mmomm25 commited on
Commit
27276f6
1 Parent(s): 26e700c

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +0 -2367
README.md CHANGED
@@ -1,2367 +0,0 @@
1
- {
2
- "best_metric": 0.058978818356990814,
3
- "best_model_checkpoint": "swin-tiny-patch4-window7-224-crack-detector30epochs2\\checkpoint-2047",
4
- "epoch": 29.846153846153847,
5
- "eval_steps": 500,
6
- "global_step": 2910,
7
- "is_hyper_param_search": false,
8
- "is_local_process_zero": true,
9
- "is_world_process_zero": true,
10
- "log_history": [
11
- {
12
- "epoch": 0.1,
13
- "learning_rate": 1.7182130584192443e-06,
14
- "loss": 0.4366,
15
- "step": 10
16
- },
17
- {
18
- "epoch": 0.21,
19
- "learning_rate": 3.4364261168384886e-06,
20
- "loss": 0.4056,
21
- "step": 20
22
- },
23
- {
24
- "epoch": 0.31,
25
- "learning_rate": 5.154639175257732e-06,
26
- "loss": 0.3739,
27
- "step": 30
28
- },
29
- {
30
- "epoch": 0.41,
31
- "learning_rate": 6.872852233676977e-06,
32
- "loss": 0.3896,
33
- "step": 40
34
- },
35
- {
36
- "epoch": 0.51,
37
- "learning_rate": 8.591065292096221e-06,
38
- "loss": 0.3859,
39
- "step": 50
40
- },
41
- {
42
- "epoch": 0.62,
43
- "learning_rate": 1.0309278350515464e-05,
44
- "loss": 0.3972,
45
- "step": 60
46
- },
47
- {
48
- "epoch": 0.72,
49
- "learning_rate": 1.2027491408934708e-05,
50
- "loss": 0.399,
51
- "step": 70
52
- },
53
- {
54
- "epoch": 0.82,
55
- "learning_rate": 1.3745704467353954e-05,
56
- "loss": 0.3618,
57
- "step": 80
58
- },
59
- {
60
- "epoch": 0.92,
61
- "learning_rate": 1.5463917525773197e-05,
62
- "loss": 0.3827,
63
- "step": 90
64
- },
65
- {
66
- "epoch": 0.99,
67
- "eval_accuracy": {
68
- "accuracy": 0.8887820512820512
69
- },
70
- "eval_f1": {
71
- "f1": 0.8844282256354783
72
- },
73
- "eval_loss": 0.30944663286209106,
74
- "eval_precision": {
75
- "precision": 0.8908184085097209
76
- },
77
- "eval_recall": {
78
- "recall": 0.8824026598746582
79
- },
80
- "eval_runtime": 17.9436,
81
- "eval_samples_per_second": 173.878,
82
- "eval_steps_per_second": 5.462,
83
- "step": 97
84
- },
85
- {
86
- "epoch": 1.03,
87
- "learning_rate": 1.7182130584192442e-05,
88
- "loss": 0.3571,
89
- "step": 100
90
- },
91
- {
92
- "epoch": 1.13,
93
- "learning_rate": 1.8900343642611683e-05,
94
- "loss": 0.313,
95
- "step": 110
96
- },
97
- {
98
- "epoch": 1.23,
99
- "learning_rate": 2.0618556701030927e-05,
100
- "loss": 0.306,
101
- "step": 120
102
- },
103
- {
104
- "epoch": 1.33,
105
- "learning_rate": 2.2336769759450175e-05,
106
- "loss": 0.3171,
107
- "step": 130
108
- },
109
- {
110
- "epoch": 1.44,
111
- "learning_rate": 2.4054982817869417e-05,
112
- "loss": 0.3036,
113
- "step": 140
114
- },
115
- {
116
- "epoch": 1.54,
117
- "learning_rate": 2.5773195876288658e-05,
118
- "loss": 0.2944,
119
- "step": 150
120
- },
121
- {
122
- "epoch": 1.64,
123
- "learning_rate": 2.749140893470791e-05,
124
- "loss": 0.2633,
125
- "step": 160
126
- },
127
- {
128
- "epoch": 1.74,
129
- "learning_rate": 2.920962199312715e-05,
130
- "loss": 0.2663,
131
- "step": 170
132
- },
133
- {
134
- "epoch": 1.85,
135
- "learning_rate": 3.0927835051546395e-05,
136
- "loss": 0.2544,
137
- "step": 180
138
- },
139
- {
140
- "epoch": 1.95,
141
- "learning_rate": 3.2646048109965636e-05,
142
- "loss": 0.3167,
143
- "step": 190
144
- },
145
- {
146
- "epoch": 2.0,
147
- "eval_accuracy": {
148
- "accuracy": 0.8907051282051283
149
- },
150
- "eval_f1": {
151
- "f1": 0.8909502908358744
152
- },
153
- "eval_loss": 0.30256137251853943,
154
- "eval_precision": {
155
- "precision": 0.8886505787148983
156
- },
157
- "eval_recall": {
158
- "recall": 0.8960294776160731
159
- },
160
- "eval_runtime": 18.1552,
161
- "eval_samples_per_second": 171.852,
162
- "eval_steps_per_second": 5.398,
163
- "step": 195
164
- },
165
- {
166
- "epoch": 2.05,
167
- "learning_rate": 3.4364261168384884e-05,
168
- "loss": 0.3464,
169
- "step": 200
170
- },
171
- {
172
- "epoch": 2.15,
173
- "learning_rate": 3.6082474226804125e-05,
174
- "loss": 0.3633,
175
- "step": 210
176
- },
177
- {
178
- "epoch": 2.26,
179
- "learning_rate": 3.7800687285223366e-05,
180
- "loss": 0.3191,
181
- "step": 220
182
- },
183
- {
184
- "epoch": 2.36,
185
- "learning_rate": 3.9518900343642614e-05,
186
- "loss": 0.2862,
187
- "step": 230
188
- },
189
- {
190
- "epoch": 2.46,
191
- "learning_rate": 4.1237113402061855e-05,
192
- "loss": 0.3021,
193
- "step": 240
194
- },
195
- {
196
- "epoch": 2.56,
197
- "learning_rate": 4.2955326460481096e-05,
198
- "loss": 0.3424,
199
- "step": 250
200
- },
201
- {
202
- "epoch": 2.67,
203
- "learning_rate": 4.467353951890035e-05,
204
- "loss": 0.3484,
205
- "step": 260
206
- },
207
- {
208
- "epoch": 2.77,
209
- "learning_rate": 4.639175257731959e-05,
210
- "loss": 0.3277,
211
- "step": 270
212
- },
213
- {
214
- "epoch": 2.87,
215
- "learning_rate": 4.810996563573883e-05,
216
- "loss": 0.2917,
217
- "step": 280
218
- },
219
- {
220
- "epoch": 2.97,
221
- "learning_rate": 4.982817869415808e-05,
222
- "loss": 0.3359,
223
- "step": 290
224
- },
225
- {
226
- "epoch": 2.99,
227
- "eval_accuracy": {
228
- "accuracy": 0.9333333333333333
229
- },
230
- "eval_f1": {
231
- "f1": 0.9307736319130557
232
- },
233
- "eval_loss": 0.1916925311088562,
234
- "eval_precision": {
235
- "precision": 0.9329256773019474
236
- },
237
- "eval_recall": {
238
- "recall": 0.9292905947776545
239
- },
240
- "eval_runtime": 17.7765,
241
- "eval_samples_per_second": 175.512,
242
- "eval_steps_per_second": 5.513,
243
- "step": 292
244
- },
245
- {
246
- "epoch": 3.08,
247
- "learning_rate": 4.982817869415808e-05,
248
- "loss": 0.3036,
249
- "step": 300
250
- },
251
- {
252
- "epoch": 3.18,
253
- "learning_rate": 4.963726613211149e-05,
254
- "loss": 0.2969,
255
- "step": 310
256
- },
257
- {
258
- "epoch": 3.28,
259
- "learning_rate": 4.944635357006491e-05,
260
- "loss": 0.2796,
261
- "step": 320
262
- },
263
- {
264
- "epoch": 3.38,
265
- "learning_rate": 4.9255441008018334e-05,
266
- "loss": 0.318,
267
- "step": 330
268
- },
269
- {
270
- "epoch": 3.49,
271
- "learning_rate": 4.9064528445971745e-05,
272
- "loss": 0.3019,
273
- "step": 340
274
- },
275
- {
276
- "epoch": 3.59,
277
- "learning_rate": 4.887361588392517e-05,
278
- "loss": 0.2571,
279
- "step": 350
280
- },
281
- {
282
- "epoch": 3.69,
283
- "learning_rate": 4.868270332187858e-05,
284
- "loss": 0.2755,
285
- "step": 360
286
- },
287
- {
288
- "epoch": 3.79,
289
- "learning_rate": 4.8491790759832e-05,
290
- "loss": 0.2548,
291
- "step": 370
292
- },
293
- {
294
- "epoch": 3.9,
295
- "learning_rate": 4.8300878197785415e-05,
296
- "loss": 0.2601,
297
- "step": 380
298
- },
299
- {
300
- "epoch": 4.0,
301
- "learning_rate": 4.810996563573883e-05,
302
- "loss": 0.2727,
303
- "step": 390
304
- },
305
- {
306
- "epoch": 4.0,
307
- "eval_accuracy": {
308
- "accuracy": 0.9355769230769231
309
- },
310
- "eval_f1": {
311
- "f1": 0.932981605076714
312
- },
313
- "eval_loss": 0.17178113758563995,
314
- "eval_precision": {
315
- "precision": 0.9337254057728852
316
- },
317
- "eval_recall": {
318
- "recall": 0.9333800262199571
319
- },
320
- "eval_runtime": 17.7852,
321
- "eval_samples_per_second": 175.426,
322
- "eval_steps_per_second": 5.51,
323
- "step": 390
324
- },
325
- {
326
- "epoch": 4.1,
327
- "learning_rate": 4.791905307369225e-05,
328
- "loss": 0.2403,
329
- "step": 400
330
- },
331
- {
332
- "epoch": 4.21,
333
- "learning_rate": 4.772814051164567e-05,
334
- "loss": 0.2257,
335
- "step": 410
336
- },
337
- {
338
- "epoch": 4.31,
339
- "learning_rate": 4.7537227949599086e-05,
340
- "loss": 0.2515,
341
- "step": 420
342
- },
343
- {
344
- "epoch": 4.41,
345
- "learning_rate": 4.7346315387552504e-05,
346
- "loss": 0.2574,
347
- "step": 430
348
- },
349
- {
350
- "epoch": 4.51,
351
- "learning_rate": 4.715540282550592e-05,
352
- "loss": 0.233,
353
- "step": 440
354
- },
355
- {
356
- "epoch": 4.62,
357
- "learning_rate": 4.696449026345934e-05,
358
- "loss": 0.2705,
359
- "step": 450
360
- },
361
- {
362
- "epoch": 4.72,
363
- "learning_rate": 4.677357770141276e-05,
364
- "loss": 0.2497,
365
- "step": 460
366
- },
367
- {
368
- "epoch": 4.82,
369
- "learning_rate": 4.6582665139366174e-05,
370
- "loss": 0.1984,
371
- "step": 470
372
- },
373
- {
374
- "epoch": 4.92,
375
- "learning_rate": 4.639175257731959e-05,
376
- "loss": 0.2538,
377
- "step": 480
378
- },
379
- {
380
- "epoch": 4.99,
381
- "eval_accuracy": {
382
- "accuracy": 0.958974358974359
383
- },
384
- "eval_f1": {
385
- "f1": 0.9582443373536393
386
- },
387
- "eval_loss": 0.12906195223331451,
388
- "eval_precision": {
389
- "precision": 0.9571693948964237
390
- },
391
- "eval_recall": {
392
- "recall": 0.9596991650961229
393
- },
394
- "eval_runtime": 17.8791,
395
- "eval_samples_per_second": 174.506,
396
- "eval_steps_per_second": 5.481,
397
- "step": 487
398
- },
399
- {
400
- "epoch": 5.03,
401
- "learning_rate": 4.6200840015273e-05,
402
- "loss": 0.2345,
403
- "step": 490
404
- },
405
- {
406
- "epoch": 5.13,
407
- "learning_rate": 4.600992745322643e-05,
408
- "loss": 0.2075,
409
- "step": 500
410
- },
411
- {
412
- "epoch": 5.23,
413
- "learning_rate": 4.581901489117984e-05,
414
- "loss": 0.1964,
415
- "step": 510
416
- },
417
- {
418
- "epoch": 5.33,
419
- "learning_rate": 4.562810232913326e-05,
420
- "loss": 0.2156,
421
- "step": 520
422
- },
423
- {
424
- "epoch": 5.44,
425
- "learning_rate": 4.5437189767086673e-05,
426
- "loss": 0.2044,
427
- "step": 530
428
- },
429
- {
430
- "epoch": 5.54,
431
- "learning_rate": 4.524627720504009e-05,
432
- "loss": 0.2315,
433
- "step": 540
434
- },
435
- {
436
- "epoch": 5.64,
437
- "learning_rate": 4.5055364642993516e-05,
438
- "loss": 0.209,
439
- "step": 550
440
- },
441
- {
442
- "epoch": 5.74,
443
- "learning_rate": 4.4864452080946926e-05,
444
- "loss": 0.2268,
445
- "step": 560
446
- },
447
- {
448
- "epoch": 5.85,
449
- "learning_rate": 4.467353951890035e-05,
450
- "loss": 0.2072,
451
- "step": 570
452
- },
453
- {
454
- "epoch": 5.95,
455
- "learning_rate": 4.448262695685376e-05,
456
- "loss": 0.2325,
457
- "step": 580
458
- },
459
- {
460
- "epoch": 6.0,
461
- "eval_accuracy": {
462
- "accuracy": 0.9618589743589744
463
- },
464
- "eval_f1": {
465
- "f1": 0.9613004013514793
466
- },
467
- "eval_loss": 0.12025142461061478,
468
- "eval_precision": {
469
- "precision": 0.9631754338520714
470
- },
471
- "eval_recall": {
472
- "recall": 0.9597128400571953
473
- },
474
- "eval_runtime": 17.8054,
475
- "eval_samples_per_second": 175.228,
476
- "eval_steps_per_second": 5.504,
477
- "step": 585
478
- },
479
- {
480
- "epoch": 6.05,
481
- "learning_rate": 4.429171439480718e-05,
482
- "loss": 0.2138,
483
- "step": 590
484
- },
485
- {
486
- "epoch": 6.15,
487
- "learning_rate": 4.41008018327606e-05,
488
- "loss": 0.1711,
489
- "step": 600
490
- },
491
- {
492
- "epoch": 6.26,
493
- "learning_rate": 4.3909889270714015e-05,
494
- "loss": 0.1854,
495
- "step": 610
496
- },
497
- {
498
- "epoch": 6.36,
499
- "learning_rate": 4.371897670866743e-05,
500
- "loss": 0.2339,
501
- "step": 620
502
- },
503
- {
504
- "epoch": 6.46,
505
- "learning_rate": 4.352806414662085e-05,
506
- "loss": 0.1882,
507
- "step": 630
508
- },
509
- {
510
- "epoch": 6.56,
511
- "learning_rate": 4.333715158457427e-05,
512
- "loss": 0.1845,
513
- "step": 640
514
- },
515
- {
516
- "epoch": 6.67,
517
- "learning_rate": 4.3146239022527685e-05,
518
- "loss": 0.1987,
519
- "step": 650
520
- },
521
- {
522
- "epoch": 6.77,
523
- "learning_rate": 4.2955326460481096e-05,
524
- "loss": 0.1729,
525
- "step": 660
526
- },
527
- {
528
- "epoch": 6.87,
529
- "learning_rate": 4.276441389843452e-05,
530
- "loss": 0.1943,
531
- "step": 670
532
- },
533
- {
534
- "epoch": 6.97,
535
- "learning_rate": 4.257350133638794e-05,
536
- "loss": 0.1881,
537
- "step": 680
538
- },
539
- {
540
- "epoch": 6.99,
541
- "eval_accuracy": {
542
- "accuracy": 0.9663461538461539
543
- },
544
- "eval_f1": {
545
- "f1": 0.9664890039586649
546
- },
547
- "eval_loss": 0.11088057607412338,
548
- "eval_precision": {
549
- "precision": 0.9656661330930015
550
- },
551
- "eval_recall": {
552
- "recall": 0.9675110508627889
553
- },
554
- "eval_runtime": 18.3707,
555
- "eval_samples_per_second": 169.836,
556
- "eval_steps_per_second": 5.335,
557
- "step": 682
558
- },
559
- {
560
- "epoch": 7.08,
561
- "learning_rate": 4.2382588774341356e-05,
562
- "loss": 0.1836,
563
- "step": 690
564
- },
565
- {
566
- "epoch": 7.18,
567
- "learning_rate": 4.2191676212294774e-05,
568
- "loss": 0.1572,
569
- "step": 700
570
- },
571
- {
572
- "epoch": 7.28,
573
- "learning_rate": 4.2000763650248184e-05,
574
- "loss": 0.1688,
575
- "step": 710
576
- },
577
- {
578
- "epoch": 7.38,
579
- "learning_rate": 4.180985108820161e-05,
580
- "loss": 0.1811,
581
- "step": 720
582
- },
583
- {
584
- "epoch": 7.49,
585
- "learning_rate": 4.161893852615502e-05,
586
- "loss": 0.2036,
587
- "step": 730
588
- },
589
- {
590
- "epoch": 7.59,
591
- "learning_rate": 4.1428025964108444e-05,
592
- "loss": 0.1844,
593
- "step": 740
594
- },
595
- {
596
- "epoch": 7.69,
597
- "learning_rate": 4.1237113402061855e-05,
598
- "loss": 0.174,
599
- "step": 750
600
- },
601
- {
602
- "epoch": 7.79,
603
- "learning_rate": 4.104620084001527e-05,
604
- "loss": 0.1628,
605
- "step": 760
606
- },
607
- {
608
- "epoch": 7.9,
609
- "learning_rate": 4.085528827796869e-05,
610
- "loss": 0.1708,
611
- "step": 770
612
- },
613
- {
614
- "epoch": 8.0,
615
- "learning_rate": 4.066437571592211e-05,
616
- "loss": 0.1675,
617
- "step": 780
618
- },
619
- {
620
- "epoch": 8.0,
621
- "eval_accuracy": {
622
- "accuracy": 0.9653846153846154
623
- },
624
- "eval_f1": {
625
- "f1": 0.9660146786449392
626
- },
627
- "eval_loss": 0.10267416387796402,
628
- "eval_precision": {
629
- "precision": 0.9647284027753311
630
- },
631
- "eval_recall": {
632
- "recall": 0.9674012564920079
633
- },
634
- "eval_runtime": 18.2557,
635
- "eval_samples_per_second": 170.906,
636
- "eval_steps_per_second": 5.368,
637
- "step": 780
638
- },
639
- {
640
- "epoch": 8.1,
641
- "learning_rate": 4.047346315387553e-05,
642
- "loss": 0.1583,
643
- "step": 790
644
- },
645
- {
646
- "epoch": 8.21,
647
- "learning_rate": 4.028255059182894e-05,
648
- "loss": 0.1673,
649
- "step": 800
650
- },
651
- {
652
- "epoch": 8.31,
653
- "learning_rate": 4.009163802978236e-05,
654
- "loss": 0.1805,
655
- "step": 810
656
- },
657
- {
658
- "epoch": 8.41,
659
- "learning_rate": 3.990072546773578e-05,
660
- "loss": 0.1702,
661
- "step": 820
662
- },
663
- {
664
- "epoch": 8.51,
665
- "learning_rate": 3.9709812905689196e-05,
666
- "loss": 0.1659,
667
- "step": 830
668
- },
669
- {
670
- "epoch": 8.62,
671
- "learning_rate": 3.9518900343642614e-05,
672
- "loss": 0.1593,
673
- "step": 840
674
- },
675
- {
676
- "epoch": 8.72,
677
- "learning_rate": 3.932798778159603e-05,
678
- "loss": 0.1556,
679
- "step": 850
680
- },
681
- {
682
- "epoch": 8.82,
683
- "learning_rate": 3.913707521954945e-05,
684
- "loss": 0.1559,
685
- "step": 860
686
- },
687
- {
688
- "epoch": 8.92,
689
- "learning_rate": 3.894616265750287e-05,
690
- "loss": 0.1584,
691
- "step": 870
692
- },
693
- {
694
- "epoch": 8.99,
695
- "eval_accuracy": {
696
- "accuracy": 0.9657051282051282
697
- },
698
- "eval_f1": {
699
- "f1": 0.9659360573535458
700
- },
701
- "eval_loss": 0.10572108626365662,
702
- "eval_precision": {
703
- "precision": 0.9647079224848474
704
- },
705
- "eval_recall": {
706
- "recall": 0.9672918952982554
707
- },
708
- "eval_runtime": 18.9889,
709
- "eval_samples_per_second": 164.306,
710
- "eval_steps_per_second": 5.161,
711
- "step": 877
712
- },
713
- {
714
- "epoch": 9.03,
715
- "learning_rate": 3.875525009545628e-05,
716
- "loss": 0.1552,
717
- "step": 880
718
- },
719
- {
720
- "epoch": 9.13,
721
- "learning_rate": 3.85643375334097e-05,
722
- "loss": 0.1536,
723
- "step": 890
724
- },
725
- {
726
- "epoch": 9.23,
727
- "learning_rate": 3.837342497136312e-05,
728
- "loss": 0.1486,
729
- "step": 900
730
- },
731
- {
732
- "epoch": 9.33,
733
- "learning_rate": 3.818251240931654e-05,
734
- "loss": 0.1404,
735
- "step": 910
736
- },
737
- {
738
- "epoch": 9.44,
739
- "learning_rate": 3.7991599847269955e-05,
740
- "loss": 0.1721,
741
- "step": 920
742
- },
743
- {
744
- "epoch": 9.54,
745
- "learning_rate": 3.7800687285223366e-05,
746
- "loss": 0.144,
747
- "step": 930
748
- },
749
- {
750
- "epoch": 9.64,
751
- "learning_rate": 3.760977472317679e-05,
752
- "loss": 0.1567,
753
- "step": 940
754
- },
755
- {
756
- "epoch": 9.74,
757
- "learning_rate": 3.74188621611302e-05,
758
- "loss": 0.1481,
759
- "step": 950
760
- },
761
- {
762
- "epoch": 9.85,
763
- "learning_rate": 3.7227949599083626e-05,
764
- "loss": 0.1601,
765
- "step": 960
766
- },
767
- {
768
- "epoch": 9.95,
769
- "learning_rate": 3.7037037037037037e-05,
770
- "loss": 0.1316,
771
- "step": 970
772
- },
773
- {
774
- "epoch": 10.0,
775
- "eval_accuracy": {
776
- "accuracy": 0.9685897435897436
777
- },
778
- "eval_f1": {
779
- "f1": 0.9692218377747261
780
- },
781
- "eval_loss": 0.09553277492523193,
782
- "eval_precision": {
783
- "precision": 0.9675383761327713
784
- },
785
- "eval_recall": {
786
- "recall": 0.9710583062570752
787
- },
788
- "eval_runtime": 18.2751,
789
- "eval_samples_per_second": 170.724,
790
- "eval_steps_per_second": 5.362,
791
- "step": 975
792
- },
793
- {
794
- "epoch": 10.05,
795
- "learning_rate": 3.6846124474990454e-05,
796
- "loss": 0.1538,
797
- "step": 980
798
- },
799
- {
800
- "epoch": 10.15,
801
- "learning_rate": 3.665521191294387e-05,
802
- "loss": 0.1388,
803
- "step": 990
804
- },
805
- {
806
- "epoch": 10.26,
807
- "learning_rate": 3.646429935089729e-05,
808
- "loss": 0.1542,
809
- "step": 1000
810
- },
811
- {
812
- "epoch": 10.36,
813
- "learning_rate": 3.6273386788850714e-05,
814
- "loss": 0.138,
815
- "step": 1010
816
- },
817
- {
818
- "epoch": 10.46,
819
- "learning_rate": 3.6082474226804125e-05,
820
- "loss": 0.1372,
821
- "step": 1020
822
- },
823
- {
824
- "epoch": 10.56,
825
- "learning_rate": 3.589156166475754e-05,
826
- "loss": 0.1344,
827
- "step": 1030
828
- },
829
- {
830
- "epoch": 10.67,
831
- "learning_rate": 3.570064910271096e-05,
832
- "loss": 0.172,
833
- "step": 1040
834
- },
835
- {
836
- "epoch": 10.77,
837
- "learning_rate": 3.550973654066438e-05,
838
- "loss": 0.1428,
839
- "step": 1050
840
- },
841
- {
842
- "epoch": 10.87,
843
- "learning_rate": 3.5318823978617795e-05,
844
- "loss": 0.1306,
845
- "step": 1060
846
- },
847
- {
848
- "epoch": 10.97,
849
- "learning_rate": 3.512791141657121e-05,
850
- "loss": 0.1441,
851
- "step": 1070
852
- },
853
- {
854
- "epoch": 10.99,
855
- "eval_accuracy": {
856
- "accuracy": 0.9730769230769231
857
- },
858
- "eval_f1": {
859
- "f1": 0.9729927507703455
860
- },
861
- "eval_loss": 0.0812140703201294,
862
- "eval_precision": {
863
- "precision": 0.9732015595516862
864
- },
865
- "eval_recall": {
866
- "recall": 0.9731353308954022
867
- },
868
- "eval_runtime": 18.3727,
869
- "eval_samples_per_second": 169.817,
870
- "eval_steps_per_second": 5.334,
871
- "step": 1072
872
- },
873
- {
874
- "epoch": 11.08,
875
- "learning_rate": 3.493699885452463e-05,
876
- "loss": 0.1297,
877
- "step": 1080
878
- },
879
- {
880
- "epoch": 11.18,
881
- "learning_rate": 3.474608629247805e-05,
882
- "loss": 0.1292,
883
- "step": 1090
884
- },
885
- {
886
- "epoch": 11.28,
887
- "learning_rate": 3.455517373043146e-05,
888
- "loss": 0.148,
889
- "step": 1100
890
- },
891
- {
892
- "epoch": 11.38,
893
- "learning_rate": 3.4364261168384884e-05,
894
- "loss": 0.1221,
895
- "step": 1110
896
- },
897
- {
898
- "epoch": 11.49,
899
- "learning_rate": 3.41733486063383e-05,
900
- "loss": 0.1326,
901
- "step": 1120
902
- },
903
- {
904
- "epoch": 11.59,
905
- "learning_rate": 3.398243604429171e-05,
906
- "loss": 0.1179,
907
- "step": 1130
908
- },
909
- {
910
- "epoch": 11.69,
911
- "learning_rate": 3.379152348224514e-05,
912
- "loss": 0.0926,
913
- "step": 1140
914
- },
915
- {
916
- "epoch": 11.79,
917
- "learning_rate": 3.360061092019855e-05,
918
- "loss": 0.1297,
919
- "step": 1150
920
- },
921
- {
922
- "epoch": 11.9,
923
- "learning_rate": 3.340969835815197e-05,
924
- "loss": 0.1514,
925
- "step": 1160
926
- },
927
- {
928
- "epoch": 12.0,
929
- "learning_rate": 3.321878579610538e-05,
930
- "loss": 0.1376,
931
- "step": 1170
932
- },
933
- {
934
- "epoch": 12.0,
935
- "eval_accuracy": {
936
- "accuracy": 0.9756410256410256
937
- },
938
- "eval_f1": {
939
- "f1": 0.9758693940548872
940
- },
941
- "eval_loss": 0.0736234188079834,
942
- "eval_precision": {
943
- "precision": 0.9760043308640731
944
- },
945
- "eval_recall": {
946
- "recall": 0.9758007649051665
947
- },
948
- "eval_runtime": 18.2656,
949
- "eval_samples_per_second": 170.813,
950
- "eval_steps_per_second": 5.365,
951
- "step": 1170
952
- },
953
- {
954
- "epoch": 12.1,
955
- "learning_rate": 3.30278732340588e-05,
956
- "loss": 0.1302,
957
- "step": 1180
958
- },
959
- {
960
- "epoch": 12.21,
961
- "learning_rate": 3.283696067201222e-05,
962
- "loss": 0.1298,
963
- "step": 1190
964
- },
965
- {
966
- "epoch": 12.31,
967
- "learning_rate": 3.2646048109965636e-05,
968
- "loss": 0.1108,
969
- "step": 1200
970
- },
971
- {
972
- "epoch": 12.41,
973
- "learning_rate": 3.2455135547919053e-05,
974
- "loss": 0.1264,
975
- "step": 1210
976
- },
977
- {
978
- "epoch": 12.51,
979
- "learning_rate": 3.226422298587247e-05,
980
- "loss": 0.1055,
981
- "step": 1220
982
- },
983
- {
984
- "epoch": 12.62,
985
- "learning_rate": 3.207331042382589e-05,
986
- "loss": 0.1196,
987
- "step": 1230
988
- },
989
- {
990
- "epoch": 12.72,
991
- "learning_rate": 3.1882397861779306e-05,
992
- "loss": 0.1183,
993
- "step": 1240
994
- },
995
- {
996
- "epoch": 12.82,
997
- "learning_rate": 3.1691485299732724e-05,
998
- "loss": 0.1104,
999
- "step": 1250
1000
- },
1001
- {
1002
- "epoch": 12.92,
1003
- "learning_rate": 3.150057273768614e-05,
1004
- "loss": 0.1222,
1005
- "step": 1260
1006
- },
1007
- {
1008
- "epoch": 12.99,
1009
- "eval_accuracy": {
1010
- "accuracy": 0.9782051282051282
1011
- },
1012
- "eval_f1": {
1013
- "f1": 0.9787252371705304
1014
- },
1015
- "eval_loss": 0.07423186302185059,
1016
- "eval_precision": {
1017
- "precision": 0.9795111438345339
1018
- },
1019
- "eval_recall": {
1020
- "recall": 0.9781211797756912
1021
- },
1022
- "eval_runtime": 18.4612,
1023
- "eval_samples_per_second": 169.003,
1024
- "eval_steps_per_second": 5.308,
1025
- "step": 1267
1026
- },
1027
- {
1028
- "epoch": 13.03,
1029
- "learning_rate": 3.130966017563956e-05,
1030
- "loss": 0.1093,
1031
- "step": 1270
1032
- },
1033
- {
1034
- "epoch": 13.13,
1035
- "learning_rate": 3.111874761359298e-05,
1036
- "loss": 0.1213,
1037
- "step": 1280
1038
- },
1039
- {
1040
- "epoch": 13.23,
1041
- "learning_rate": 3.0927835051546395e-05,
1042
- "loss": 0.1017,
1043
- "step": 1290
1044
- },
1045
- {
1046
- "epoch": 13.33,
1047
- "learning_rate": 3.0736922489499806e-05,
1048
- "loss": 0.1413,
1049
- "step": 1300
1050
- },
1051
- {
1052
- "epoch": 13.44,
1053
- "learning_rate": 3.054600992745323e-05,
1054
- "loss": 0.1114,
1055
- "step": 1310
1056
- },
1057
- {
1058
- "epoch": 13.54,
1059
- "learning_rate": 3.0355097365406644e-05,
1060
- "loss": 0.1155,
1061
- "step": 1320
1062
- },
1063
- {
1064
- "epoch": 13.64,
1065
- "learning_rate": 3.0164184803360062e-05,
1066
- "loss": 0.1414,
1067
- "step": 1330
1068
- },
1069
- {
1070
- "epoch": 13.74,
1071
- "learning_rate": 2.9973272241313483e-05,
1072
- "loss": 0.1061,
1073
- "step": 1340
1074
- },
1075
- {
1076
- "epoch": 13.85,
1077
- "learning_rate": 2.9782359679266897e-05,
1078
- "loss": 0.1474,
1079
- "step": 1350
1080
- },
1081
- {
1082
- "epoch": 13.95,
1083
- "learning_rate": 2.9591447117220318e-05,
1084
- "loss": 0.1195,
1085
- "step": 1360
1086
- },
1087
- {
1088
- "epoch": 14.0,
1089
- "eval_accuracy": {
1090
- "accuracy": 0.9746794871794872
1091
- },
1092
- "eval_f1": {
1093
- "f1": 0.9751503673932421
1094
- },
1095
- "eval_loss": 0.07223396748304367,
1096
- "eval_precision": {
1097
- "precision": 0.9748002335329808
1098
- },
1099
- "eval_recall": {
1100
- "recall": 0.9758973037646403
1101
- },
1102
- "eval_runtime": 18.5238,
1103
- "eval_samples_per_second": 168.432,
1104
- "eval_steps_per_second": 5.29,
1105
- "step": 1365
1106
- },
1107
- {
1108
- "epoch": 14.05,
1109
- "learning_rate": 2.9400534555173732e-05,
1110
- "loss": 0.1095,
1111
- "step": 1370
1112
- },
1113
- {
1114
- "epoch": 14.15,
1115
- "learning_rate": 2.920962199312715e-05,
1116
- "loss": 0.1021,
1117
- "step": 1380
1118
- },
1119
- {
1120
- "epoch": 14.26,
1121
- "learning_rate": 2.9018709431080564e-05,
1122
- "loss": 0.1239,
1123
- "step": 1390
1124
- },
1125
- {
1126
- "epoch": 14.36,
1127
- "learning_rate": 2.8827796869033985e-05,
1128
- "loss": 0.1192,
1129
- "step": 1400
1130
- },
1131
- {
1132
- "epoch": 14.46,
1133
- "learning_rate": 2.86368843069874e-05,
1134
- "loss": 0.103,
1135
- "step": 1410
1136
- },
1137
- {
1138
- "epoch": 14.56,
1139
- "learning_rate": 2.844597174494082e-05,
1140
- "loss": 0.1191,
1141
- "step": 1420
1142
- },
1143
- {
1144
- "epoch": 14.67,
1145
- "learning_rate": 2.8255059182894235e-05,
1146
- "loss": 0.114,
1147
- "step": 1430
1148
- },
1149
- {
1150
- "epoch": 14.77,
1151
- "learning_rate": 2.8064146620847653e-05,
1152
- "loss": 0.1096,
1153
- "step": 1440
1154
- },
1155
- {
1156
- "epoch": 14.87,
1157
- "learning_rate": 2.7873234058801074e-05,
1158
- "loss": 0.1124,
1159
- "step": 1450
1160
- },
1161
- {
1162
- "epoch": 14.97,
1163
- "learning_rate": 2.7682321496754488e-05,
1164
- "loss": 0.113,
1165
- "step": 1460
1166
- },
1167
- {
1168
- "epoch": 14.99,
1169
- "eval_accuracy": {
1170
- "accuracy": 0.9775641025641025
1171
- },
1172
- "eval_f1": {
1173
- "f1": 0.9779218456371974
1174
- },
1175
- "eval_loss": 0.0729188397526741,
1176
- "eval_precision": {
1177
- "precision": 0.979047027933676
1178
- },
1179
- "eval_recall": {
1180
- "recall": 0.9770918627040228
1181
- },
1182
- "eval_runtime": 18.5885,
1183
- "eval_samples_per_second": 167.846,
1184
- "eval_steps_per_second": 5.272,
1185
- "step": 1462
1186
- },
1187
- {
1188
- "epoch": 15.08,
1189
- "learning_rate": 2.749140893470791e-05,
1190
- "loss": 0.1092,
1191
- "step": 1470
1192
- },
1193
- {
1194
- "epoch": 15.18,
1195
- "learning_rate": 2.7300496372661323e-05,
1196
- "loss": 0.1213,
1197
- "step": 1480
1198
- },
1199
- {
1200
- "epoch": 15.28,
1201
- "learning_rate": 2.710958381061474e-05,
1202
- "loss": 0.0856,
1203
- "step": 1490
1204
- },
1205
- {
1206
- "epoch": 15.38,
1207
- "learning_rate": 2.6918671248568155e-05,
1208
- "loss": 0.1165,
1209
- "step": 1500
1210
- },
1211
- {
1212
- "epoch": 15.49,
1213
- "learning_rate": 2.6727758686521576e-05,
1214
- "loss": 0.1017,
1215
- "step": 1510
1216
- },
1217
- {
1218
- "epoch": 15.59,
1219
- "learning_rate": 2.653684612447499e-05,
1220
- "loss": 0.1003,
1221
- "step": 1520
1222
- },
1223
- {
1224
- "epoch": 15.69,
1225
- "learning_rate": 2.634593356242841e-05,
1226
- "loss": 0.1121,
1227
- "step": 1530
1228
- },
1229
- {
1230
- "epoch": 15.79,
1231
- "learning_rate": 2.6155021000381826e-05,
1232
- "loss": 0.1338,
1233
- "step": 1540
1234
- },
1235
- {
1236
- "epoch": 15.9,
1237
- "learning_rate": 2.5964108438335243e-05,
1238
- "loss": 0.113,
1239
- "step": 1550
1240
- },
1241
- {
1242
- "epoch": 16.0,
1243
- "learning_rate": 2.5773195876288658e-05,
1244
- "loss": 0.1111,
1245
- "step": 1560
1246
- },
1247
- {
1248
- "epoch": 16.0,
1249
- "eval_accuracy": {
1250
- "accuracy": 0.9772435897435897
1251
- },
1252
- "eval_f1": {
1253
- "f1": 0.9778308709990938
1254
- },
1255
- "eval_loss": 0.06960152089595795,
1256
- "eval_precision": {
1257
- "precision": 0.9779158764095219
1258
- },
1259
- "eval_recall": {
1260
- "recall": 0.9778124028636349
1261
- },
1262
- "eval_runtime": 18.5763,
1263
- "eval_samples_per_second": 167.956,
1264
- "eval_steps_per_second": 5.276,
1265
- "step": 1560
1266
- },
1267
- {
1268
- "epoch": 16.1,
1269
- "learning_rate": 2.558228331424208e-05,
1270
- "loss": 0.1015,
1271
- "step": 1570
1272
- },
1273
- {
1274
- "epoch": 16.21,
1275
- "learning_rate": 2.5391370752195496e-05,
1276
- "loss": 0.1049,
1277
- "step": 1580
1278
- },
1279
- {
1280
- "epoch": 16.31,
1281
- "learning_rate": 2.520045819014891e-05,
1282
- "loss": 0.094,
1283
- "step": 1590
1284
- },
1285
- {
1286
- "epoch": 16.41,
1287
- "learning_rate": 2.500954562810233e-05,
1288
- "loss": 0.1041,
1289
- "step": 1600
1290
- },
1291
- {
1292
- "epoch": 16.51,
1293
- "learning_rate": 2.4818633066055746e-05,
1294
- "loss": 0.0772,
1295
- "step": 1610
1296
- },
1297
- {
1298
- "epoch": 16.62,
1299
- "learning_rate": 2.4627720504009167e-05,
1300
- "loss": 0.1079,
1301
- "step": 1620
1302
- },
1303
- {
1304
- "epoch": 16.72,
1305
- "learning_rate": 2.4436807941962585e-05,
1306
- "loss": 0.0942,
1307
- "step": 1630
1308
- },
1309
- {
1310
- "epoch": 16.82,
1311
- "learning_rate": 2.4245895379916e-05,
1312
- "loss": 0.1068,
1313
- "step": 1640
1314
- },
1315
- {
1316
- "epoch": 16.92,
1317
- "learning_rate": 2.4054982817869417e-05,
1318
- "loss": 0.1016,
1319
- "step": 1650
1320
- },
1321
- {
1322
- "epoch": 16.99,
1323
- "eval_accuracy": {
1324
- "accuracy": 0.9737179487179487
1325
- },
1326
- "eval_f1": {
1327
- "f1": 0.9745961111652444
1328
- },
1329
- "eval_loss": 0.08040237426757812,
1330
- "eval_precision": {
1331
- "precision": 0.9733897161308854
1332
- },
1333
- "eval_recall": {
1334
- "recall": 0.9760199723830122
1335
- },
1336
- "eval_runtime": 18.4943,
1337
- "eval_samples_per_second": 168.701,
1338
- "eval_steps_per_second": 5.299,
1339
- "step": 1657
1340
- },
1341
- {
1342
- "epoch": 17.03,
1343
- "learning_rate": 2.3864070255822834e-05,
1344
- "loss": 0.0902,
1345
- "step": 1660
1346
- },
1347
- {
1348
- "epoch": 17.13,
1349
- "learning_rate": 2.3673157693776252e-05,
1350
- "loss": 0.084,
1351
- "step": 1670
1352
- },
1353
- {
1354
- "epoch": 17.23,
1355
- "learning_rate": 2.348224513172967e-05,
1356
- "loss": 0.117,
1357
- "step": 1680
1358
- },
1359
- {
1360
- "epoch": 17.33,
1361
- "learning_rate": 2.3291332569683087e-05,
1362
- "loss": 0.0899,
1363
- "step": 1690
1364
- },
1365
- {
1366
- "epoch": 17.44,
1367
- "learning_rate": 2.31004200076365e-05,
1368
- "loss": 0.1025,
1369
- "step": 1700
1370
- },
1371
- {
1372
- "epoch": 17.54,
1373
- "learning_rate": 2.290950744558992e-05,
1374
- "loss": 0.0987,
1375
- "step": 1710
1376
- },
1377
- {
1378
- "epoch": 17.64,
1379
- "learning_rate": 2.2718594883543337e-05,
1380
- "loss": 0.1038,
1381
- "step": 1720
1382
- },
1383
- {
1384
- "epoch": 17.74,
1385
- "learning_rate": 2.2527682321496758e-05,
1386
- "loss": 0.1018,
1387
- "step": 1730
1388
- },
1389
- {
1390
- "epoch": 17.85,
1391
- "learning_rate": 2.2336769759450175e-05,
1392
- "loss": 0.1092,
1393
- "step": 1740
1394
- },
1395
- {
1396
- "epoch": 17.95,
1397
- "learning_rate": 2.214585719740359e-05,
1398
- "loss": 0.1212,
1399
- "step": 1750
1400
- },
1401
- {
1402
- "epoch": 18.0,
1403
- "eval_accuracy": {
1404
- "accuracy": 0.978525641025641
1405
- },
1406
- "eval_f1": {
1407
- "f1": 0.9792426784918802
1408
- },
1409
- "eval_loss": 0.07058186829090118,
1410
- "eval_precision": {
1411
- "precision": 0.9780537479064654
1412
- },
1413
- "eval_recall": {
1414
- "recall": 0.980517981186892
1415
- },
1416
- "eval_runtime": 18.5461,
1417
- "eval_samples_per_second": 168.229,
1418
- "eval_steps_per_second": 5.284,
1419
- "step": 1755
1420
- },
1421
- {
1422
- "epoch": 18.05,
1423
- "learning_rate": 2.1954944635357007e-05,
1424
- "loss": 0.1046,
1425
- "step": 1760
1426
- },
1427
- {
1428
- "epoch": 18.15,
1429
- "learning_rate": 2.1764032073310425e-05,
1430
- "loss": 0.0763,
1431
- "step": 1770
1432
- },
1433
- {
1434
- "epoch": 18.26,
1435
- "learning_rate": 2.1573119511263843e-05,
1436
- "loss": 0.1083,
1437
- "step": 1780
1438
- },
1439
- {
1440
- "epoch": 18.36,
1441
- "learning_rate": 2.138220694921726e-05,
1442
- "loss": 0.0966,
1443
- "step": 1790
1444
- },
1445
- {
1446
- "epoch": 18.46,
1447
- "learning_rate": 2.1191294387170678e-05,
1448
- "loss": 0.0917,
1449
- "step": 1800
1450
- },
1451
- {
1452
- "epoch": 18.56,
1453
- "learning_rate": 2.1000381825124092e-05,
1454
- "loss": 0.0893,
1455
- "step": 1810
1456
- },
1457
- {
1458
- "epoch": 18.67,
1459
- "learning_rate": 2.080946926307751e-05,
1460
- "loss": 0.1147,
1461
- "step": 1820
1462
- },
1463
- {
1464
- "epoch": 18.77,
1465
- "learning_rate": 2.0618556701030927e-05,
1466
- "loss": 0.1105,
1467
- "step": 1830
1468
- },
1469
- {
1470
- "epoch": 18.87,
1471
- "learning_rate": 2.0427644138984345e-05,
1472
- "loss": 0.1257,
1473
- "step": 1840
1474
- },
1475
- {
1476
- "epoch": 18.97,
1477
- "learning_rate": 2.0236731576937766e-05,
1478
- "loss": 0.0773,
1479
- "step": 1850
1480
- },
1481
- {
1482
- "epoch": 18.99,
1483
- "eval_accuracy": {
1484
- "accuracy": 0.9782051282051282
1485
- },
1486
- "eval_f1": {
1487
- "f1": 0.9790037742544285
1488
- },
1489
- "eval_loss": 0.06617825478315353,
1490
- "eval_precision": {
1491
- "precision": 0.9789744968624101
1492
- },
1493
- "eval_recall": {
1494
- "recall": 0.9791140613622205
1495
- },
1496
- "eval_runtime": 18.5258,
1497
- "eval_samples_per_second": 168.414,
1498
- "eval_steps_per_second": 5.29,
1499
- "step": 1852
1500
- },
1501
- {
1502
- "epoch": 19.08,
1503
- "learning_rate": 2.004581901489118e-05,
1504
- "loss": 0.0801,
1505
- "step": 1860
1506
- },
1507
- {
1508
- "epoch": 19.18,
1509
- "learning_rate": 1.9854906452844598e-05,
1510
- "loss": 0.1068,
1511
- "step": 1870
1512
- },
1513
- {
1514
- "epoch": 19.28,
1515
- "learning_rate": 1.9663993890798016e-05,
1516
- "loss": 0.0728,
1517
- "step": 1880
1518
- },
1519
- {
1520
- "epoch": 19.38,
1521
- "learning_rate": 1.9473081328751433e-05,
1522
- "loss": 0.0868,
1523
- "step": 1890
1524
- },
1525
- {
1526
- "epoch": 19.49,
1527
- "learning_rate": 1.928216876670485e-05,
1528
- "loss": 0.0996,
1529
- "step": 1900
1530
- },
1531
- {
1532
- "epoch": 19.59,
1533
- "learning_rate": 1.909125620465827e-05,
1534
- "loss": 0.0821,
1535
- "step": 1910
1536
- },
1537
- {
1538
- "epoch": 19.69,
1539
- "learning_rate": 1.8900343642611683e-05,
1540
- "loss": 0.0732,
1541
- "step": 1920
1542
- },
1543
- {
1544
- "epoch": 19.79,
1545
- "learning_rate": 1.87094310805651e-05,
1546
- "loss": 0.0977,
1547
- "step": 1930
1548
- },
1549
- {
1550
- "epoch": 19.9,
1551
- "learning_rate": 1.8518518518518518e-05,
1552
- "loss": 0.0773,
1553
- "step": 1940
1554
- },
1555
- {
1556
- "epoch": 20.0,
1557
- "learning_rate": 1.8327605956471936e-05,
1558
- "loss": 0.0807,
1559
- "step": 1950
1560
- },
1561
- {
1562
- "epoch": 20.0,
1563
- "eval_accuracy": {
1564
- "accuracy": 0.9782051282051282
1565
- },
1566
- "eval_f1": {
1567
- "f1": 0.978849268707168
1568
- },
1569
- "eval_loss": 0.06757049262523651,
1570
- "eval_precision": {
1571
- "precision": 0.9789056277749657
1572
- },
1573
- "eval_recall": {
1574
- "recall": 0.9788691085482305
1575
- },
1576
- "eval_runtime": 18.3657,
1577
- "eval_samples_per_second": 169.882,
1578
- "eval_steps_per_second": 5.336,
1579
- "step": 1950
1580
- },
1581
- {
1582
- "epoch": 20.1,
1583
- "learning_rate": 1.8136693394425357e-05,
1584
- "loss": 0.0913,
1585
- "step": 1960
1586
- },
1587
- {
1588
- "epoch": 20.21,
1589
- "learning_rate": 1.794578083237877e-05,
1590
- "loss": 0.0977,
1591
- "step": 1970
1592
- },
1593
- {
1594
- "epoch": 20.31,
1595
- "learning_rate": 1.775486827033219e-05,
1596
- "loss": 0.0854,
1597
- "step": 1980
1598
- },
1599
- {
1600
- "epoch": 20.41,
1601
- "learning_rate": 1.7563955708285607e-05,
1602
- "loss": 0.0764,
1603
- "step": 1990
1604
- },
1605
- {
1606
- "epoch": 20.51,
1607
- "learning_rate": 1.7373043146239024e-05,
1608
- "loss": 0.0925,
1609
- "step": 2000
1610
- },
1611
- {
1612
- "epoch": 20.62,
1613
- "learning_rate": 1.7182130584192442e-05,
1614
- "loss": 0.0888,
1615
- "step": 2010
1616
- },
1617
- {
1618
- "epoch": 20.72,
1619
- "learning_rate": 1.6991218022145856e-05,
1620
- "loss": 0.0684,
1621
- "step": 2020
1622
- },
1623
- {
1624
- "epoch": 20.82,
1625
- "learning_rate": 1.6800305460099274e-05,
1626
- "loss": 0.0826,
1627
- "step": 2030
1628
- },
1629
- {
1630
- "epoch": 20.92,
1631
- "learning_rate": 1.660939289805269e-05,
1632
- "loss": 0.1051,
1633
- "step": 2040
1634
- },
1635
- {
1636
- "epoch": 20.99,
1637
- "eval_accuracy": {
1638
- "accuracy": 0.9801282051282051
1639
- },
1640
- "eval_f1": {
1641
- "f1": 0.9809129407528451
1642
- },
1643
- "eval_loss": 0.058978818356990814,
1644
- "eval_precision": {
1645
- "precision": 0.98154804184011
1646
- },
1647
- "eval_recall": {
1648
- "recall": 0.9802969452515257
1649
- },
1650
- "eval_runtime": 18.6326,
1651
- "eval_samples_per_second": 167.449,
1652
- "eval_steps_per_second": 5.26,
1653
- "step": 2047
1654
- },
1655
- {
1656
- "epoch": 21.03,
1657
- "learning_rate": 1.641848033600611e-05,
1658
- "loss": 0.0828,
1659
- "step": 2050
1660
- },
1661
- {
1662
- "epoch": 21.13,
1663
- "learning_rate": 1.6227567773959527e-05,
1664
- "loss": 0.0801,
1665
- "step": 2060
1666
- },
1667
- {
1668
- "epoch": 21.23,
1669
- "learning_rate": 1.6036655211912944e-05,
1670
- "loss": 0.0999,
1671
- "step": 2070
1672
- },
1673
- {
1674
- "epoch": 21.33,
1675
- "learning_rate": 1.5845742649866362e-05,
1676
- "loss": 0.0773,
1677
- "step": 2080
1678
- },
1679
- {
1680
- "epoch": 21.44,
1681
- "learning_rate": 1.565483008781978e-05,
1682
- "loss": 0.0726,
1683
- "step": 2090
1684
- },
1685
- {
1686
- "epoch": 21.54,
1687
- "learning_rate": 1.5463917525773197e-05,
1688
- "loss": 0.0941,
1689
- "step": 2100
1690
- },
1691
- {
1692
- "epoch": 21.64,
1693
- "learning_rate": 1.5273004963726615e-05,
1694
- "loss": 0.0859,
1695
- "step": 2110
1696
- },
1697
- {
1698
- "epoch": 21.74,
1699
- "learning_rate": 1.5082092401680031e-05,
1700
- "loss": 0.0713,
1701
- "step": 2120
1702
- },
1703
- {
1704
- "epoch": 21.85,
1705
- "learning_rate": 1.4891179839633449e-05,
1706
- "loss": 0.0847,
1707
- "step": 2130
1708
- },
1709
- {
1710
- "epoch": 21.95,
1711
- "learning_rate": 1.4700267277586866e-05,
1712
- "loss": 0.0941,
1713
- "step": 2140
1714
- },
1715
- {
1716
- "epoch": 22.0,
1717
- "eval_accuracy": {
1718
- "accuracy": 0.982051282051282
1719
- },
1720
- "eval_f1": {
1721
- "f1": 0.9825777588621161
1722
- },
1723
- "eval_loss": 0.06584476679563522,
1724
- "eval_precision": {
1725
- "precision": 0.9828165773418366
1726
- },
1727
- "eval_recall": {
1728
- "recall": 0.9824456736033335
1729
- },
1730
- "eval_runtime": 18.6363,
1731
- "eval_samples_per_second": 167.415,
1732
- "eval_steps_per_second": 5.259,
1733
- "step": 2145
1734
- },
1735
- {
1736
- "epoch": 22.05,
1737
- "learning_rate": 1.4509354715540282e-05,
1738
- "loss": 0.0874,
1739
- "step": 2150
1740
- },
1741
- {
1742
- "epoch": 22.15,
1743
- "learning_rate": 1.43184421534937e-05,
1744
- "loss": 0.0807,
1745
- "step": 2160
1746
- },
1747
- {
1748
- "epoch": 22.26,
1749
- "learning_rate": 1.4127529591447117e-05,
1750
- "loss": 0.0662,
1751
- "step": 2170
1752
- },
1753
- {
1754
- "epoch": 22.36,
1755
- "learning_rate": 1.3936617029400537e-05,
1756
- "loss": 0.0826,
1757
- "step": 2180
1758
- },
1759
- {
1760
- "epoch": 22.46,
1761
- "learning_rate": 1.3745704467353954e-05,
1762
- "loss": 0.0903,
1763
- "step": 2190
1764
- },
1765
- {
1766
- "epoch": 22.56,
1767
- "learning_rate": 1.355479190530737e-05,
1768
- "loss": 0.0781,
1769
- "step": 2200
1770
- },
1771
- {
1772
- "epoch": 22.67,
1773
- "learning_rate": 1.3363879343260788e-05,
1774
- "loss": 0.0813,
1775
- "step": 2210
1776
- },
1777
- {
1778
- "epoch": 22.77,
1779
- "learning_rate": 1.3172966781214206e-05,
1780
- "loss": 0.078,
1781
- "step": 2220
1782
- },
1783
- {
1784
- "epoch": 22.87,
1785
- "learning_rate": 1.2982054219167622e-05,
1786
- "loss": 0.0969,
1787
- "step": 2230
1788
- },
1789
- {
1790
- "epoch": 22.97,
1791
- "learning_rate": 1.279114165712104e-05,
1792
- "loss": 0.0849,
1793
- "step": 2240
1794
- },
1795
- {
1796
- "epoch": 22.99,
1797
- "eval_accuracy": {
1798
- "accuracy": 0.9807692307692307
1799
- },
1800
- "eval_f1": {
1801
- "f1": 0.9814188968048345
1802
- },
1803
- "eval_loss": 0.06442565470933914,
1804
- "eval_precision": {
1805
- "precision": 0.981735779695144
1806
- },
1807
- "eval_recall": {
1808
- "recall": 0.9811890642850706
1809
- },
1810
- "eval_runtime": 18.5305,
1811
- "eval_samples_per_second": 168.371,
1812
- "eval_steps_per_second": 5.289,
1813
- "step": 2242
1814
- },
1815
- {
1816
- "epoch": 23.08,
1817
- "learning_rate": 1.2600229095074455e-05,
1818
- "loss": 0.0853,
1819
- "step": 2250
1820
- },
1821
- {
1822
- "epoch": 23.18,
1823
- "learning_rate": 1.2409316533027873e-05,
1824
- "loss": 0.0493,
1825
- "step": 2260
1826
- },
1827
- {
1828
- "epoch": 23.28,
1829
- "learning_rate": 1.2218403970981292e-05,
1830
- "loss": 0.0917,
1831
- "step": 2270
1832
- },
1833
- {
1834
- "epoch": 23.38,
1835
- "learning_rate": 1.2027491408934708e-05,
1836
- "loss": 0.0708,
1837
- "step": 2280
1838
- },
1839
- {
1840
- "epoch": 23.49,
1841
- "learning_rate": 1.1836578846888126e-05,
1842
- "loss": 0.0671,
1843
- "step": 2290
1844
- },
1845
- {
1846
- "epoch": 23.59,
1847
- "learning_rate": 1.1645666284841544e-05,
1848
- "loss": 0.0724,
1849
- "step": 2300
1850
- },
1851
- {
1852
- "epoch": 23.69,
1853
- "learning_rate": 1.145475372279496e-05,
1854
- "loss": 0.0674,
1855
- "step": 2310
1856
- },
1857
- {
1858
- "epoch": 23.79,
1859
- "learning_rate": 1.1263841160748379e-05,
1860
- "loss": 0.0712,
1861
- "step": 2320
1862
- },
1863
- {
1864
- "epoch": 23.9,
1865
- "learning_rate": 1.1072928598701795e-05,
1866
- "loss": 0.0812,
1867
- "step": 2330
1868
- },
1869
- {
1870
- "epoch": 24.0,
1871
- "learning_rate": 1.0882016036655212e-05,
1872
- "loss": 0.0754,
1873
- "step": 2340
1874
- },
1875
- {
1876
- "epoch": 24.0,
1877
- "eval_accuracy": {
1878
- "accuracy": 0.9814102564102564
1879
- },
1880
- "eval_f1": {
1881
- "f1": 0.9818759081081004
1882
- },
1883
- "eval_loss": 0.06598126143217087,
1884
- "eval_precision": {
1885
- "precision": 0.9823138375067749
1886
- },
1887
- "eval_recall": {
1888
- "recall": 0.9815646778958121
1889
- },
1890
- "eval_runtime": 18.3926,
1891
- "eval_samples_per_second": 169.633,
1892
- "eval_steps_per_second": 5.328,
1893
- "step": 2340
1894
- },
1895
- {
1896
- "epoch": 24.1,
1897
- "learning_rate": 1.069110347460863e-05,
1898
- "loss": 0.0577,
1899
- "step": 2350
1900
- },
1901
- {
1902
- "epoch": 24.21,
1903
- "learning_rate": 1.0500190912562046e-05,
1904
- "loss": 0.0724,
1905
- "step": 2360
1906
- },
1907
- {
1908
- "epoch": 24.31,
1909
- "learning_rate": 1.0309278350515464e-05,
1910
- "loss": 0.0816,
1911
- "step": 2370
1912
- },
1913
- {
1914
- "epoch": 24.41,
1915
- "learning_rate": 1.0118365788468883e-05,
1916
- "loss": 0.0759,
1917
- "step": 2380
1918
- },
1919
- {
1920
- "epoch": 24.51,
1921
- "learning_rate": 9.927453226422299e-06,
1922
- "loss": 0.0851,
1923
- "step": 2390
1924
- },
1925
- {
1926
- "epoch": 24.62,
1927
- "learning_rate": 9.736540664375717e-06,
1928
- "loss": 0.068,
1929
- "step": 2400
1930
- },
1931
- {
1932
- "epoch": 24.72,
1933
- "learning_rate": 9.545628102329134e-06,
1934
- "loss": 0.0685,
1935
- "step": 2410
1936
- },
1937
- {
1938
- "epoch": 24.82,
1939
- "learning_rate": 9.35471554028255e-06,
1940
- "loss": 0.084,
1941
- "step": 2420
1942
- },
1943
- {
1944
- "epoch": 24.92,
1945
- "learning_rate": 9.163802978235968e-06,
1946
- "loss": 0.0693,
1947
- "step": 2430
1948
- },
1949
- {
1950
- "epoch": 24.99,
1951
- "eval_accuracy": {
1952
- "accuracy": 0.9810897435897435
1953
- },
1954
- "eval_f1": {
1955
- "f1": 0.9818163582669245
1956
- },
1957
- "eval_loss": 0.05909496173262596,
1958
- "eval_precision": {
1959
- "precision": 0.9821763874529763
1960
- },
1961
- "eval_recall": {
1962
- "recall": 0.9815015240688245
1963
- },
1964
- "eval_runtime": 18.5026,
1965
- "eval_samples_per_second": 168.625,
1966
- "eval_steps_per_second": 5.297,
1967
- "step": 2437
1968
- },
1969
- {
1970
- "epoch": 25.03,
1971
- "learning_rate": 8.972890416189386e-06,
1972
- "loss": 0.0717,
1973
- "step": 2440
1974
- },
1975
- {
1976
- "epoch": 25.13,
1977
- "learning_rate": 8.781977854142803e-06,
1978
- "loss": 0.0631,
1979
- "step": 2450
1980
- },
1981
- {
1982
- "epoch": 25.23,
1983
- "learning_rate": 8.591065292096221e-06,
1984
- "loss": 0.0514,
1985
- "step": 2460
1986
- },
1987
- {
1988
- "epoch": 25.33,
1989
- "learning_rate": 8.400152730049637e-06,
1990
- "loss": 0.0706,
1991
- "step": 2470
1992
- },
1993
- {
1994
- "epoch": 25.44,
1995
- "learning_rate": 8.209240168003055e-06,
1996
- "loss": 0.0626,
1997
- "step": 2480
1998
- },
1999
- {
2000
- "epoch": 25.54,
2001
- "learning_rate": 8.018327605956472e-06,
2002
- "loss": 0.0548,
2003
- "step": 2490
2004
- },
2005
- {
2006
- "epoch": 25.64,
2007
- "learning_rate": 7.82741504390989e-06,
2008
- "loss": 0.0657,
2009
- "step": 2500
2010
- },
2011
- {
2012
- "epoch": 25.74,
2013
- "learning_rate": 7.636502481863307e-06,
2014
- "loss": 0.0993,
2015
- "step": 2510
2016
- },
2017
- {
2018
- "epoch": 25.85,
2019
- "learning_rate": 7.445589919816724e-06,
2020
- "loss": 0.0688,
2021
- "step": 2520
2022
- },
2023
- {
2024
- "epoch": 25.95,
2025
- "learning_rate": 7.254677357770141e-06,
2026
- "loss": 0.0468,
2027
- "step": 2530
2028
- },
2029
- {
2030
- "epoch": 26.0,
2031
- "eval_accuracy": {
2032
- "accuracy": 0.9810897435897435
2033
- },
2034
- "eval_f1": {
2035
- "f1": 0.9817492175527575
2036
- },
2037
- "eval_loss": 0.06363129615783691,
2038
- "eval_precision": {
2039
- "precision": 0.9815769384870415
2040
- },
2041
- "eval_recall": {
2042
- "recall": 0.9820001468845976
2043
- },
2044
- "eval_runtime": 18.4562,
2045
- "eval_samples_per_second": 169.049,
2046
- "eval_steps_per_second": 5.31,
2047
- "step": 2535
2048
- },
2049
- {
2050
- "epoch": 26.05,
2051
- "learning_rate": 7.063764795723559e-06,
2052
- "loss": 0.071,
2053
- "step": 2540
2054
- },
2055
- {
2056
- "epoch": 26.15,
2057
- "learning_rate": 6.872852233676977e-06,
2058
- "loss": 0.0776,
2059
- "step": 2550
2060
- },
2061
- {
2062
- "epoch": 26.26,
2063
- "learning_rate": 6.681939671630394e-06,
2064
- "loss": 0.061,
2065
- "step": 2560
2066
- },
2067
- {
2068
- "epoch": 26.36,
2069
- "learning_rate": 6.491027109583811e-06,
2070
- "loss": 0.0563,
2071
- "step": 2570
2072
- },
2073
- {
2074
- "epoch": 26.46,
2075
- "learning_rate": 6.300114547537228e-06,
2076
- "loss": 0.0673,
2077
- "step": 2580
2078
- },
2079
- {
2080
- "epoch": 26.56,
2081
- "learning_rate": 6.109201985490646e-06,
2082
- "loss": 0.0569,
2083
- "step": 2590
2084
- },
2085
- {
2086
- "epoch": 26.67,
2087
- "learning_rate": 5.918289423444063e-06,
2088
- "loss": 0.069,
2089
- "step": 2600
2090
- },
2091
- {
2092
- "epoch": 26.77,
2093
- "learning_rate": 5.72737686139748e-06,
2094
- "loss": 0.0703,
2095
- "step": 2610
2096
- },
2097
- {
2098
- "epoch": 26.87,
2099
- "learning_rate": 5.536464299350897e-06,
2100
- "loss": 0.0753,
2101
- "step": 2620
2102
- },
2103
- {
2104
- "epoch": 26.97,
2105
- "learning_rate": 5.345551737304315e-06,
2106
- "loss": 0.0626,
2107
- "step": 2630
2108
- },
2109
- {
2110
- "epoch": 26.99,
2111
- "eval_accuracy": {
2112
- "accuracy": 0.9823717948717948
2113
- },
2114
- "eval_f1": {
2115
- "f1": 0.9828710628760066
2116
- },
2117
- "eval_loss": 0.06039305403828621,
2118
- "eval_precision": {
2119
- "precision": 0.9833379982045161
2120
- },
2121
- "eval_recall": {
2122
- "recall": 0.982495297966067
2123
- },
2124
- "eval_runtime": 18.2954,
2125
- "eval_samples_per_second": 170.535,
2126
- "eval_steps_per_second": 5.357,
2127
- "step": 2632
2128
- },
2129
- {
2130
- "epoch": 27.08,
2131
- "learning_rate": 5.154639175257732e-06,
2132
- "loss": 0.0665,
2133
- "step": 2640
2134
- },
2135
- {
2136
- "epoch": 27.18,
2137
- "learning_rate": 4.9637266132111495e-06,
2138
- "loss": 0.0657,
2139
- "step": 2650
2140
- },
2141
- {
2142
- "epoch": 27.28,
2143
- "learning_rate": 4.772814051164567e-06,
2144
- "loss": 0.0555,
2145
- "step": 2660
2146
- },
2147
- {
2148
- "epoch": 27.38,
2149
- "learning_rate": 4.581901489117984e-06,
2150
- "loss": 0.0379,
2151
- "step": 2670
2152
- },
2153
- {
2154
- "epoch": 27.49,
2155
- "learning_rate": 4.390988927071402e-06,
2156
- "loss": 0.0643,
2157
- "step": 2680
2158
- },
2159
- {
2160
- "epoch": 27.59,
2161
- "learning_rate": 4.2000763650248184e-06,
2162
- "loss": 0.0777,
2163
- "step": 2690
2164
- },
2165
- {
2166
- "epoch": 27.69,
2167
- "learning_rate": 4.009163802978236e-06,
2168
- "loss": 0.0632,
2169
- "step": 2700
2170
- },
2171
- {
2172
- "epoch": 27.79,
2173
- "learning_rate": 3.818251240931654e-06,
2174
- "loss": 0.0439,
2175
- "step": 2710
2176
- },
2177
- {
2178
- "epoch": 27.9,
2179
- "learning_rate": 3.6273386788850705e-06,
2180
- "loss": 0.0736,
2181
- "step": 2720
2182
- },
2183
- {
2184
- "epoch": 28.0,
2185
- "learning_rate": 3.4364261168384886e-06,
2186
- "loss": 0.0565,
2187
- "step": 2730
2188
- },
2189
- {
2190
- "epoch": 28.0,
2191
- "eval_accuracy": {
2192
- "accuracy": 0.9817307692307692
2193
- },
2194
- "eval_f1": {
2195
- "f1": 0.9824528709031027
2196
- },
2197
- "eval_loss": 0.0612463541328907,
2198
- "eval_precision": {
2199
- "precision": 0.9824245712456766
2200
- },
2201
- "eval_recall": {
2202
- "recall": 0.9825037931142255
2203
- },
2204
- "eval_runtime": 18.4139,
2205
- "eval_samples_per_second": 169.437,
2206
- "eval_steps_per_second": 5.322,
2207
- "step": 2730
2208
- },
2209
- {
2210
- "epoch": 28.1,
2211
- "learning_rate": 3.2455135547919054e-06,
2212
- "loss": 0.0765,
2213
- "step": 2740
2214
- },
2215
- {
2216
- "epoch": 28.21,
2217
- "learning_rate": 3.054600992745323e-06,
2218
- "loss": 0.057,
2219
- "step": 2750
2220
- },
2221
- {
2222
- "epoch": 28.31,
2223
- "learning_rate": 2.86368843069874e-06,
2224
- "loss": 0.0602,
2225
- "step": 2760
2226
- },
2227
- {
2228
- "epoch": 28.41,
2229
- "learning_rate": 2.6727758686521575e-06,
2230
- "loss": 0.0624,
2231
- "step": 2770
2232
- },
2233
- {
2234
- "epoch": 28.51,
2235
- "learning_rate": 2.4818633066055748e-06,
2236
- "loss": 0.0588,
2237
- "step": 2780
2238
- },
2239
- {
2240
- "epoch": 28.62,
2241
- "learning_rate": 2.290950744558992e-06,
2242
- "loss": 0.0553,
2243
- "step": 2790
2244
- },
2245
- {
2246
- "epoch": 28.72,
2247
- "learning_rate": 2.1000381825124092e-06,
2248
- "loss": 0.0551,
2249
- "step": 2800
2250
- },
2251
- {
2252
- "epoch": 28.82,
2253
- "learning_rate": 1.909125620465827e-06,
2254
- "loss": 0.0579,
2255
- "step": 2810
2256
- },
2257
- {
2258
- "epoch": 28.92,
2259
- "learning_rate": 1.7182130584192443e-06,
2260
- "loss": 0.0428,
2261
- "step": 2820
2262
- },
2263
- {
2264
- "epoch": 28.99,
2265
- "eval_accuracy": {
2266
- "accuracy": 0.9814102564102564
2267
- },
2268
- "eval_f1": {
2269
- "f1": 0.9820040000167444
2270
- },
2271
- "eval_loss": 0.060263678431510925,
2272
- "eval_precision": {
2273
- "precision": 0.9821559799758683
2274
- },
2275
- "eval_recall": {
2276
- "recall": 0.9818903265571605
2277
- },
2278
- "eval_runtime": 18.3859,
2279
- "eval_samples_per_second": 169.696,
2280
- "eval_steps_per_second": 5.33,
2281
- "step": 2827
2282
- },
2283
- {
2284
- "epoch": 29.03,
2285
- "learning_rate": 1.5273004963726615e-06,
2286
- "loss": 0.0647,
2287
- "step": 2830
2288
- },
2289
- {
2290
- "epoch": 29.13,
2291
- "learning_rate": 1.3363879343260788e-06,
2292
- "loss": 0.0592,
2293
- "step": 2840
2294
- },
2295
- {
2296
- "epoch": 29.23,
2297
- "learning_rate": 1.145475372279496e-06,
2298
- "loss": 0.0607,
2299
- "step": 2850
2300
- },
2301
- {
2302
- "epoch": 29.33,
2303
- "learning_rate": 9.545628102329134e-07,
2304
- "loss": 0.0531,
2305
- "step": 2860
2306
- },
2307
- {
2308
- "epoch": 29.44,
2309
- "learning_rate": 7.636502481863308e-07,
2310
- "loss": 0.0679,
2311
- "step": 2870
2312
- },
2313
- {
2314
- "epoch": 29.54,
2315
- "learning_rate": 5.72737686139748e-07,
2316
- "loss": 0.0672,
2317
- "step": 2880
2318
- },
2319
- {
2320
- "epoch": 29.64,
2321
- "learning_rate": 3.818251240931654e-07,
2322
- "loss": 0.039,
2323
- "step": 2890
2324
- },
2325
- {
2326
- "epoch": 29.74,
2327
- "learning_rate": 1.909125620465827e-07,
2328
- "loss": 0.0397,
2329
- "step": 2900
2330
- },
2331
- {
2332
- "epoch": 29.85,
2333
- "learning_rate": 0.0,
2334
- "loss": 0.0628,
2335
- "step": 2910
2336
- },
2337
- {
2338
- "epoch": 29.85,
2339
- "eval_accuracy": {
2340
- "accuracy": 0.9814102564102564
2341
- },
2342
- "eval_f1": {
2343
- "f1": 0.9820040000167444
2344
- },
2345
- "eval_loss": 0.05972651019692421,
2346
- "eval_precision": {
2347
- "precision": 0.9821559799758683
2348
- },
2349
- "eval_recall": {
2350
- "recall": 0.9818903265571605
2351
- },
2352
- "eval_runtime": 18.3177,
2353
- "eval_samples_per_second": 170.327,
2354
- "eval_steps_per_second": 5.35,
2355
- "step": 2910
2356
- }
2357
- ],
2358
- "logging_steps": 10,
2359
- "max_steps": 2910,
2360
- "num_input_tokens_seen": 0,
2361
- "num_train_epochs": 30,
2362
- "save_steps": 500,
2363
- "total_flos": 9.258876179133235e+18,
2364
- "train_batch_size": 32,
2365
- "trial_name": null,
2366
- "trial_params": null
2367
- }