wuff-mann commited on
Commit
9025b8e
·
verified ·
1 Parent(s): 65f6097

Upload logs/history.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. logs/history.json +330 -0
logs/history.json CHANGED
@@ -328,5 +328,335 @@
328
  "active_slots": 6,
329
  "grad_norm": 0.129,
330
  "sec": 137.6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
331
  }
332
  ]
 
328
  "active_slots": 6,
329
  "grad_norm": 0.129,
330
  "sec": 137.6
331
+ },
332
+ {
333
+ "epoch": 11,
334
+ "beta": 0.000132,
335
+ "lr": 0.0002657,
336
+ "train_loss": 0.044122,
337
+ "train_recon": 0.039005,
338
+ "train_kl": 38.7652,
339
+ "val_loss": 0.049256,
340
+ "val_recon": 0.03757,
341
+ "val_kl": 38.9557,
342
+ "active_dims": 30,
343
+ "active_per_slot": [
344
+ 5,
345
+ 5,
346
+ 5,
347
+ 5,
348
+ 5,
349
+ 5
350
+ ],
351
+ "usage_max": 0.1667,
352
+ "usage_min": 0.1667,
353
+ "usage": [
354
+ 0.1667,
355
+ 0.1667,
356
+ 0.1667,
357
+ 0.1667,
358
+ 0.1667,
359
+ 0.1667
360
+ ],
361
+ "active_slots": 6,
362
+ "grad_norm": 0.12,
363
+ "sec": 138.1
364
+ },
365
+ {
366
+ "epoch": 12,
367
+ "beta": 0.000144,
368
+ "lr": 0.0002595,
369
+ "train_loss": 0.044382,
370
+ "train_recon": 0.038939,
371
+ "train_kl": 37.8008,
372
+ "val_loss": 0.048549,
373
+ "val_recon": 0.037244,
374
+ "val_kl": 37.6819,
375
+ "active_dims": 30,
376
+ "active_per_slot": [
377
+ 5,
378
+ 5,
379
+ 5,
380
+ 5,
381
+ 5,
382
+ 5
383
+ ],
384
+ "usage_max": 0.1667,
385
+ "usage_min": 0.1667,
386
+ "usage": [
387
+ 0.1667,
388
+ 0.1667,
389
+ 0.1667,
390
+ 0.1667,
391
+ 0.1667,
392
+ 0.1667
393
+ ],
394
+ "active_slots": 6,
395
+ "grad_norm": 0.121,
396
+ "sec": 138.0
397
+ },
398
+ {
399
+ "epoch": 13,
400
+ "beta": 0.000156,
401
+ "lr": 0.0002528,
402
+ "train_loss": 0.044585,
403
+ "train_recon": 0.038834,
404
+ "train_kl": 36.8619,
405
+ "val_loss": 0.04834,
406
+ "val_recon": 0.03728,
407
+ "val_kl": 36.8646,
408
+ "active_dims": 30,
409
+ "active_per_slot": [
410
+ 5,
411
+ 5,
412
+ 5,
413
+ 5,
414
+ 5,
415
+ 5
416
+ ],
417
+ "usage_max": 0.1667,
418
+ "usage_min": 0.1667,
419
+ "usage": [
420
+ 0.1667,
421
+ 0.1667,
422
+ 0.1667,
423
+ 0.1667,
424
+ 0.1667,
425
+ 0.1667
426
+ ],
427
+ "active_slots": 6,
428
+ "grad_norm": 0.121,
429
+ "sec": 138.6
430
+ },
431
+ {
432
+ "epoch": 14,
433
+ "beta": 0.000168,
434
+ "lr": 0.0002458,
435
+ "train_loss": 0.044787,
436
+ "train_recon": 0.038737,
437
+ "train_kl": 36.0114,
438
+ "val_loss": 0.048191,
439
+ "val_recon": 0.03729,
440
+ "val_kl": 36.3355,
441
+ "active_dims": 30,
442
+ "active_per_slot": [
443
+ 5,
444
+ 5,
445
+ 5,
446
+ 5,
447
+ 5,
448
+ 5
449
+ ],
450
+ "usage_max": 0.1667,
451
+ "usage_min": 0.1667,
452
+ "usage": [
453
+ 0.1667,
454
+ 0.1667,
455
+ 0.1667,
456
+ 0.1667,
457
+ 0.1667,
458
+ 0.1667
459
+ ],
460
+ "active_slots": 6,
461
+ "grad_norm": 0.112,
462
+ "sec": 138.3
463
+ },
464
+ {
465
+ "epoch": 15,
466
+ "beta": 0.00018,
467
+ "lr": 0.0002384,
468
+ "train_loss": 0.044982,
469
+ "train_recon": 0.03861,
470
+ "train_kl": 35.3991,
471
+ "val_loss": 0.047675,
472
+ "val_recon": 0.036956,
473
+ "val_kl": 35.731,
474
+ "active_dims": 30,
475
+ "active_per_slot": [
476
+ 5,
477
+ 5,
478
+ 5,
479
+ 5,
480
+ 5,
481
+ 5
482
+ ],
483
+ "usage_max": 0.1667,
484
+ "usage_min": 0.1667,
485
+ "usage": [
486
+ 0.1667,
487
+ 0.1667,
488
+ 0.1667,
489
+ 0.1667,
490
+ 0.1667,
491
+ 0.1667
492
+ ],
493
+ "active_slots": 6,
494
+ "grad_norm": 0.106,
495
+ "sec": 137.5
496
+ },
497
+ {
498
+ "epoch": 16,
499
+ "beta": 0.000192,
500
+ "lr": 0.0002306,
501
+ "train_loss": 0.045248,
502
+ "train_recon": 0.038584,
503
+ "train_kl": 34.7073,
504
+ "val_loss": 0.047217,
505
+ "val_recon": 0.036878,
506
+ "val_kl": 34.4648,
507
+ "active_dims": 60,
508
+ "active_per_slot": [
509
+ 10,
510
+ 10,
511
+ 10,
512
+ 10,
513
+ 10,
514
+ 10
515
+ ],
516
+ "usage_max": 0.1667,
517
+ "usage_min": 0.1667,
518
+ "usage": [
519
+ 0.1667,
520
+ 0.1667,
521
+ 0.1667,
522
+ 0.1667,
523
+ 0.1667,
524
+ 0.1667
525
+ ],
526
+ "active_slots": 6,
527
+ "grad_norm": 0.106,
528
+ "sec": 137.2
529
+ },
530
+ {
531
+ "epoch": 17,
532
+ "beta": 0.000204,
533
+ "lr": 0.0002225,
534
+ "train_loss": 0.045363,
535
+ "train_recon": 0.038386,
536
+ "train_kl": 34.1997,
537
+ "val_loss": 0.046849,
538
+ "val_recon": 0.036407,
539
+ "val_kl": 34.8051,
540
+ "active_dims": 60,
541
+ "active_per_slot": [
542
+ 10,
543
+ 10,
544
+ 10,
545
+ 10,
546
+ 10,
547
+ 10
548
+ ],
549
+ "usage_max": 0.1667,
550
+ "usage_min": 0.1667,
551
+ "usage": [
552
+ 0.1667,
553
+ 0.1667,
554
+ 0.1667,
555
+ 0.1667,
556
+ 0.1667,
557
+ 0.1667
558
+ ],
559
+ "active_slots": 6,
560
+ "grad_norm": 0.101,
561
+ "sec": 137.2
562
+ },
563
+ {
564
+ "epoch": 18,
565
+ "beta": 0.000216,
566
+ "lr": 0.0002142,
567
+ "train_loss": 0.045662,
568
+ "train_recon": 0.038352,
569
+ "train_kl": 33.8434,
570
+ "val_loss": 0.046294,
571
+ "val_recon": 0.035753,
572
+ "val_kl": 35.138,
573
+ "active_dims": 60,
574
+ "active_per_slot": [
575
+ 10,
576
+ 10,
577
+ 10,
578
+ 10,
579
+ 10,
580
+ 10
581
+ ],
582
+ "usage_max": 0.1667,
583
+ "usage_min": 0.1667,
584
+ "usage": [
585
+ 0.1667,
586
+ 0.1667,
587
+ 0.1667,
588
+ 0.1667,
589
+ 0.1667,
590
+ 0.1667
591
+ ],
592
+ "active_slots": 6,
593
+ "grad_norm": 0.108,
594
+ "sec": 137.2
595
+ },
596
+ {
597
+ "epoch": 19,
598
+ "beta": 0.000228,
599
+ "lr": 0.0002055,
600
+ "train_loss": 0.0458,
601
+ "train_recon": 0.038176,
602
+ "train_kl": 33.4403,
603
+ "val_loss": 0.045988,
604
+ "val_recon": 0.036047,
605
+ "val_kl": 33.1367,
606
+ "active_dims": 60,
607
+ "active_per_slot": [
608
+ 10,
609
+ 10,
610
+ 10,
611
+ 10,
612
+ 10,
613
+ 10
614
+ ],
615
+ "usage_max": 0.1667,
616
+ "usage_min": 0.1667,
617
+ "usage": [
618
+ 0.1667,
619
+ 0.1667,
620
+ 0.1667,
621
+ 0.1667,
622
+ 0.1667,
623
+ 0.1667
624
+ ],
625
+ "active_slots": 6,
626
+ "grad_norm": 0.099,
627
+ "sec": 137.3
628
+ },
629
+ {
630
+ "epoch": 20,
631
+ "beta": 0.00024,
632
+ "lr": 0.0001967,
633
+ "train_loss": 0.046018,
634
+ "train_recon": 0.038115,
635
+ "train_kl": 32.9295,
636
+ "val_loss": 0.046326,
637
+ "val_recon": 0.036254,
638
+ "val_kl": 33.573,
639
+ "active_dims": 60,
640
+ "active_per_slot": [
641
+ 10,
642
+ 10,
643
+ 10,
644
+ 10,
645
+ 10,
646
+ 10
647
+ ],
648
+ "usage_max": 0.1667,
649
+ "usage_min": 0.1667,
650
+ "usage": [
651
+ 0.1667,
652
+ 0.1667,
653
+ 0.1667,
654
+ 0.1667,
655
+ 0.1667,
656
+ 0.1667
657
+ ],
658
+ "active_slots": 6,
659
+ "grad_norm": 0.096,
660
+ "sec": 137.7
661
  }
662
  ]