
- },
2104
- {
2105
- "epoch": 12.87,
2106
- "learning_rate": 0.00016666666666666666,
2107
- "loss": 2.4135,
2108
- "step": 175000
2109
- },
2110
- {
2111
- "epoch": 12.91,
2112
- "learning_rate": 0.00016555555555555556,
2113
- "loss": 2.411,
2114
- "step": 175500
2115
- },
2116
- {
2117
- "epoch": 12.94,
2118
- "learning_rate": 0.00016444444444444446,
2119
- "loss": 2.4108,
2120
- "step": 176000
2121
- },
2122
- {
2123
- "epoch": 12.98,
2124
- "learning_rate": 0.00016333333333333334,
2125
- "loss": 2.4105,
2126
- "step": 176500
2127
- },
2128
- {
2129
- "epoch": 13.02,
2130
- "learning_rate": 0.0001622222222222222,
2131
- "loss": 2.4026,
2132
- "step": 177000
2133
- },
2134
- {
2135
- "epoch": 13.05,
2136
- "learning_rate": 0.0001611111111111111,
2137
- "loss": 2.38,
2138
- "step": 177500
2139
- },
2140
- {
2141
- "epoch": 13.09,
2142
- "learning_rate": 0.00016,
2143
- "loss": 2.3798,
2144
- "step": 178000
2145
- },
2146
- {
2147
- "epoch": 13.13,
2148
- "learning_rate": 0.0001588888888888889,
2149
- "loss": 2.3848,
2150
- "step": 178500
2151
- },
2152
- {
2153
- "epoch": 13.16,
2154
- "learning_rate": 0.00015777777777777776,
2155
- "loss": 2.3865,
2156
- "step": 179000
2157
- },
2158
- {
2159
- "epoch": 13.2,
2160
- "learning_rate": 0.0001566666666666667,
2161
- "loss": 2.3872,
2162
- "step": 179500
2163
- },
2164
- {
2165
- "epoch": 13.24,
2166
- "learning_rate": 0.00015555555555555556,
2167
- "loss": 2.3894,
2168
- "step": 180000
2169
- },
2170
- {
2171
- "epoch": 13.27,
2172
- "learning_rate": 0.00015444444444444444,
2173
- "loss": 2.392,
2174
- "step": 180500
2175
- },
2176
- {
2177
- "epoch": 13.31,
2178
- "learning_rate": 0.00015333333333333334,
2179
- "loss": 2.3872,
2180
- "step": 181000
2181
- },
2182
- {
2183
- "epoch": 13.35,
2184
- "learning_rate": 0.00015222222222222224,
2185
- "loss": 2.3893,
2186
- "step": 181500
2187
- },
2188
- {
2189
- "epoch": 13.38,
2190
- "learning_rate": 0.0001511111111111111,
2191
- "loss": 2.391,
2192
- "step": 182000
2193
- },
2194
- {
2195
- "epoch": 13.42,
2196
- "learning_rate": 0.00015,
2197
- "loss": 2.3948,
2198
- "step": 182500
2199
- },
2200
- {
2201
- "epoch": 13.46,
2202
- "learning_rate": 0.0001488888888888889,
2203
- "loss": 2.3915,
2204
- "step": 183000
2205
- },
2206
- {
2207
- "epoch": 13.49,
2208
- "learning_rate": 0.0001477777777777778,
2209
- "loss": 2.3893,
2210
- "step": 183500
2211
- },
2212
- {
2213
- "epoch": 13.53,
2214
- "learning_rate": 0.00014666666666666666,
2215
- "loss": 2.3917,
2216
- "step": 184000
2217
- },
2218
- {
2219
- "epoch": 13.57,
2220
- "learning_rate": 0.00014555555555555556,
2221
- "loss": 2.395,
2222
- "step": 184500
2223
- },
2224
- {
2225
- "epoch": 13.6,
2226
- "learning_rate": 0.00014444444444444444,
2227
- "loss": 2.3926,
2228
- "step": 185000
2229
- },
2230
- {
2231
- "epoch": 13.64,
2232
- "learning_rate": 0.00014333333333333334,
2233
- "loss": 2.3938,
2234
- "step": 185500
2235
- },
2236
- {
2237
- "epoch": 13.68,
2238
- "learning_rate": 0.0001422222222222222,
2239
- "loss": 2.3934,
2240
- "step": 186000
2241
- },
2242
- {
2243
- "epoch": 13.71,
2244
- "learning_rate": 0.00014111111111111111,
2245
- "loss": 2.3982,
2246
- "step": 186500
2247
- },
2248
- {
2249
- "epoch": 13.75,
2250
- "learning_rate": 0.00014000000000000001,
2251
- "loss": 2.397,
2252
- "step": 187000
2253
- },
2254
- {
2255
- "epoch": 13.79,
2256
- "learning_rate": 0.0001388888888888889,
2257
- "loss": 2.3969,
2258
- "step": 187500
2259
- },
2260
- {
2261
- "epoch": 13.82,
2262
- "learning_rate": 0.0001377777777777778,
2263
- "loss": 2.3969,
2264
- "step": 188000
2265
- },
2266
- {
2267
- "epoch": 13.86,
2268
- "learning_rate": 0.00013666666666666666,
2269
- "loss": 2.3997,
2270
- "step": 188500
2271
- },
2272
- {
2273
- "epoch": 13.9,
2274
- "learning_rate": 0.00013555555555555556,
2275
- "loss": 2.3966,
2276
- "step": 189000
2277
- },
2278
- {
2279
- "epoch": 13.93,
2280
- "learning_rate": 0.00013444444444444447,
2281
- "loss": 2.3976,
2282
- "step": 189500
2283
- },
2284
- {
2285
- "epoch": 13.97,
2286
- "learning_rate": 0.00013333333333333334,
2287
- "loss": 2.3981,
2288
- "step": 190000
2289
- },
2290
- {
2291
- "epoch": 14.01,
2292
- "learning_rate": 0.00013222222222222221,
2293
- "loss": 2.3907,
2294
- "step": 190500
2295
- },
2296
- {
2297
- "epoch": 14.05,
2298
- "learning_rate": 0.00013111111111111111,
2299
- "loss": 2.3665,
2300
- "step": 191000
2301
- },
2302
- {
2303
- "epoch": 14.08,
2304
- "learning_rate": 0.00013000000000000002,
2305
- "loss": 2.3677,
2306
- "step": 191500
2307
- },
2308
- {
2309
- "epoch": 14.12,
2310
- "learning_rate": 0.0001288888888888889,
2311
- "loss": 2.3667,
2312
- "step": 192000
2313
- },
2314
- {
2315
- "epoch": 14.16,
2316
- "learning_rate": 0.00012777777777777776,
2317
- "loss": 2.3746,
2318
- "step": 192500
2319
- },
2320
- {
2321
- "epoch": 14.19,
2322
- "learning_rate": 0.0001266666666666667,
2323
- "loss": 2.3693,
2324
- "step": 193000
2325
- },
2326
- {
2327
- "epoch": 14.23,
2328
- "learning_rate": 0.00012555555555555557,
2329
- "loss": 2.3721,
2330
- "step": 193500
2331
- },
2332
- {
2333
- "epoch": 14.27,
2334
- "learning_rate": 0.00012444444444444444,
2335
- "loss": 2.3729,
2336
- "step": 194000
2337
- },
2338
- {
2339
- "epoch": 14.3,
2340
- "learning_rate": 0.00012333333333333334,
2341
- "loss": 2.3782,
2342
- "step": 194500
2343
- },
2344
- {
2345
- "epoch": 14.34,
2346
- "learning_rate": 0.00012222222222222221,
2347
- "loss": 2.3751,
2348
- "step": 195000
2349
- },
2350
- {
2351
- "epoch": 14.38,
2352
- "learning_rate": 0.00012111111111111112,
2353
- "loss": 2.3777,
2354
- "step": 195500
2355
- },
2356
- {
2357
- "epoch": 14.41,
2358
- "learning_rate": 0.00012,
2359
- "loss": 2.3767,
2360
- "step": 196000
2361
- },
2362
- {
2363
- "epoch": 14.45,
2364
- "learning_rate": 0.00011888888888888889,
2365
- "loss": 2.3761,
2366
- "step": 196500
2367
- },
2368
- {
2369
- "epoch": 14.49,
2370
- "learning_rate": 0.00011777777777777778,
2371
- "loss": 2.3787,
2372
- "step": 197000
2373
- },
2374
- {
2375
- "epoch": 14.52,
2376
- "learning_rate": 0.00011666666666666667,
2377
- "loss": 2.3796,
2378
- "step": 197500
2379
- },
2380
- {
2381
- "epoch": 14.56,
2382
- "learning_rate": 0.00011555555555555555,
2383
- "loss": 2.3796,
2384
- "step": 198000
2385
- },
2386
- {
2387
- "epoch": 14.6,
2388
- "learning_rate": 0.00011444444444444445,
2389
- "loss": 2.3791,
2390
- "step": 198500
2391
- },
2392
- {
2393
- "epoch": 14.63,
2394
- "learning_rate": 0.00011333333333333333,
2395
- "loss": 2.3789,
2396
- "step": 199000
2397
- },
2398
- {
2399
- "epoch": 14.67,
2400
- "learning_rate": 0.00011222222222222223,
2401
- "loss": 2.3818,
2402
- "step": 199500
2403
- },
2404
- {
2405
- "epoch": 14.71,
2406
- "learning_rate": 0.0001111111111111111,
2407
- "loss": 2.3813,
2408
- "step": 200000
2409
- },
2410
- {
2411
- "epoch": 14.74,
2412
- "learning_rate": 0.00011,
2413
- "loss": 2.3815,
2414
- "step": 200500
2415
- },
2416
- {
2417
- "epoch": 14.78,
2418
- "learning_rate": 0.00010888888888888888,
2419
- "loss": 2.3779,
2420
- "step": 201000
2421
- },
2422
- {
2423
- "epoch": 14.82,
2424
- "learning_rate": 0.00010777777777777778,
2425
- "loss": 2.3818,
2426
- "step": 201500
2427
- },
2428
- {
2429
- "epoch": 14.85,
2430
- "learning_rate": 0.00010666666666666668,
2431
- "loss": 2.3796,
2432
- "step": 202000
2433
- },
2434
- {
2435
- "epoch": 14.89,
2436
- "learning_rate": 0.00010555555555555555,
2437
- "loss": 2.3822,
2438
- "step": 202500
2439
- },
2440
- {
2441
- "epoch": 14.93,
2442
- "learning_rate": 0.00010444444444444445,
2443
- "loss": 2.3829,
2444
- "step": 203000
2445
- },
2446
- {
2447
- "epoch": 14.96,
2448
- "learning_rate": 0.00010333333333333333,
2449
- "loss": 2.3798,
2450
- "step": 203500
2451
- },
2452
- {
2453
- "epoch": 15.0,
2454
- "learning_rate": 0.00010222222222222223,
2455
- "loss": 2.3833,
2456
- "step": 204000
2457
- },
2458
- {
2459
- "epoch": 15.04,
2460
- "learning_rate": 0.00010111111111111112,
2461
- "loss": 2.3523,
2462
- "step": 204500
2463
- },
2464
- {
2465
- "epoch": 15.07,
2466
- "learning_rate": 0.0001,
2467
- "loss": 2.3534,
2468
- "step": 205000
2469
- },
2470
- {
2471
- "epoch": 15.11,
2472
- "learning_rate": 9.888888888888889e-05,
2473
- "loss": 2.3552,
2474
- "step": 205500
2475
- },
2476
- {
2477
- "epoch": 15.15,
2478
- "learning_rate": 9.777777777777778e-05,
2479
- "loss": 2.3573,
2480
- "step": 206000
2481
- },
2482
- {
2483
- "epoch": 15.18,
2484
- "learning_rate": 9.666666666666667e-05,
2485
- "loss": 2.3597,
2486
- "step": 206500
2487
- },
2488
- {
2489
- "epoch": 15.22,
2490
- "learning_rate": 9.555555555555557e-05,
2491
- "loss": 2.3594,
2492
- "step": 207000
2493
- },
2494
- {
2495
- "epoch": 15.26,
2496
- "learning_rate": 9.444444444444444e-05,
2497
- "loss": 2.3626,
2498
- "step": 207500
2499
- },
2500
- {
2501
- "epoch": 15.3,
2502
- "learning_rate": 9.333333333333334e-05,
2503
- "loss": 2.3613,
2504
- "step": 208000
2505
- },
2506
- {
2507
- "epoch": 15.33,
2508
- "learning_rate": 9.222222222222222e-05,
2509
- "loss": 2.3594,
2510
- "step": 208500
2511
- },
2512
- {
2513
- "epoch": 15.37,
2514
- "learning_rate": 9.111111111111112e-05,
2515
- "loss": 2.3609,
2516
- "step": 209000
2517
- },
2518
- {
2519
- "epoch": 15.41,
2520
- "learning_rate": 8.999999999999999e-05,
2521
- "loss": 2.3618,
2522
- "step": 209500
2523
- },
2524
- {
2525
- "epoch": 15.44,
2526
- "learning_rate": 8.888888888888889e-05,
2527
- "loss": 2.3643,
2528
- "step": 210000
2529
- },
2530
- {
2531
- "epoch": 15.48,
2532
- "learning_rate": 8.777777777777778e-05,
2533
- "loss": 2.3646,
2534
- "step": 210500
2535
- },
2536
- {
2537
- "epoch": 15.52,
2538
- "learning_rate": 8.666666666666667e-05,
2539
- "loss": 2.3599,
2540
- "step": 211000
2541
- },
2542
- {
2543
- "epoch": 15.55,
2544
- "learning_rate": 8.555555555555556e-05,
2545
- "loss": 2.3623,
2546
- "step": 211500
2547
- },
2548
- {
2549
- "epoch": 15.59,
2550
- "learning_rate": 8.444444444444444e-05,
2551
- "loss": 2.364,
2552
- "step": 212000
2553
- },
2554
- {
2555
- "epoch": 15.63,
2556
- "learning_rate": 8.333333333333333e-05,
2557
- "loss": 2.3654,
2558
- "step": 212500
2559
- },
2560
- {
2561
- "epoch": 15.66,
2562
- "learning_rate": 8.222222222222223e-05,
2563
- "loss": 2.3626,
2564
- "step": 213000
2565
- },
2566
- {
2567
- "epoch": 15.7,
2568
- "learning_rate": 8.11111111111111e-05,
2569
- "loss": 2.3663,
2570
- "step": 213500
2571
- },
2572
- {
2573
- "epoch": 15.74,
2574
- "learning_rate": 8e-05,
2575
- "loss": 2.3669,
2576
- "step": 214000
2577
- },
2578
- {
2579
- "epoch": 15.77,
2580
- "learning_rate": 7.888888888888888e-05,
2581
- "loss": 2.3662,
2582
- "step": 214500
2583
- },
2584
- {
2585
- "epoch": 15.81,
2586
- "learning_rate": 7.777777777777778e-05,
2587
- "loss": 2.3641,
2588
- "step": 215000
2589
- },
2590
- {
2591
- "epoch": 15.85,
2592
- "learning_rate": 7.666666666666667e-05,
2593
- "loss": 2.365,
2594
- "step": 215500
2595
- },
2596
- {
2597
- "epoch": 15.88,
2598
- "learning_rate": 7.555555555555556e-05,
2599
- "loss": 2.3685,
2600
- "step": 216000
2601
- },
2602
- {
2603
- "epoch": 15.92,
2604
- "learning_rate": 7.444444444444444e-05,
2605
- "loss": 2.3667,
2606
- "step": 216500
2607
- },
2608
- {
2609
- "epoch": 15.96,
2610
- "learning_rate": 7.333333333333333e-05,
2611
- "loss": 2.3641,
2612
- "step": 217000
2613
- },
2614
- {
2615
- "epoch": 15.99,
2616
- "learning_rate": 7.222222222222222e-05,
2617
- "loss": 2.3651,
2618
- "step": 217500
2619
- },
2620
- {
2621
- "epoch": 16.03,
2622
- "learning_rate": 7.11111111111111e-05,
2623
- "loss": 2.3486,
2624
- "step": 218000
2625
- },
2626
- {
2627
- "epoch": 16.07,
2628
- "learning_rate": 7.000000000000001e-05,
2629
- "loss": 2.3431,
2630
- "step": 218500
2631
- },
2632
- {
2633
- "epoch": 16.1,
2634
- "learning_rate": 6.88888888888889e-05,
2635
- "loss": 2.3439,
2636
- "step": 219000
2637
- },
2638
- {
2639
- "epoch": 16.14,
2640
- "learning_rate": 6.777777777777778e-05,
2641
- "loss": 2.3438,
2642
- "step": 219500
2643
- },
2644
- {
2645
- "epoch": 16.18,
2646
- "learning_rate": 6.666666666666667e-05,
2647
- "loss": 2.3426,
2648
- "step": 220000
2649
- },
2650
- {
2651
- "epoch": 16.21,
2652
- "learning_rate": 6.555555555555556e-05,
2653
- "loss": 2.3463,
2654
- "step": 220500
2655
- },
2656
- {
2657
- "epoch": 16.25,
2658
- "learning_rate": 6.444444444444444e-05,
2659
- "loss": 2.3486,
2660
- "step": 221000
2661
- },
2662
- {
2663
- "epoch": 16.29,
2664
- "learning_rate": 6.333333333333335e-05,
2665
- "loss": 2.3476,
2666
- "step": 221500
2667
- },
2668
- {
2669
- "epoch": 16.32,
2670
- "learning_rate": 6.222222222222222e-05,
2671
- "loss": 2.3469,
2672
- "step": 222000
2673
- },
2674
- {
2675
- "epoch": 16.36,
2676
- "learning_rate": 6.111111111111111e-05,
2677
- "loss": 2.3519,
2678
- "step": 222500
2679
- },
2680
- {
2681
- "epoch": 16.4,
2682
- "learning_rate": 6e-05,
2683
- "loss": 2.3485,
2684
- "step": 223000
2685
- },
2686
- {
2687
- "epoch": 16.44,
2688
- "learning_rate": 5.888888888888889e-05,
2689
- "loss": 2.3464,
2690
- "step": 223500
2691
- },
2692
- {
2693
- "epoch": 16.47,
2694
- "learning_rate": 5.7777777777777776e-05,
2695
- "loss": 2.3524,
2696
- "step": 224000
2697
- },
2698
- {
2699
- "epoch": 16.51,
2700
- "learning_rate": 5.6666666666666664e-05,
2701
- "loss": 2.3482,
2702
- "step": 224500
2703
- },
2704
- {
2705
- "epoch": 16.55,
2706
- "learning_rate": 5.555555555555555e-05,
2707
- "loss": 2.3485,
2708
- "step": 225000
2709
- },
2710
- {
2711
- "epoch": 16.58,
2712
- "learning_rate": 5.444444444444444e-05,
2713
- "loss": 2.3476,
2714
- "step": 225500
2715
- },
2716
- {
2717
- "epoch": 16.62,
2718
- "learning_rate": 5.333333333333334e-05,
2719
- "loss": 2.3452,
2720
- "step": 226000
2721
- },
2722
- {
2723
- "epoch": 16.66,
2724
- "learning_rate": 5.222222222222223e-05,
2725
- "loss": 2.3491,
2726
- "step": 226500
2727
- },
2728
- {
2729
- "epoch": 16.69,
2730
- "learning_rate": 5.1111111111111115e-05,
2731
- "loss": 2.3504,
2732
- "step": 227000
2733
- },
2734
- {
2735
- "epoch": 16.73,
2736
- "learning_rate": 5e-05,
2737
- "loss": 2.3517,
2738
- "step": 227500
2739
- },
2740
- {
2741
- "epoch": 16.77,
2742
- "learning_rate": 4.888888888888889e-05,
2743
- "loss": 2.3508,
2744
- "step": 228000
2745
- },
2746
- {
2747
- "epoch": 16.8,
2748
- "learning_rate": 4.7777777777777784e-05,
2749
- "loss": 2.3522,
2750
- "step": 228500
2751
- },
2752
- {
2753
- "epoch": 16.84,
2754
- "learning_rate": 4.666666666666667e-05,
2755
- "loss": 2.3494,
2756
- "step": 229000
2757
- },
2758
- {
2759
- "epoch": 16.88,
2760
- "learning_rate": 4.555555555555556e-05,
2761
- "loss": 2.3508,
2762
- "step": 229500
2763
- },
2764
- {
2765
- "epoch": 16.91,
2766
- "learning_rate": 4.4444444444444447e-05,
2767
- "loss": 2.349,
2768
- "step": 230000
2769
- },
2770
- {
2771
- "epoch": 16.95,
2772
- "learning_rate": 4.3333333333333334e-05,
2773
- "loss": 2.3499,
2774
- "step": 230500
2775
- },
2776
- {
2777
- "epoch": 16.99,
2778
- "learning_rate": 4.222222222222222e-05,
2779
- "loss": 2.3474,
2780
- "step": 231000
2781
- },
2782
- {
2783
- "epoch": 17.02,
2784
- "learning_rate": 4.1111111111111116e-05,
2785
- "loss": 2.3398,
2786
- "step": 231500
2787
- },
2788
- {
2789
- "epoch": 17.06,
2790
- "learning_rate": 4e-05,
2791
- "loss": 2.3326,
2792
- "step": 232000
2793
- },
2794
- {
2795
- "epoch": 17.1,
2796
- "learning_rate": 3.888888888888889e-05,
2797
- "loss": 2.332,
2798
- "step": 232500
2799
- },
2800
- {
2801
- "epoch": 17.13,
2802
- "learning_rate": 3.777777777777778e-05,
2803
- "loss": 2.3296,
2804
- "step": 233000
2805
- },
2806
- {
2807
- "epoch": 17.17,
2808
- "learning_rate": 3.6666666666666666e-05,
2809
- "loss": 2.3336,
2810
- "step": 233500
2811
- },
2812
- {
2813
- "epoch": 17.21,
2814
- "learning_rate": 3.555555555555555e-05,
2815
- "loss": 2.3328,
2816
- "step": 234000
2817
- },
2818
- {
2819
- "epoch": 17.24,
2820
- "learning_rate": 3.444444444444445e-05,
2821
- "loss": 2.3386,
2822
- "step": 234500
2823
- },
2824
- {
2825
- "epoch": 17.28,
2826
- "learning_rate": 3.3333333333333335e-05,
2827
- "loss": 2.3343,
2828
- "step": 235000
2829
- },
2830
- {
2831
- "epoch": 17.32,
2832
- "learning_rate": 3.222222222222222e-05,
2833
- "loss": 2.3354,
2834
- "step": 235500
2835
- },
2836
- {
2837
- "epoch": 17.35,
2838
- "learning_rate": 3.111111111111111e-05,
2839
- "loss": 2.3331,
2840
- "step": 236000
2841
- },
2842
- {
2843
- "epoch": 17.39,
2844
- "learning_rate": 3e-05,
2845
- "loss": 2.3351,
2846
- "step": 236500
2847
- },
2848
- {
2849
- "epoch": 17.43,
2850
- "learning_rate": 2.8888888888888888e-05,
2851
- "loss": 2.3319,
2852
- "step": 237000
2853
- },
2854
- {
2855
- "epoch": 17.46,
2856
- "learning_rate": 2.7777777777777776e-05,
2857
- "loss": 2.3351,
2858
- "step": 237500
2859
- },
2860
- {
2861
- "epoch": 17.5,
2862
- "learning_rate": 2.666666666666667e-05,
2863
- "loss": 2.3342,
2864
- "step": 238000
2865
- },
2866
- {
2867
- "epoch": 17.54,
2868
- "learning_rate": 2.5555555555555557e-05,
2869
- "loss": 2.3358,
2870
- "step": 238500
2871
- },
2872
- {
2873
- "epoch": 17.57,
2874
- "learning_rate": 2.4444444444444445e-05,
2875
- "loss": 2.3334,
2876
- "step": 239000
2877
- },
2878
- {
2879
- "epoch": 17.61,
2880
- "learning_rate": 2.3333333333333336e-05,
2881
- "loss": 2.3349,
2882
- "step": 239500
2883
- },
2884
- {
2885
- "epoch": 17.65,
2886
- "learning_rate": 2.2222222222222223e-05,
2887
- "loss": 2.3337,
2888
- "step": 240000
2889
- },
2890
- {
2891
- "epoch": 17.69,
2892
- "learning_rate": 2.111111111111111e-05,
2893
- "loss": 2.3357,
2894
- "step": 240500
2895
- },
2896
- {
2897
- "epoch": 17.72,
2898
- "learning_rate": 2e-05,
2899
- "loss": 2.3369,
2900
- "step": 241000
2901
- },
2902
- {
2903
- "epoch": 17.76,
2904
- "learning_rate": 1.888888888888889e-05,
2905
- "loss": 2.3365,
2906
- "step": 241500
2907
- },
2908
- {
2909
- "epoch": 17.8,
2910
- "learning_rate": 1.7777777777777777e-05,
2911
- "loss": 2.3347,
2912
- "step": 242000
2913
- },
2914
- {
2915
- "epoch": 17.83,
2916
- "learning_rate": 1.6666666666666667e-05,
2917
- "loss": 2.336,
2918
- "step": 242500
2919
- },
2920
- {
2921
- "epoch": 17.87,
2922
- "learning_rate": 1.5555555555555555e-05,
2923
- "loss": 2.334,
2924
- "step": 243000
2925
- },
2926
- {
2927
- "epoch": 17.91,
2928
- "learning_rate": 1.4444444444444444e-05,
2929
- "loss": 2.3343,
2930
- "step": 243500
2931
- },
2932
- {
2933
- "epoch": 17.94,
2934
- "learning_rate": 1.3333333333333335e-05,
2935
- "loss": 2.3332,
2936
- "step": 244000
2937
- },
2938
- {
2939
- "epoch": 17.98,
2940
- "learning_rate": 1.2222222222222222e-05,
2941
- "loss": 2.3376,
2942
- "step": 244500
2943
- },
2944
- {
2945
- "epoch": 18.02,
2946
- "learning_rate": 1.1111111111111112e-05,
2947
- "loss": 2.3319,
2948
- "step": 245000
2949
- },
2950
- {
2951
- "epoch": 18.05,
2952
- "learning_rate": 1e-05,
2953
- "loss": 2.3224,
2954
- "step": 245500
2955
- },
2956
- {
2957
- "epoch": 18.09,
2958
- "learning_rate": 8.888888888888888e-06,
2959
- "loss": 2.3249,
2960
- "step": 246000
2961
- },
2962
- {
2963
- "epoch": 18.13,
2964
- "learning_rate": 7.777777777777777e-06,
2965
- "loss": 2.3241,
2966
- "step": 246500
2967
- },
2968
- {
2969
- "epoch": 18.16,
2970
- "learning_rate": 6.6666666666666675e-06,
2971
- "loss": 2.3222,
2972
- "step": 247000
2973
- },
2974
- {
2975
- "epoch": 18.2,
2976
- "learning_rate": 5.555555555555556e-06,
2977
- "loss": 2.3221,
2978
- "step": 247500
2979
- },
2980
- {
2981
- "epoch": 18.24,
2982
- "learning_rate": 4.444444444444444e-06,
2983
- "loss": 2.324,
2984
- "step": 248000
2985
- },
2986
- {
2987
- "epoch": 18.27,
2988
- "learning_rate": 3.3333333333333337e-06,
2989
- "loss": 2.3234,
2990
- "step": 248500
2991
- },
2992
- {
2993
- "epoch": 18.31,
2994
- "learning_rate": 2.222222222222222e-06,
2995
- "loss": 2.3225,
2996
- "step": 249000
2997
- },
2998
- {
2999
- "epoch": 18.35,
3000
- "learning_rate": 1.111111111111111e-06,
3001
- "loss": 2.3234,
3002
- "step": 249500
3003
- },
3004
- {
3005
- "epoch": 18.38,
3006
- "learning_rate": 0.0,
3007
- "loss": 2.3237,
3008
- "step": 250000
3009
- },
3010
- {
3011
- "epoch": 18.38,
3012
- "step": 250000,
3013
- "total_flos": 3.482052847164457e+19,
3014
- "train_loss": 0.7083540009765625,
3015
- "train_runtime": 869248.5122,
3016
- "train_samples_per_second": 73.627,
3017
- "train_steps_per_second": 0.288
3018
- }
3019
- ],
3020
- "max_steps": 250000,
3021
- "num_train_epochs": 19,
3022
- "total_flos": 3.482052847164457e+19,
3023
- "trial_name": null,
3024
- "trial_params": null
3025
- }