wuff-mann commited on
Commit
f27f6f2
·
verified ·
1 Parent(s): f48fad7

Upload logs/history.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. logs/history.json +233 -1203
logs/history.json CHANGED
@@ -3,1300 +3,330 @@
3
  "epoch": 1,
4
  "beta": 1.2e-05,
5
  "lr": 0.0002997,
6
- "train_loss": 0.080518,
7
- "train_recon": 0.079909,
8
- "train_kl": 50.7819,
9
- "val_loss": 0.080803,
10
- "val_recon": 0.053361,
11
- "val_kl": 91.4745,
12
- "active_dims": 48,
13
  "active_per_slot": [
 
 
 
14
  16,
15
  16,
16
  16
17
  ],
18
- "usage_max": 0.3368,
19
- "usage_min": 0.3276,
20
  "usage": [
21
- 0.3276,
22
- 0.3356,
23
- 0.3368
 
 
 
24
  ],
25
- "grad_norm": 0.163,
26
- "sec": 72.7
 
27
  },
28
  {
29
  "epoch": 2,
30
  "beta": 2.4e-05,
31
  "lr": 0.0002988,
32
- "train_loss": 0.050594,
33
- "train_recon": 0.048677,
34
- "train_kl": 79.885,
35
- "val_loss": 0.067565,
36
- "val_recon": 0.043665,
37
- "val_kl": 79.6684,
38
- "active_dims": 48,
39
  "active_per_slot": [
 
 
 
40
  16,
41
  16,
42
  16
43
  ],
44
- "usage_max": 0.3352,
45
- "usage_min": 0.3322,
46
  "usage": [
47
- 0.3352,
48
- 0.3322,
49
- 0.3325
 
 
 
50
  ],
51
- "grad_norm": 0.17,
52
- "sec": 72.9
 
53
  },
54
  {
55
  "epoch": 3,
56
  "beta": 3.6e-05,
57
  "lr": 0.0002974,
58
- "train_loss": 0.045462,
59
- "train_recon": 0.043025,
60
- "train_kl": 67.691,
61
- "val_loss": 0.059956,
62
- "val_recon": 0.041135,
63
- "val_kl": 62.736,
64
- "active_dims": 48,
65
  "active_per_slot": [
 
 
 
66
  16,
67
  16,
68
  16
69
  ],
70
- "usage_max": 0.3365,
71
- "usage_min": 0.3306,
72
  "usage": [
73
- 0.3306,
74
- 0.3365,
75
- 0.3329
 
 
 
76
  ],
77
- "grad_norm": 0.159,
78
- "sec": 74.6
 
79
  },
80
  {
81
  "epoch": 4,
82
  "beta": 4.8e-05,
83
  "lr": 0.0002953,
84
- "train_loss": 0.043859,
85
- "train_recon": 0.041226,
86
- "train_kl": 54.8556,
87
- "val_loss": 0.055058,
88
- "val_recon": 0.039352,
89
- "val_kl": 52.3535,
90
- "active_dims": 48,
91
  "active_per_slot": [
92
- 16,
93
- 16,
94
- 16
 
 
 
95
  ],
96
- "usage_max": 0.3343,
97
- "usage_min": 0.3319,
98
  "usage": [
99
- 0.3343,
100
- 0.3319,
101
- 0.3337
 
 
 
102
  ],
103
- "grad_norm": 0.156,
104
- "sec": 75.8
 
105
  },
106
  {
107
  "epoch": 5,
108
  "beta": 6e-05,
109
  "lr": 0.0002927,
110
- "train_loss": 0.042942,
111
- "train_recon": 0.040082,
112
- "train_kl": 47.6777,
113
- "val_loss": 0.052435,
114
- "val_recon": 0.038535,
115
- "val_kl": 46.3312,
116
- "active_dims": 48,
117
  "active_per_slot": [
118
- 16,
119
- 16,
120
- 16
 
 
 
121
  ],
122
- "usage_max": 0.3393,
123
- "usage_min": 0.3298,
124
  "usage": [
125
- 0.3309,
126
- 0.3298,
127
- 0.3393
 
 
 
128
  ],
129
- "grad_norm": 0.15,
130
- "sec": 76.2
 
131
  },
132
  {
133
  "epoch": 6,
134
  "beta": 7.2e-05,
135
  "lr": 0.0002895,
136
- "train_loss": 0.042367,
137
- "train_recon": 0.039275,
138
- "train_kl": 42.9507,
139
- "val_loss": 0.05137,
140
- "val_recon": 0.038831,
141
- "val_kl": 41.7972,
142
- "active_dims": 45,
143
- "active_per_slot": [
144
- 15,
145
- 15,
146
- 15
147
- ],
148
- "usage_max": 0.3342,
149
- "usage_min": 0.3318,
150
- "usage": [
151
- 0.3342,
152
- 0.3318,
153
- 0.334
154
- ],
155
- "grad_norm": 0.152,
156
- "sec": 76.3
 
 
 
 
 
 
 
157
  },
158
  {
159
  "epoch": 7,
160
  "beta": 8.4e-05,
161
  "lr": 0.0002858,
162
- "train_loss": 0.041705,
163
- "train_recon": 0.038374,
164
- "train_kl": 39.6486,
165
- "val_loss": 0.049004,
166
- "val_recon": 0.037198,
167
- "val_kl": 39.3539,
168
- "active_dims": 39,
169
- "active_per_slot": [
170
- 13,
171
- 13,
172
- 13
173
- ],
174
- "usage_max": 0.3384,
175
- "usage_min": 0.3299,
176
- "usage": [
177
- 0.3299,
178
- 0.3384,
179
- 0.3317
180
- ],
181
- "grad_norm": 0.152,
182
- "sec": 76.3
 
 
 
 
 
 
 
183
  },
184
  {
185
  "epoch": 8,
186
  "beta": 9.6e-05,
187
  "lr": 0.0002815,
188
- "train_loss": 0.041342,
189
- "train_recon": 0.037728,
190
- "train_kl": 37.6374,
191
- "val_loss": 0.047678,
192
- "val_recon": 0.036624,
193
- "val_kl": 36.8492,
194
- "active_dims": 27,
195
  "active_per_slot": [
 
 
 
196
  9,
197
  9,
198
  9
199
  ],
200
- "usage_max": 0.3353,
201
- "usage_min": 0.332,
202
  "usage": [
203
- 0.332,
204
- 0.3353,
205
- 0.3327
 
 
 
206
  ],
207
- "grad_norm": 0.158,
208
- "sec": 76.3
 
209
  },
210
  {
211
  "epoch": 9,
212
  "beta": 0.000108,
213
  "lr": 0.0002767,
214
- "train_loss": 0.041069,
215
- "train_recon": 0.037139,
216
- "train_kl": 36.3906,
217
- "val_loss": 0.046538,
218
- "val_recon": 0.035567,
219
- "val_kl": 36.57,
220
- "active_dims": 24,
221
- "active_per_slot": [
222
- 8,
223
- 8,
224
- 8
225
- ],
226
- "usage_max": 0.3362,
227
- "usage_min": 0.3307,
228
- "usage": [
229
- 0.3362,
230
- 0.333,
231
- 0.3307
232
- ],
233
- "grad_norm": 0.155,
234
- "sec": 76.4
 
 
 
 
 
 
 
235
  },
236
  {
237
  "epoch": 10,
238
  "beta": 0.00012,
239
  "lr": 0.0002714,
240
- "train_loss": 0.041053,
241
- "train_recon": 0.03683,
242
- "train_kl": 35.1953,
243
- "val_loss": 0.045933,
244
- "val_recon": 0.035353,
245
- "val_kl": 35.2668,
246
- "active_dims": 18,
247
- "active_per_slot": [
248
- 6,
249
- 6,
250
- 6
251
- ],
252
- "usage_max": 0.3347,
253
- "usage_min": 0.3312,
254
- "usage": [
255
- 0.3312,
256
- 0.334,
257
- 0.3347
258
- ],
259
- "grad_norm": 0.159,
260
- "sec": 76.2
261
- },
262
- {
263
- "epoch": 11,
264
- "beta": 0.000132,
265
- "lr": 0.0002657,
266
- "train_loss": 0.04284,
267
- "train_recon": 0.038392,
268
- "train_kl": 33.6972,
269
- "val_loss": 0.049249,
270
- "val_recon": 0.038784,
271
- "val_kl": 34.8832,
272
- "active_dims": 18,
273
- "active_per_slot": [
274
- 6,
275
- 6,
276
- 6
277
- ],
278
- "usage_max": 0.3351,
279
- "usage_min": 0.3319,
280
- "usage": [
281
- 0.3351,
282
- 0.333,
283
- 0.3319
284
- ],
285
- "grad_norm": 0.171,
286
- "sec": 76.4
287
- },
288
- {
289
- "epoch": 12,
290
- "beta": 0.000144,
291
- "lr": 0.0002595,
292
- "train_loss": 0.04225,
293
- "train_recon": 0.037525,
294
- "train_kl": 32.8078,
295
- "val_loss": 0.045718,
296
- "val_recon": 0.035839,
297
- "val_kl": 32.9317,
298
- "active_dims": 15,
299
- "active_per_slot": [
300
- 5,
301
- 5,
302
- 5
303
- ],
304
- "usage_max": 0.3386,
305
- "usage_min": 0.3305,
306
- "usage": [
307
- 0.3309,
308
- 0.3305,
309
- 0.3386
310
- ],
311
- "grad_norm": 0.164,
312
- "sec": 76.3
313
- },
314
- {
315
- "epoch": 13,
316
- "beta": 0.000156,
317
- "lr": 0.0002528,
318
- "train_loss": 0.04128,
319
- "train_recon": 0.036243,
320
- "train_kl": 32.2887,
321
- "val_loss": 0.04463,
322
- "val_recon": 0.034931,
323
- "val_kl": 32.3313,
324
- "active_dims": 15,
325
- "active_per_slot": [
326
- 5,
327
- 5,
328
- 5
329
- ],
330
- "usage_max": 0.3363,
331
- "usage_min": 0.3315,
332
- "usage": [
333
- 0.3322,
334
- 0.3315,
335
- 0.3363
336
- ],
337
- "grad_norm": 0.156,
338
- "sec": 76.3
339
- },
340
- {
341
- "epoch": 14,
342
- "beta": 0.000168,
343
- "lr": 0.0002458,
344
- "train_loss": 0.041259,
345
- "train_recon": 0.03598,
346
- "train_kl": 31.422,
347
- "val_loss": 0.043682,
348
- "val_recon": 0.034225,
349
- "val_kl": 31.5214,
350
- "active_dims": 15,
351
- "active_per_slot": [
352
- 5,
353
- 5,
354
- 5
355
- ],
356
- "usage_max": 0.3367,
357
- "usage_min": 0.3313,
358
- "usage": [
359
- 0.332,
360
- 0.3313,
361
- 0.3367
362
- ],
363
- "grad_norm": 0.152,
364
- "sec": 76.4
365
- },
366
- {
367
- "epoch": 15,
368
- "beta": 0.00018,
369
- "lr": 0.0002384,
370
- "train_loss": 0.041078,
371
- "train_recon": 0.035554,
372
- "train_kl": 30.6929,
373
- "val_loss": 0.043337,
374
- "val_recon": 0.034313,
375
- "val_kl": 30.0813,
376
- "active_dims": 15,
377
- "active_per_slot": [
378
- 5,
379
- 5,
380
- 5
381
- ],
382
- "usage_max": 0.3361,
383
- "usage_min": 0.3289,
384
- "usage": [
385
- 0.335,
386
- 0.3361,
387
- 0.3289
388
- ],
389
- "grad_norm": 0.153,
390
- "sec": 76.2
391
- },
392
- {
393
- "epoch": 16,
394
- "beta": 0.000192,
395
- "lr": 0.0002306,
396
- "train_loss": 0.041467,
397
- "train_recon": 0.035776,
398
- "train_kl": 29.6391,
399
- "val_loss": 0.042472,
400
- "val_recon": 0.033582,
401
- "val_kl": 29.6334,
402
- "active_dims": 15,
403
- "active_per_slot": [
404
- 5,
405
- 5,
406
- 5
407
- ],
408
- "usage_max": 0.3352,
409
- "usage_min": 0.3321,
410
- "usage": [
411
- 0.3327,
412
- 0.3352,
413
- 0.3321
414
- ],
415
- "grad_norm": 0.163,
416
- "sec": 76.4
417
- },
418
- {
419
- "epoch": 17,
420
- "beta": 0.000204,
421
- "lr": 0.0002225,
422
- "train_loss": 0.04141,
423
- "train_recon": 0.035644,
424
- "train_kl": 28.2652,
425
- "val_loss": 0.041835,
426
- "val_recon": 0.033433,
427
- "val_kl": 28.0079,
428
- "active_dims": 15,
429
  "active_per_slot": [
430
  5,
431
- 5,
432
- 5
433
- ],
434
- "usage_max": 0.3344,
435
- "usage_min": 0.3327,
436
- "usage": [
437
- 0.3344,
438
- 0.3327,
439
- 0.3329
440
- ],
441
- "grad_norm": 0.168,
442
- "sec": 76.3
443
- },
444
- {
445
- "epoch": 18,
446
- "beta": 0.000216,
447
- "lr": 0.0002142,
448
- "train_loss": 0.041097,
449
- "train_recon": 0.035241,
450
- "train_kl": 27.1111,
451
- "val_loss": 0.041358,
452
- "val_recon": 0.033374,
453
- "val_kl": 26.6155,
454
- "active_dims": 15,
455
- "active_per_slot": [
456
  5,
457
  5,
458
- 5
459
- ],
460
- "usage_max": 0.3359,
461
- "usage_min": 0.3316,
462
- "usage": [
463
- 0.3359,
464
- 0.3325,
465
- 0.3316
466
- ],
467
- "grad_norm": 0.161,
468
- "sec": 76.4
469
- },
470
- {
471
- "epoch": 19,
472
- "beta": 0.000228,
473
- "lr": 0.0002055,
474
- "train_loss": 0.040617,
475
- "train_recon": 0.034562,
476
- "train_kl": 26.5581,
477
- "val_loss": 0.041156,
478
- "val_recon": 0.033222,
479
- "val_kl": 26.4477,
480
- "active_dims": 15,
481
- "active_per_slot": [
482
  5,
483
  5,
484
  5
485
  ],
486
- "usage_max": 0.3391,
487
- "usage_min": 0.3303,
488
- "usage": [
489
- 0.3303,
490
- 0.3305,
491
- 0.3391
492
- ],
493
- "grad_norm": 0.153,
494
- "sec": 76.3
495
- },
496
- {
497
- "epoch": 20,
498
- "beta": 0.00024,
499
- "lr": 0.0001967,
500
- "train_loss": 0.041227,
501
- "train_recon": 0.034961,
502
- "train_kl": 26.1056,
503
- "val_loss": 0.040926,
504
- "val_recon": 0.033059,
505
- "val_kl": 26.2245,
506
- "active_dims": 15,
507
- "active_per_slot": [
508
- 5,
509
- 5,
510
- 5
511
- ],
512
- "usage_max": 0.3358,
513
- "usage_min": 0.3307,
514
- "usage": [
515
- 0.3335,
516
- 0.3307,
517
- 0.3358
518
- ],
519
- "grad_norm": 0.18,
520
- "sec": 76.4
521
- },
522
- {
523
- "epoch": 21,
524
- "beta": 0.000252,
525
- "lr": 0.0001877,
526
- "train_loss": 0.041545,
527
- "train_recon": 0.035107,
528
- "train_kl": 25.5483,
529
- "val_loss": 0.041037,
530
- "val_recon": 0.033349,
531
- "val_kl": 25.629,
532
- "active_dims": 15,
533
- "active_per_slot": [
534
- 5,
535
- 5,
536
- 5
537
- ],
538
- "usage_max": 0.3362,
539
- "usage_min": 0.3307,
540
- "usage": [
541
- 0.3362,
542
- 0.3307,
543
- 0.3331
544
- ],
545
- "grad_norm": 0.163,
546
- "sec": 76.4
547
- },
548
- {
549
- "epoch": 22,
550
- "beta": 0.000264,
551
- "lr": 0.0001785,
552
- "train_loss": 0.041687,
553
- "train_recon": 0.035005,
554
- "train_kl": 25.3086,
555
- "val_loss": 0.041052,
556
- "val_recon": 0.033407,
557
- "val_kl": 25.4829,
558
- "active_dims": 15,
559
- "active_per_slot": [
560
- 5,
561
- 5,
562
- 5
563
- ],
564
- "usage_max": 0.334,
565
- "usage_min": 0.3327,
566
- "usage": [
567
- 0.3327,
568
- 0.3333,
569
- 0.334
570
- ],
571
- "grad_norm": 0.173,
572
- "sec": 76.4
573
- },
574
- {
575
- "epoch": 23,
576
- "beta": 0.000276,
577
- "lr": 0.0001692,
578
- "train_loss": 0.04165,
579
- "train_recon": 0.034746,
580
- "train_kl": 25.0122,
581
- "val_loss": 0.040556,
582
- "val_recon": 0.033062,
583
- "val_kl": 24.9796,
584
- "active_dims": 12,
585
- "active_per_slot": [
586
- 4,
587
- 4,
588
- 4
589
- ],
590
- "usage_max": 0.3351,
591
- "usage_min": 0.3318,
592
- "usage": [
593
- 0.3318,
594
- 0.3331,
595
- 0.3351
596
- ],
597
- "grad_norm": 0.158,
598
- "sec": 76.3
599
- },
600
- {
601
- "epoch": 24,
602
- "beta": 0.000288,
603
- "lr": 0.0001599,
604
- "train_loss": 0.041886,
605
- "train_recon": 0.034761,
606
- "train_kl": 24.7398,
607
- "val_loss": 0.040211,
608
- "val_recon": 0.032709,
609
- "val_kl": 25.0077,
610
- "active_dims": 12,
611
- "active_per_slot": [
612
- 4,
613
- 4,
614
- 4
615
- ],
616
- "usage_max": 0.3353,
617
- "usage_min": 0.3299,
618
  "usage": [
619
- 0.3348,
620
- 0.3353,
621
- 0.3299
622
- ],
623
- "grad_norm": 0.179,
624
- "sec": 76.3
625
- },
626
- {
627
- "epoch": 25,
628
- "beta": 0.0003,
629
- "lr": 0.0001505,
630
- "train_loss": 0.041811,
631
- "train_recon": 0.034457,
632
- "train_kl": 24.513,
633
- "val_loss": 0.040276,
634
- "val_recon": 0.032941,
635
- "val_kl": 24.4481,
636
- "active_dims": 12,
637
- "active_per_slot": [
638
- 4,
639
- 4,
640
- 4
641
- ],
642
- "usage_max": 0.3366,
643
- "usage_min": 0.3288,
644
- "usage": [
645
- 0.3288,
646
- 0.3346,
647
- 0.3366
648
- ],
649
- "grad_norm": 0.161,
650
- "sec": 76.3
651
- },
652
- {
653
- "epoch": 26,
654
- "beta": 0.0003,
655
- "lr": 0.0001411,
656
- "train_loss": 0.041752,
657
- "train_recon": 0.034418,
658
- "train_kl": 24.447,
659
- "val_loss": 0.039923,
660
- "val_recon": 0.032614,
661
- "val_kl": 24.3621,
662
- "active_dims": 12,
663
- "active_per_slot": [
664
- 4,
665
- 4,
666
- 4
667
- ],
668
- "usage_max": 0.3342,
669
- "usage_min": 0.3326,
670
- "usage": [
671
- 0.3326,
672
- 0.3332,
673
- 0.3342
674
- ],
675
- "grad_norm": 0.171,
676
- "sec": 76.4
677
- },
678
- {
679
- "epoch": 27,
680
- "beta": 0.0003,
681
- "lr": 0.0001318,
682
- "train_loss": 0.041613,
683
- "train_recon": 0.034288,
684
- "train_kl": 24.4168,
685
- "val_loss": 0.04007,
686
- "val_recon": 0.032646,
687
- "val_kl": 24.7462,
688
- "active_dims": 12,
689
- "active_per_slot": [
690
- 4,
691
- 4,
692
- 4
693
- ],
694
- "usage_max": 0.3364,
695
- "usage_min": 0.33,
696
- "usage": [
697
- 0.3336,
698
- 0.3364,
699
- 0.33
700
- ],
701
- "grad_norm": 0.173,
702
- "sec": 76.2
703
- },
704
- {
705
- "epoch": 28,
706
- "beta": 0.0003,
707
- "lr": 0.0001225,
708
- "train_loss": 0.041215,
709
- "train_recon": 0.033873,
710
- "train_kl": 24.4724,
711
- "val_loss": 0.039428,
712
- "val_recon": 0.032019,
713
- "val_kl": 24.6982,
714
- "active_dims": 12,
715
- "active_per_slot": [
716
- 4,
717
- 4,
718
- 4
719
- ],
720
- "usage_max": 0.3339,
721
- "usage_min": 0.3327,
722
- "usage": [
723
- 0.3327,
724
- 0.3339,
725
- 0.3334
726
- ],
727
- "grad_norm": 0.15,
728
- "sec": 76.2
729
- },
730
- {
731
- "epoch": 29,
732
- "beta": 0.0003,
733
- "lr": 0.0001133,
734
- "train_loss": 0.041672,
735
- "train_recon": 0.034353,
736
- "train_kl": 24.3948,
737
- "val_loss": 0.039853,
738
- "val_recon": 0.032617,
739
- "val_kl": 24.118,
740
- "active_dims": 12,
741
- "active_per_slot": [
742
- 4,
743
- 4,
744
- 4
745
- ],
746
- "usage_max": 0.3369,
747
- "usage_min": 0.3308,
748
- "usage": [
749
- 0.3308,
750
- 0.3323,
751
- 0.3369
752
- ],
753
- "grad_norm": 0.239,
754
- "sec": 76.2
755
- },
756
- {
757
- "epoch": 30,
758
- "beta": 0.0003,
759
- "lr": 0.0001043,
760
- "train_loss": 0.041163,
761
- "train_recon": 0.033864,
762
- "train_kl": 24.3285,
763
- "val_loss": 0.03923,
764
- "val_recon": 0.031955,
765
- "val_kl": 24.2529,
766
- "active_dims": 12,
767
- "active_per_slot": [
768
- 4,
769
- 4,
770
- 4
771
- ],
772
- "usage_max": 0.3347,
773
- "usage_min": 0.3315,
774
- "usage": [
775
- 0.3315,
776
- 0.3347,
777
- 0.3339
778
- ],
779
- "grad_norm": 0.17,
780
- "sec": 76.3
781
- },
782
- {
783
- "epoch": 31,
784
- "beta": 0.0003,
785
- "lr": 9.55e-05,
786
- "train_loss": 0.041065,
787
- "train_recon": 0.033755,
788
- "train_kl": 24.3662,
789
- "val_loss": 0.039105,
790
- "val_recon": 0.031836,
791
- "val_kl": 24.2311,
792
- "active_dims": 12,
793
- "active_per_slot": [
794
- 4,
795
- 4,
796
- 4
797
- ],
798
- "usage_max": 0.3355,
799
- "usage_min": 0.3308,
800
- "usage": [
801
- 0.3337,
802
- 0.3355,
803
- 0.3308
804
- ],
805
- "grad_norm": 0.143,
806
- "sec": 76.4
807
- },
808
- {
809
- "epoch": 32,
810
- "beta": 0.0003,
811
- "lr": 8.68e-05,
812
- "train_loss": 0.041097,
813
- "train_recon": 0.03376,
814
- "train_kl": 24.4566,
815
- "val_loss": 0.039323,
816
- "val_recon": 0.031966,
817
- "val_kl": 24.5241,
818
- "active_dims": 12,
819
- "active_per_slot": [
820
- 4,
821
- 4,
822
- 4
823
- ],
824
- "usage_max": 0.3371,
825
- "usage_min": 0.3299,
826
- "usage": [
827
- 0.3371,
828
- 0.333,
829
- 0.3299
830
- ],
831
- "grad_norm": 0.172,
832
- "sec": 76.3
833
- },
834
- {
835
- "epoch": 33,
836
- "beta": 0.0003,
837
- "lr": 7.85e-05,
838
- "train_loss": 0.04074,
839
- "train_recon": 0.03338,
840
- "train_kl": 24.5343,
841
- "val_loss": 0.039178,
842
- "val_recon": 0.03189,
843
- "val_kl": 24.295,
844
- "active_dims": 12,
845
- "active_per_slot": [
846
- 4,
847
- 4,
848
- 4
849
- ],
850
- "usage_max": 0.3344,
851
- "usage_min": 0.3323,
852
- "usage": [
853
- 0.3344,
854
- 0.3333,
855
- 0.3323
856
- ],
857
- "grad_norm": 0.158,
858
- "sec": 76.3
859
- },
860
- {
861
- "epoch": 34,
862
- "beta": 0.0003,
863
- "lr": 7.04e-05,
864
- "train_loss": 0.040739,
865
- "train_recon": 0.033378,
866
- "train_kl": 24.5378,
867
- "val_loss": 0.039307,
868
- "val_recon": 0.032037,
869
- "val_kl": 24.232,
870
- "active_dims": 12,
871
- "active_per_slot": [
872
- 4,
873
- 4,
874
- 4
875
- ],
876
- "usage_max": 0.3346,
877
- "usage_min": 0.3324,
878
- "usage": [
879
- 0.333,
880
- 0.3346,
881
- 0.3324
882
- ],
883
- "grad_norm": 0.162,
884
- "sec": 76.3
885
- },
886
- {
887
- "epoch": 35,
888
- "beta": 0.0003,
889
- "lr": 6.26e-05,
890
- "train_loss": 0.040503,
891
- "train_recon": 0.033156,
892
- "train_kl": 24.4894,
893
- "val_loss": 0.038795,
894
- "val_recon": 0.031507,
895
- "val_kl": 24.2947,
896
- "active_dims": 12,
897
- "active_per_slot": [
898
- 4,
899
- 4,
900
- 4
901
- ],
902
- "usage_max": 0.3368,
903
- "usage_min": 0.3308,
904
- "usage": [
905
- 0.3308,
906
- 0.3324,
907
- 0.3368
908
- ],
909
- "grad_norm": 0.135,
910
- "sec": 76.3
911
- },
912
- {
913
- "epoch": 36,
914
- "beta": 0.0003,
915
- "lr": 5.52e-05,
916
- "train_loss": 0.040407,
917
- "train_recon": 0.033044,
918
- "train_kl": 24.5425,
919
- "val_loss": 0.039208,
920
- "val_recon": 0.031802,
921
- "val_kl": 24.6873,
922
- "active_dims": 12,
923
- "active_per_slot": [
924
- 4,
925
- 4,
926
- 4
927
- ],
928
- "usage_max": 0.3352,
929
- "usage_min": 0.3305,
930
- "usage": [
931
- 0.3305,
932
- 0.3352,
933
- 0.3344
934
- ],
935
- "grad_norm": 0.162,
936
- "sec": 76.3
937
- },
938
- {
939
- "epoch": 37,
940
- "beta": 0.0003,
941
- "lr": 4.82e-05,
942
- "train_loss": 0.040371,
943
- "train_recon": 0.033021,
944
- "train_kl": 24.5023,
945
- "val_loss": 0.038672,
946
- "val_recon": 0.031364,
947
- "val_kl": 24.357,
948
- "active_dims": 12,
949
- "active_per_slot": [
950
- 4,
951
- 4,
952
- 4
953
- ],
954
- "usage_max": 0.3353,
955
- "usage_min": 0.3312,
956
- "usage": [
957
- 0.3335,
958
- 0.3312,
959
- 0.3353
960
- ],
961
- "grad_norm": 0.155,
962
- "sec": 76.3
963
- },
964
- {
965
- "epoch": 38,
966
- "beta": 0.0003,
967
- "lr": 4.15e-05,
968
- "train_loss": 0.0403,
969
- "train_recon": 0.032937,
970
- "train_kl": 24.5434,
971
- "val_loss": 0.03885,
972
- "val_recon": 0.031494,
973
- "val_kl": 24.5214,
974
- "active_dims": 12,
975
- "active_per_slot": [
976
- 4,
977
- 4,
978
- 4
979
- ],
980
- "usage_max": 0.3388,
981
- "usage_min": 0.3301,
982
- "usage": [
983
- 0.3301,
984
- 0.3388,
985
- 0.3311
986
- ],
987
- "grad_norm": 0.188,
988
- "sec": 76.3
989
- },
990
- {
991
- "epoch": 39,
992
- "beta": 0.0003,
993
- "lr": 3.53e-05,
994
- "train_loss": 0.040317,
995
- "train_recon": 0.03296,
996
- "train_kl": 24.5236,
997
- "val_loss": 0.038564,
998
- "val_recon": 0.03124,
999
- "val_kl": 24.4149,
1000
- "active_dims": 12,
1001
- "active_per_slot": [
1002
- 4,
1003
- 4,
1004
- 4
1005
- ],
1006
- "usage_max": 0.337,
1007
- "usage_min": 0.3315,
1008
- "usage": [
1009
- 0.337,
1010
- 0.3316,
1011
- 0.3315
1012
- ],
1013
- "grad_norm": 0.172,
1014
- "sec": 76.4
1015
- },
1016
- {
1017
- "epoch": 40,
1018
- "beta": 0.0003,
1019
- "lr": 2.96e-05,
1020
- "train_loss": 0.040195,
1021
- "train_recon": 0.032811,
1022
- "train_kl": 24.6104,
1023
- "val_loss": 0.038846,
1024
- "val_recon": 0.031379,
1025
- "val_kl": 24.8893,
1026
- "active_dims": 12,
1027
- "active_per_slot": [
1028
- 4,
1029
- 4,
1030
- 4
1031
- ],
1032
- "usage_max": 0.335,
1033
- "usage_min": 0.3314,
1034
- "usage": [
1035
- 0.335,
1036
- 0.3314,
1037
- 0.3336
1038
- ],
1039
- "grad_norm": 0.195,
1040
- "sec": 76.3
1041
- },
1042
- {
1043
- "epoch": 41,
1044
- "beta": 0.0003,
1045
- "lr": 2.43e-05,
1046
- "train_loss": 0.040033,
1047
- "train_recon": 0.032655,
1048
- "train_kl": 24.5927,
1049
- "val_loss": 0.038497,
1050
- "val_recon": 0.031199,
1051
- "val_kl": 24.3254,
1052
- "active_dims": 12,
1053
- "active_per_slot": [
1054
- 4,
1055
- 4,
1056
- 4
1057
- ],
1058
- "usage_max": 0.3343,
1059
- "usage_min": 0.3318,
1060
- "usage": [
1061
- 0.3318,
1062
- 0.3339,
1063
- 0.3343
1064
- ],
1065
- "grad_norm": 0.164,
1066
- "sec": 76.3
1067
- },
1068
- {
1069
- "epoch": 42,
1070
- "beta": 0.0003,
1071
- "lr": 1.95e-05,
1072
- "train_loss": 0.039939,
1073
- "train_recon": 0.032578,
1074
- "train_kl": 24.5383,
1075
- "val_loss": 0.038448,
1076
- "val_recon": 0.031064,
1077
- "val_kl": 24.6102,
1078
- "active_dims": 12,
1079
- "active_per_slot": [
1080
- 4,
1081
- 4,
1082
- 4
1083
- ],
1084
- "usage_max": 0.3378,
1085
- "usage_min": 0.3303,
1086
- "usage": [
1087
- 0.3378,
1088
- 0.3303,
1089
- 0.3318
1090
- ],
1091
- "grad_norm": 0.148,
1092
- "sec": 76.2
1093
- },
1094
- {
1095
- "epoch": 43,
1096
- "beta": 0.0003,
1097
- "lr": 1.52e-05,
1098
- "train_loss": 0.039851,
1099
- "train_recon": 0.032464,
1100
- "train_kl": 24.6233,
1101
- "val_loss": 0.038465,
1102
- "val_recon": 0.031036,
1103
- "val_kl": 24.7623,
1104
- "active_dims": 12,
1105
- "active_per_slot": [
1106
- 4,
1107
- 4,
1108
- 4
1109
- ],
1110
- "usage_max": 0.3358,
1111
- "usage_min": 0.33,
1112
- "usage": [
1113
- 0.33,
1114
- 0.3342,
1115
- 0.3358
1116
- ],
1117
- "grad_norm": 0.158,
1118
- "sec": 76.4
1119
- },
1120
- {
1121
- "epoch": 44,
1122
- "beta": 0.0003,
1123
- "lr": 1.15e-05,
1124
- "train_loss": 0.039883,
1125
- "train_recon": 0.032494,
1126
- "train_kl": 24.627,
1127
- "val_loss": 0.038375,
1128
- "val_recon": 0.030965,
1129
- "val_kl": 24.7015,
1130
- "active_dims": 12,
1131
- "active_per_slot": [
1132
- 4,
1133
- 4,
1134
- 4
1135
- ],
1136
- "usage_max": 0.3373,
1137
- "usage_min": 0.3302,
1138
- "usage": [
1139
- 0.3325,
1140
- 0.3302,
1141
- 0.3373
1142
- ],
1143
- "grad_norm": 0.194,
1144
- "sec": 76.3
1145
- },
1146
- {
1147
- "epoch": 45,
1148
- "beta": 0.0003,
1149
- "lr": 8.3e-06,
1150
- "train_loss": 0.039779,
1151
- "train_recon": 0.032389,
1152
- "train_kl": 24.6331,
1153
- "val_loss": 0.038241,
1154
- "val_recon": 0.030881,
1155
- "val_kl": 24.5332,
1156
- "active_dims": 12,
1157
- "active_per_slot": [
1158
- 4,
1159
- 4,
1160
- 4
1161
- ],
1162
- "usage_max": 0.3363,
1163
- "usage_min": 0.3313,
1164
- "usage": [
1165
- 0.3363,
1166
- 0.3313,
1167
- 0.3324
1168
  ],
 
1169
  "grad_norm": 0.129,
1170
- "sec": 76.4
1171
- },
1172
- {
1173
- "epoch": 46,
1174
- "beta": 0.0003,
1175
- "lr": 5.7e-06,
1176
- "train_loss": 0.039808,
1177
- "train_recon": 0.0324,
1178
- "train_kl": 24.6945,
1179
- "val_loss": 0.038271,
1180
- "val_recon": 0.030882,
1181
- "val_kl": 24.6279,
1182
- "active_dims": 12,
1183
- "active_per_slot": [
1184
- 4,
1185
- 4,
1186
- 4
1187
- ],
1188
- "usage_max": 0.3353,
1189
- "usage_min": 0.3317,
1190
- "usage": [
1191
- 0.333,
1192
- 0.3317,
1193
- 0.3353
1194
- ],
1195
- "grad_norm": 0.287,
1196
- "sec": 76.2
1197
- },
1198
- {
1199
- "epoch": 47,
1200
- "beta": 0.0003,
1201
- "lr": 3.6e-06,
1202
- "train_loss": 0.039715,
1203
- "train_recon": 0.032329,
1204
- "train_kl": 24.6202,
1205
- "val_loss": 0.038265,
1206
- "val_recon": 0.030881,
1207
- "val_kl": 24.6125,
1208
- "active_dims": 12,
1209
- "active_per_slot": [
1210
- 4,
1211
- 4,
1212
- 4
1213
- ],
1214
- "usage_max": 0.3355,
1215
- "usage_min": 0.3317,
1216
- "usage": [
1217
- 0.3328,
1218
- 0.3355,
1219
- 0.3317
1220
- ],
1221
- "grad_norm": 0.125,
1222
- "sec": 76.3
1223
- },
1224
- {
1225
- "epoch": 48,
1226
- "beta": 0.0003,
1227
- "lr": 2.2e-06,
1228
- "train_loss": 0.039696,
1229
- "train_recon": 0.032302,
1230
- "train_kl": 24.6479,
1231
- "val_loss": 0.03823,
1232
- "val_recon": 0.030834,
1233
- "val_kl": 24.6528,
1234
- "active_dims": 12,
1235
- "active_per_slot": [
1236
- 4,
1237
- 4,
1238
- 4
1239
- ],
1240
- "usage_max": 0.3369,
1241
- "usage_min": 0.331,
1242
- "usage": [
1243
- 0.3369,
1244
- 0.331,
1245
- 0.3322
1246
- ],
1247
- "grad_norm": 0.126,
1248
- "sec": 76.3
1249
- },
1250
- {
1251
- "epoch": 49,
1252
- "beta": 0.0003,
1253
- "lr": 1.3e-06,
1254
- "train_loss": 0.039684,
1255
- "train_recon": 0.032283,
1256
- "train_kl": 24.6717,
1257
- "val_loss": 0.038225,
1258
- "val_recon": 0.030839,
1259
- "val_kl": 24.6194,
1260
- "active_dims": 12,
1261
- "active_per_slot": [
1262
- 4,
1263
- 4,
1264
- 4
1265
- ],
1266
- "usage_max": 0.3378,
1267
- "usage_min": 0.3302,
1268
- "usage": [
1269
- 0.3378,
1270
- 0.3302,
1271
- 0.332
1272
- ],
1273
- "grad_norm": 0.114,
1274
- "sec": 76.4
1275
- },
1276
- {
1277
- "epoch": 50,
1278
- "beta": 0.0003,
1279
- "lr": 1e-06,
1280
- "train_loss": 0.039684,
1281
- "train_recon": 0.032295,
1282
- "train_kl": 24.6304,
1283
- "val_loss": 0.038194,
1284
- "val_recon": 0.030802,
1285
- "val_kl": 24.6384,
1286
- "active_dims": 12,
1287
- "active_per_slot": [
1288
- 4,
1289
- 4,
1290
- 4
1291
- ],
1292
- "usage_max": 0.3344,
1293
- "usage_min": 0.3327,
1294
- "usage": [
1295
- 0.3344,
1296
- 0.3328,
1297
- 0.3327
1298
- ],
1299
- "grad_norm": 0.145,
1300
- "sec": 76.3
1301
  }
1302
  ]
 
3
  "epoch": 1,
4
  "beta": 1.2e-05,
5
  "lr": 0.0002997,
6
+ "train_loss": 0.080587,
7
+ "train_recon": 0.079828,
8
+ "train_kl": 63.2749,
9
+ "val_loss": 0.089129,
10
+ "val_recon": 0.055516,
11
+ "val_kl": 112.0422,
12
+ "active_dims": 96,
13
  "active_per_slot": [
14
+ 16,
15
+ 16,
16
+ 16,
17
  16,
18
  16,
19
  16
20
  ],
21
+ "usage_max": 0.1667,
22
+ "usage_min": 0.1666,
23
  "usage": [
24
+ 0.1667,
25
+ 0.1667,
26
+ 0.1667,
27
+ 0.1667,
28
+ 0.1666,
29
+ 0.1667
30
  ],
31
+ "active_slots": 6,
32
+ "grad_norm": 0.174,
33
+ "sec": 130.9
34
  },
35
  {
36
  "epoch": 2,
37
  "beta": 2.4e-05,
38
  "lr": 0.0002988,
39
+ "train_loss": 0.051385,
40
+ "train_recon": 0.049252,
41
+ "train_kl": 88.8779,
42
+ "val_loss": 0.071474,
43
+ "val_recon": 0.04581,
44
+ "val_kl": 85.5472,
45
+ "active_dims": 96,
46
  "active_per_slot": [
47
+ 16,
48
+ 16,
49
+ 16,
50
  16,
51
  16,
52
  16
53
  ],
54
+ "usage_max": 0.1667,
55
+ "usage_min": 0.1666,
56
  "usage": [
57
+ 0.1667,
58
+ 0.1667,
59
+ 0.1667,
60
+ 0.1667,
61
+ 0.1667,
62
+ 0.1666
63
  ],
64
+ "active_slots": 6,
65
+ "grad_norm": 0.182,
66
+ "sec": 136.3
67
  },
68
  {
69
  "epoch": 3,
70
  "beta": 3.6e-05,
71
  "lr": 0.0002974,
72
+ "train_loss": 0.047323,
73
+ "train_recon": 0.044807,
74
+ "train_kl": 69.889,
75
+ "val_loss": 0.063513,
76
+ "val_recon": 0.043855,
77
+ "val_kl": 65.5257,
78
+ "active_dims": 96,
79
  "active_per_slot": [
80
+ 16,
81
+ 16,
82
+ 16,
83
  16,
84
  16,
85
  16
86
  ],
87
+ "usage_max": 0.1667,
88
+ "usage_min": 0.1666,
89
  "usage": [
90
+ 0.1667,
91
+ 0.1667,
92
+ 0.1667,
93
+ 0.1667,
94
+ 0.1667,
95
+ 0.1666
96
  ],
97
+ "active_slots": 6,
98
+ "grad_norm": 0.16,
99
+ "sec": 137.4
100
  },
101
  {
102
  "epoch": 4,
103
  "beta": 4.8e-05,
104
  "lr": 0.0002953,
105
+ "train_loss": 0.045265,
106
+ "train_recon": 0.042434,
107
+ "train_kl": 58.9728,
108
+ "val_loss": 0.059306,
109
+ "val_recon": 0.041996,
110
+ "val_kl": 57.6993,
111
+ "active_dims": 90,
112
  "active_per_slot": [
113
+ 15,
114
+ 15,
115
+ 15,
116
+ 15,
117
+ 15,
118
+ 15
119
  ],
120
+ "usage_max": 0.1667,
121
+ "usage_min": 0.1667,
122
  "usage": [
123
+ 0.1667,
124
+ 0.1667,
125
+ 0.1667,
126
+ 0.1667,
127
+ 0.1667,
128
+ 0.1667
129
  ],
130
+ "active_slots": 6,
131
+ "grad_norm": 0.16,
132
+ "sec": 137.5
133
  },
134
  {
135
  "epoch": 5,
136
  "beta": 6e-05,
137
  "lr": 0.0002927,
138
+ "train_loss": 0.044393,
139
+ "train_recon": 0.04123,
140
+ "train_kl": 52.7117,
141
+ "val_loss": 0.055317,
142
+ "val_recon": 0.039975,
143
+ "val_kl": 51.1392,
144
+ "active_dims": 90,
145
  "active_per_slot": [
146
+ 15,
147
+ 15,
148
+ 15,
149
+ 15,
150
+ 15,
151
+ 15
152
  ],
153
+ "usage_max": 0.1667,
154
+ "usage_min": 0.1667,
155
  "usage": [
156
+ 0.1667,
157
+ 0.1667,
158
+ 0.1667,
159
+ 0.1667,
160
+ 0.1667,
161
+ 0.1667
162
  ],
163
+ "active_slots": 6,
164
+ "grad_norm": 0.151,
165
+ "sec": 137.0
166
  },
167
  {
168
  "epoch": 6,
169
  "beta": 7.2e-05,
170
  "lr": 0.0002895,
171
+ "train_loss": 0.043934,
172
+ "train_recon": 0.040449,
173
+ "train_kl": 48.409,
174
+ "val_loss": 0.053754,
175
+ "val_recon": 0.03942,
176
+ "val_kl": 47.7803,
177
+ "active_dims": 84,
178
+ "active_per_slot": [
179
+ 14,
180
+ 14,
181
+ 14,
182
+ 14,
183
+ 14,
184
+ 14
185
+ ],
186
+ "usage_max": 0.1667,
187
+ "usage_min": 0.1667,
188
+ "usage": [
189
+ 0.1667,
190
+ 0.1667,
191
+ 0.1667,
192
+ 0.1667,
193
+ 0.1667,
194
+ 0.1667
195
+ ],
196
+ "active_slots": 6,
197
+ "grad_norm": 0.14,
198
+ "sec": 137.1
199
  },
200
  {
201
  "epoch": 7,
202
  "beta": 8.4e-05,
203
  "lr": 0.0002858,
204
+ "train_loss": 0.04374,
205
+ "train_recon": 0.039955,
206
+ "train_kl": 45.0568,
207
+ "val_loss": 0.053639,
208
+ "val_recon": 0.040031,
209
+ "val_kl": 45.3613,
210
+ "active_dims": 66,
211
+ "active_per_slot": [
212
+ 11,
213
+ 11,
214
+ 11,
215
+ 11,
216
+ 11,
217
+ 11
218
+ ],
219
+ "usage_max": 0.1667,
220
+ "usage_min": 0.1667,
221
+ "usage": [
222
+ 0.1667,
223
+ 0.1667,
224
+ 0.1667,
225
+ 0.1667,
226
+ 0.1667,
227
+ 0.1667
228
+ ],
229
+ "active_slots": 6,
230
+ "grad_norm": 0.138,
231
+ "sec": 137.6
232
  },
233
  {
234
  "epoch": 8,
235
  "beta": 9.6e-05,
236
  "lr": 0.0002815,
237
+ "train_loss": 0.043649,
238
+ "train_recon": 0.039552,
239
+ "train_kl": 42.6693,
240
+ "val_loss": 0.051192,
241
+ "val_recon": 0.038312,
242
+ "val_kl": 42.9359,
243
+ "active_dims": 54,
244
  "active_per_slot": [
245
+ 9,
246
+ 9,
247
+ 9,
248
  9,
249
  9,
250
  9
251
  ],
252
+ "usage_max": 0.1667,
253
+ "usage_min": 0.1666,
254
  "usage": [
255
+ 0.1667,
256
+ 0.1667,
257
+ 0.1667,
258
+ 0.1667,
259
+ 0.1667,
260
+ 0.1666
261
  ],
262
+ "active_slots": 6,
263
+ "grad_norm": 0.135,
264
+ "sec": 137.6
265
  },
266
  {
267
  "epoch": 9,
268
  "beta": 0.000108,
269
  "lr": 0.0002767,
270
+ "train_loss": 0.04384,
271
+ "train_recon": 0.039407,
272
+ "train_kl": 41.0424,
273
+ "val_loss": 0.051036,
274
+ "val_recon": 0.03877,
275
+ "val_kl": 40.889,
276
+ "active_dims": 42,
277
+ "active_per_slot": [
278
+ 7,
279
+ 7,
280
+ 7,
281
+ 7,
282
+ 7,
283
+ 7
284
+ ],
285
+ "usage_max": 0.1667,
286
+ "usage_min": 0.1666,
287
+ "usage": [
288
+ 0.1666,
289
+ 0.1667,
290
+ 0.1666,
291
+ 0.1667,
292
+ 0.1667,
293
+ 0.1667
294
+ ],
295
+ "active_slots": 6,
296
+ "grad_norm": 0.135,
297
+ "sec": 137.0
298
  },
299
  {
300
  "epoch": 10,
301
  "beta": 0.00012,
302
  "lr": 0.0002714,
303
+ "train_loss": 0.043939,
304
+ "train_recon": 0.039159,
305
+ "train_kl": 39.8347,
306
+ "val_loss": 0.049877,
307
+ "val_recon": 0.037889,
308
+ "val_kl": 39.9588,
309
+ "active_dims": 30,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
310
  "active_per_slot": [
311
  5,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
312
  5,
313
  5,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
314
  5,
315
  5,
316
  5
317
  ],
318
+ "usage_max": 0.1667,
319
+ "usage_min": 0.1666,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
320
  "usage": [
321
+ 0.1666,
322
+ 0.1667,
323
+ 0.1667,
324
+ 0.1667,
325
+ 0.1667,
326
+ 0.1667
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
327
  ],
328
+ "active_slots": 6,
329
  "grad_norm": 0.129,
330
+ "sec": 137.6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
331
  }
332
  ]