PowerMachine commited on
Commit
cbb0020
·
verified ·
1 Parent(s): 60e4753

V6.5-7ds cleanup: remove legacy v6_3_training_metrics.json

Browse files
Files changed (1) hide show
  1. v6_3_training_metrics.json +0 -1737
v6_3_training_metrics.json DELETED
@@ -1,1737 +0,0 @@
1
- {
2
- "version": "V6.3",
3
- "steps": [
4
- {
5
- "step": 0,
6
- "epoch": 0,
7
- "dataset": "TucanoBR/GigaVerbo",
8
- "quality": {
9
- "1.1_train_loss": -1.0,
10
- "1.2_train_acc": 1.0,
11
- "1.3_val_loss": -1.0,
12
- "1.4_val_acc": 1.0,
13
- "1.5_perplexity": 0.36787944123356736
14
- },
15
- "speed": {
16
- "2.1_throughput_sps": 1.1736643998581224,
17
- "2.2_step_time_ms": 13632.517099380493,
18
- "2.3_epoch_progress": 0.5,
19
- "2.4_rss_mb": 509.828125,
20
- "2.5_xeon_tflops": 1.408584045428496
21
- },
22
- "kohonen": {
23
- "sigma_t": 1.3846745195799537,
24
- "alpha_t": 0.09607894391523232,
25
- "t": 80,
26
- "n_neurons": 8,
27
- "fisher_w_mean": 0.0,
28
- "has_ewc_reference": false
29
- },
30
- "hypothesis": {
31
- "classifier_trained": false,
32
- "punishment_count": 0,
33
- "success_count": 1,
34
- "training_ready": true,
35
- "buffer_size": 16,
36
- "required_new_samples": 0
37
- }
38
- },
39
- {
40
- "step": 1,
41
- "epoch": 0,
42
- "dataset": "TucanoBR/GigaVerbo",
43
- "quality": {
44
- "1.1_train_loss": -0.9842509842514764,
45
- "1.2_train_acc": 0.96875,
46
- "1.3_val_loss": -0.9842509842514764,
47
- "1.4_val_acc": 0.96875,
48
- "1.5_perplexity": 0.3737190436413242
49
- },
50
- "speed": {
51
- "2.1_throughput_sps": 1.986820563440586,
52
- "2.2_step_time_ms": 8053.067445755005,
53
- "2.3_epoch_progress": 0.5,
54
- "2.4_rss_mb": 564.65234375,
55
- "2.5_xeon_tflops": 1.408584045428496
56
- },
57
- "kohonen": {
58
- "sigma_t": 1.3846745195799537,
59
- "alpha_t": 0.09607894391523232,
60
- "t": 80,
61
- "n_neurons": 8,
62
- "fisher_w_mean": 0.0,
63
- "has_ewc_reference": false
64
- },
65
- "hypothesis": {
66
- "classifier_trained": true,
67
- "punishment_count": 1,
68
- "success_count": 0,
69
- "training_ready": true,
70
- "buffer_size": 32,
71
- "required_new_samples": 0
72
- }
73
- },
74
- {
75
- "step": 2,
76
- "epoch": 0,
77
- "dataset": "TucanoBR/GigaVerbo",
78
- "quality": {
79
- "1.1_train_loss": -1.0,
80
- "1.2_train_acc": 1.0,
81
- "1.3_val_loss": -1.0,
82
- "1.4_val_acc": 1.0,
83
- "1.5_perplexity": 0.36787944123356736
84
- },
85
- "speed": {
86
- "2.1_throughput_sps": 2.973347394980598,
87
- "2.2_step_time_ms": 5381.140470504761,
88
- "2.3_epoch_progress": 0.5,
89
- "2.4_rss_mb": 564.65234375,
90
- "2.5_xeon_tflops": 1.408584045428496
91
- },
92
- "kohonen": {
93
- "sigma_t": 1.3846745195799537,
94
- "alpha_t": 0.09607894391523232,
95
- "t": 80,
96
- "n_neurons": 8,
97
- "fisher_w_mean": 0.09977994114160538,
98
- "has_ewc_reference": true
99
- },
100
- "hypothesis": {
101
- "classifier_trained": true,
102
- "punishment_count": 0,
103
- "success_count": 0,
104
- "training_ready": false,
105
- "buffer_size": 0,
106
- "required_new_samples": 10
107
- }
108
- },
109
- {
110
- "step": 3,
111
- "epoch": 0,
112
- "dataset": "TucanoBR/GigaVerbo",
113
- "quality": {
114
- "1.1_train_loss": -1.0,
115
- "1.2_train_acc": 1.0,
116
- "1.3_val_loss": -1.0,
117
- "1.4_val_acc": 1.0,
118
- "1.5_perplexity": 0.36787944123356736
119
- },
120
- "speed": {
121
- "2.1_throughput_sps": 3.964087453289865,
122
- "2.2_step_time_ms": 4036.237895488739,
123
- "2.3_epoch_progress": 0.5,
124
- "2.4_rss_mb": 564.65234375,
125
- "2.5_xeon_tflops": 1.408584045428496
126
- },
127
- "kohonen": {
128
- "sigma_t": 1.3846745195799537,
129
- "alpha_t": 0.09607894391523232,
130
- "t": 80,
131
- "n_neurons": 8,
132
- "fisher_w_mean": 0.09977994114160538,
133
- "has_ewc_reference": true
134
- },
135
- "hypothesis": {
136
- "classifier_trained": true,
137
- "punishment_count": 0,
138
- "success_count": 0,
139
- "training_ready": false,
140
- "buffer_size": 2,
141
- "required_new_samples": 10
142
- }
143
- },
144
- {
145
- "step": 4,
146
- "epoch": 0,
147
- "dataset": "dominguesm/restore-punctuation-ptbr-dataset",
148
- "quality": {
149
- "1.1_train_loss": -0.97182531580755,
150
- "1.2_train_acc": 0.9444444444444444,
151
- "1.3_val_loss": -0.97182531580755,
152
- "1.4_val_acc": 0.9444444444444444,
153
- "1.5_perplexity": 0.3783917230298394
154
- },
155
- "speed": {
156
- "2.1_throughput_sps": 3.1067598433248054,
157
- "2.2_step_time_ms": 5150.060129165649,
158
- "2.3_epoch_progress": 0.5,
159
- "2.4_rss_mb": 610.62109375,
160
- "2.5_xeon_tflops": 1.408584045428496
161
- },
162
- "kohonen": {
163
- "sigma_t": 1.2654972248945755,
164
- "alpha_t": 0.09185122844014575,
165
- "t": 170,
166
- "n_neurons": 8,
167
- "fisher_w_mean": 0.09977994114160538,
168
- "has_ewc_reference": true
169
- },
170
- "hypothesis": {
171
- "classifier_trained": true,
172
- "punishment_count": 1,
173
- "success_count": 0,
174
- "training_ready": true,
175
- "buffer_size": 18,
176
- "required_new_samples": 10
177
- }
178
- },
179
- {
180
- "step": 5,
181
- "epoch": 0,
182
- "dataset": "dominguesm/restore-punctuation-ptbr-dataset",
183
- "quality": {
184
- "1.1_train_loss": -1.0,
185
- "1.2_train_acc": 1.0,
186
- "1.3_val_loss": -1.0,
187
- "1.4_val_acc": 1.0,
188
- "1.5_perplexity": 0.36787944123356736
189
- },
190
- "speed": {
191
- "2.1_throughput_sps": 3.726793894229844,
192
- "2.2_step_time_ms": 4293.234467506409,
193
- "2.3_epoch_progress": 0.5,
194
- "2.4_rss_mb": 610.62109375,
195
- "2.5_xeon_tflops": 1.408584045428496
196
- },
197
- "kohonen": {
198
- "sigma_t": 1.2654972248945755,
199
- "alpha_t": 0.09185122844014575,
200
- "t": 170,
201
- "n_neurons": 8,
202
- "fisher_w_mean": 0.0,
203
- "has_ewc_reference": true
204
- },
205
- "hypothesis": {
206
- "classifier_trained": true,
207
- "punishment_count": 0,
208
- "success_count": 0,
209
- "training_ready": false,
210
- "buffer_size": 0,
211
- "required_new_samples": 10
212
- }
213
- },
214
- {
215
- "step": 6,
216
- "epoch": 0,
217
- "dataset": "dominguesm/restore-punctuation-ptbr-dataset",
218
- "quality": {
219
- "1.1_train_loss": -1.0,
220
- "1.2_train_acc": 1.0,
221
- "1.3_val_loss": -1.0,
222
- "1.4_val_acc": 1.0,
223
- "1.5_perplexity": 0.36787944123356736
224
- },
225
- "speed": {
226
- "2.1_throughput_sps": 4.3450685684645,
227
- "2.2_step_time_ms": 3682.335444859096,
228
- "2.3_epoch_progress": 0.5,
229
- "2.4_rss_mb": 610.62109375,
230
- "2.5_xeon_tflops": 1.408584045428496
231
- },
232
- "kohonen": {
233
- "sigma_t": 1.1682011746071073,
234
- "alpha_t": 0.08824969025845955,
235
- "t": 250,
236
- "n_neurons": 8,
237
- "fisher_w_mean": 0.0,
238
- "has_ewc_reference": true
239
- },
240
- "hypothesis": {
241
- "classifier_trained": true,
242
- "punishment_count": 0,
243
- "success_count": 1,
244
- "training_ready": true,
245
- "buffer_size": 16,
246
- "required_new_samples": 10
247
- }
248
- },
249
- {
250
- "step": 7,
251
- "epoch": 0,
252
- "dataset": "dominguesm/restore-punctuation-ptbr-dataset",
253
- "quality": {
254
- "1.1_train_loss": -1.0,
255
- "1.2_train_acc": 1.0,
256
- "1.3_val_loss": -1.0,
257
- "1.4_val_acc": 1.0,
258
- "1.5_perplexity": 0.36787944123356736
259
- },
260
- "speed": {
261
- "2.1_throughput_sps": 4.964894674442569,
262
- "2.2_step_time_ms": 3222.626268863678,
263
- "2.3_epoch_progress": 0.5,
264
- "2.4_rss_mb": 610.62109375,
265
- "2.5_xeon_tflops": 1.408584045428496
266
- },
267
- "kohonen": {
268
- "sigma_t": 1.1682011746071073,
269
- "alpha_t": 0.08824969025845955,
270
- "t": 250,
271
- "n_neurons": 8,
272
- "fisher_w_mean": 0.0,
273
- "has_ewc_reference": true
274
- },
275
- "hypothesis": {
276
- "classifier_trained": true,
277
- "punishment_count": 0,
278
- "success_count": 2,
279
- "training_ready": true,
280
- "buffer_size": 18,
281
- "required_new_samples": 10
282
- }
283
- },
284
- {
285
- "step": 8,
286
- "epoch": 0,
287
- "dataset": "Madras1/corpus-ptbr-v2",
288
- "quality": {
289
- "1.1_train_loss": -0.985184366143778,
290
- "1.2_train_acc": 0.9705882352941176,
291
- "1.3_val_loss": -0.985184366143778,
292
- "1.4_val_acc": 0.9705882352941176,
293
- "1.5_perplexity": 0.3733703837949463
294
- },
295
- "speed": {
296
- "2.1_throughput_sps": 3.845537710828318,
297
- "2.2_step_time_ms": 4160.666518741184,
298
- "2.3_epoch_progress": 0.5,
299
- "2.4_rss_mb": 880.08984375,
300
- "2.5_xeon_tflops": 1.408584045428496
301
- },
302
- "kohonen": {
303
- "sigma_t": 1.1682011746071073,
304
- "alpha_t": 0.08824969025845955,
305
- "t": 250,
306
- "n_neurons": 8,
307
- "fisher_w_mean": 0.0,
308
- "has_ewc_reference": true
309
- },
310
- "hypothesis": {
311
- "classifier_trained": true,
312
- "punishment_count": 1,
313
- "success_count": 0,
314
- "training_ready": true,
315
- "buffer_size": 34,
316
- "required_new_samples": 10
317
- }
318
- },
319
- {
320
- "step": 9,
321
- "epoch": 0,
322
- "dataset": "Madras1/corpus-ptbr-v2",
323
- "quality": {
324
- "1.1_train_loss": -1.0,
325
- "1.2_train_acc": 1.0,
326
- "1.3_val_loss": -1.0,
327
- "1.4_val_acc": 1.0,
328
- "1.5_perplexity": 0.36787944123356736
329
- },
330
- "speed": {
331
- "2.1_throughput_sps": 4.271432862755954,
332
- "2.2_step_time_ms": 3745.815634727478,
333
- "2.3_epoch_progress": 0.5,
334
- "2.4_rss_mb": 880.08984375,
335
- "2.5_xeon_tflops": 1.408584045428496
336
- },
337
- "kohonen": {
338
- "sigma_t": 1.1682011746071073,
339
- "alpha_t": 0.08824969025845955,
340
- "t": 250,
341
- "n_neurons": 8,
342
- "fisher_w_mean": 0.0,
343
- "has_ewc_reference": true
344
- },
345
- "hypothesis": {
346
- "classifier_trained": true,
347
- "punishment_count": 0,
348
- "success_count": 0,
349
- "training_ready": false,
350
- "buffer_size": 0,
351
- "required_new_samples": 10
352
- }
353
- },
354
- {
355
- "step": 10,
356
- "epoch": 0,
357
- "dataset": "Madras1/corpus-ptbr-v2",
358
- "quality": {
359
- "1.1_train_loss": -1.0,
360
- "1.2_train_acc": 1.0,
361
- "1.3_val_loss": -1.0,
362
- "1.4_val_acc": 1.0,
363
- "1.5_perplexity": 0.36787944123356736
364
- },
365
- "speed": {
366
- "2.1_throughput_sps": 4.696214305358404,
367
- "2.2_step_time_ms": 3406.9995446638627,
368
- "2.3_epoch_progress": 0.5,
369
- "2.4_rss_mb": 880.08984375,
370
- "2.5_xeon_tflops": 1.408584045428496
371
- },
372
- "kohonen": {
373
- "sigma_t": 1.0783856001478893,
374
- "alpha_t": 0.08478937040879159,
375
- "t": 330,
376
- "n_neurons": 8,
377
- "fisher_w_mean": 0.0,
378
- "has_ewc_reference": true
379
- },
380
- "hypothesis": {
381
- "classifier_trained": true,
382
- "punishment_count": 0,
383
- "success_count": 1,
384
- "training_ready": true,
385
- "buffer_size": 16,
386
- "required_new_samples": 10
387
- }
388
- },
389
- {
390
- "step": 11,
391
- "epoch": 0,
392
- "dataset": "Madras1/corpus-ptbr-v2",
393
- "quality": {
394
- "1.1_train_loss": -0.97182531580755,
395
- "1.2_train_acc": 0.9444444444444444,
396
- "1.3_val_loss": -0.97182531580755,
397
- "1.4_val_acc": 0.9444444444444444,
398
- "1.5_perplexity": 0.3783917230298394
399
- },
400
- "speed": {
401
- "2.1_throughput_sps": 5.10548252403601,
402
- "2.2_step_time_ms": 3133.8859597841897,
403
- "2.3_epoch_progress": 0.5,
404
- "2.4_rss_mb": 880.60546875,
405
- "2.5_xeon_tflops": 1.408584045428496
406
- },
407
- "kohonen": {
408
- "sigma_t": 1.0783856001478893,
409
- "alpha_t": 0.08478937040879159,
410
- "t": 330,
411
- "n_neurons": 8,
412
- "fisher_w_mean": 0.0,
413
- "has_ewc_reference": true
414
- },
415
- "hypothesis": {
416
- "classifier_trained": true,
417
- "punishment_count": 1,
418
- "success_count": 0,
419
- "training_ready": true,
420
- "buffer_size": 18,
421
- "required_new_samples": 10
422
- }
423
- },
424
- {
425
- "step": 12,
426
- "epoch": 0,
427
- "dataset": "CEIA-POSITIVO/ultrachat_br_clustred_balanced_v1",
428
- "quality": {
429
- "1.1_train_loss": -1.0,
430
- "1.2_train_acc": 1.0,
431
- "1.3_val_loss": -1.0,
432
- "1.4_val_acc": 1.0,
433
- "1.5_perplexity": 0.36787944123356736
434
- },
435
- "speed": {
436
- "2.1_throughput_sps": 3.8902537378603372,
437
- "2.2_step_time_ms": 4112.842266376202,
438
- "2.3_epoch_progress": 0.5,
439
- "2.4_rss_mb": 1121.4453125,
440
- "2.5_xeon_tflops": 1.408584045428496
441
- },
442
- "kohonen": {
443
- "sigma_t": 1.0783856001478893,
444
- "alpha_t": 0.08478937040879159,
445
- "t": 330,
446
- "n_neurons": 8,
447
- "fisher_w_mean": 0.0,
448
- "has_ewc_reference": true
449
- },
450
- "hypothesis": {
451
- "classifier_trained": true,
452
- "punishment_count": 0,
453
- "success_count": 0,
454
- "training_ready": false,
455
- "buffer_size": 0,
456
- "required_new_samples": 10
457
- }
458
- },
459
- {
460
- "step": 13,
461
- "epoch": 0,
462
- "dataset": "CEIA-POSITIVO/ultrachat_br_clustred_balanced_v1",
463
- "quality": {
464
- "1.1_train_loss": -1.0,
465
- "1.2_train_acc": 1.0,
466
- "1.3_val_loss": -1.0,
467
- "1.4_val_acc": 1.0,
468
- "1.5_perplexity": 0.36787944123356736
469
- },
470
- "speed": {
471
- "2.1_throughput_sps": 4.187973653467053,
472
- "2.2_step_time_ms": 3820.463384900774,
473
- "2.3_epoch_progress": 0.5,
474
- "2.4_rss_mb": 1121.4453125,
475
- "2.5_xeon_tflops": 1.408584045428496
476
- },
477
- "kohonen": {
478
- "sigma_t": 0.9954753752044792,
479
- "alpha_t": 0.08146473164114146,
480
- "t": 410,
481
- "n_neurons": 8,
482
- "fisher_w_mean": 0.0,
483
- "has_ewc_reference": true
484
- },
485
- "hypothesis": {
486
- "classifier_trained": true,
487
- "punishment_count": 0,
488
- "success_count": 1,
489
- "training_ready": true,
490
- "buffer_size": 16,
491
- "required_new_samples": 10
492
- }
493
- },
494
- {
495
- "step": 14,
496
- "epoch": 0,
497
- "dataset": "CEIA-POSITIVO/ultrachat_br_clustred_balanced_v1",
498
- "quality": {
499
- "1.1_train_loss": -1.0,
500
- "1.2_train_acc": 1.0,
501
- "1.3_val_loss": -1.0,
502
- "1.4_val_acc": 1.0,
503
- "1.5_perplexity": 0.36787944123356736
504
- },
505
- "speed": {
506
- "2.1_throughput_sps": 4.486083466924311,
507
- "2.2_step_time_ms": 3566.5854454040527,
508
- "2.3_epoch_progress": 0.5,
509
- "2.4_rss_mb": 1121.4453125,
510
- "2.5_xeon_tflops": 1.408584045428496
511
- },
512
- "kohonen": {
513
- "sigma_t": 0.9954753752044792,
514
- "alpha_t": 0.08146473164114146,
515
- "t": 410,
516
- "n_neurons": 8,
517
- "fisher_w_mean": 0.0,
518
- "has_ewc_reference": true
519
- },
520
- "hypothesis": {
521
- "classifier_trained": true,
522
- "punishment_count": 0,
523
- "success_count": 2,
524
- "training_ready": true,
525
- "buffer_size": 32,
526
- "required_new_samples": 10
527
- }
528
- },
529
- {
530
- "step": 15,
531
- "epoch": 0,
532
- "dataset": "CEIA-POSITIVO/ultrachat_br_clustred_balanced_v1",
533
- "quality": {
534
- "1.1_train_loss": -1.0,
535
- "1.2_train_acc": 1.0,
536
- "1.3_val_loss": -1.0,
537
- "1.4_val_acc": 1.0,
538
- "1.5_perplexity": 0.36787944123356736
539
- },
540
- "speed": {
541
- "2.1_throughput_sps": 4.784451730184544,
542
- "2.2_step_time_ms": 3344.1658318042755,
543
- "2.3_epoch_progress": 0.5,
544
- "2.4_rss_mb": 1121.4453125,
545
- "2.5_xeon_tflops": 1.408584045428496
546
- },
547
- "kohonen": {
548
- "sigma_t": 0.9954753752044792,
549
- "alpha_t": 0.08146473164114146,
550
- "t": 410,
551
- "n_neurons": 8,
552
- "fisher_w_mean": 0.0,
553
- "has_ewc_reference": true
554
- },
555
- "hypothesis": {
556
- "classifier_trained": true,
557
- "punishment_count": 0,
558
- "success_count": 3,
559
- "training_ready": true,
560
- "buffer_size": 34,
561
- "required_new_samples": 10
562
- }
563
- },
564
- {
565
- "step": 16,
566
- "epoch": 0,
567
- "dataset": "nvidia/OpenMathInstruct-2",
568
- "quality": {
569
- "1.1_train_loss": -1.0,
570
- "1.2_train_acc": 1.0,
571
- "1.3_val_loss": -1.0,
572
- "1.4_val_acc": 1.0,
573
- "1.5_perplexity": 0.36787944123356736
574
- },
575
- "speed": {
576
- "2.1_throughput_sps": 4.378943522026081,
577
- "2.2_step_time_ms": 3653.8493633270264,
578
- "2.3_epoch_progress": 0.5,
579
- "2.4_rss_mb": 1164.43359375,
580
- "2.5_xeon_tflops": 1.408584045428496
581
- },
582
- "kohonen": {
583
- "sigma_t": 0.9954753752044792,
584
- "alpha_t": 0.08146473164114146,
585
- "t": 410,
586
- "n_neurons": 8,
587
- "fisher_w_mean": 0.0,
588
- "has_ewc_reference": true
589
- },
590
- "hypothesis": {
591
- "classifier_trained": true,
592
- "punishment_count": 0,
593
- "success_count": 4,
594
- "training_ready": true,
595
- "buffer_size": 50,
596
- "required_new_samples": 10
597
- }
598
- },
599
- {
600
- "step": 17,
601
- "epoch": 0,
602
- "dataset": "nvidia/OpenMathInstruct-2",
603
- "quality": {
604
- "1.1_train_loss": -1.0,
605
- "1.2_train_acc": 1.0,
606
- "1.3_val_loss": -1.0,
607
- "1.4_val_acc": 1.0,
608
- "1.5_perplexity": 0.36787944123356736
609
- },
610
- "speed": {
611
- "2.1_throughput_sps": 4.635223450423612,
612
- "2.2_step_time_ms": 3451.829274495443,
613
- "2.3_epoch_progress": 0.5,
614
- "2.4_rss_mb": 1164.43359375,
615
- "2.5_xeon_tflops": 1.408584045428496
616
- },
617
- "kohonen": {
618
- "sigma_t": 0.9954753752044792,
619
- "alpha_t": 0.08146473164114146,
620
- "t": 410,
621
- "n_neurons": 8,
622
- "fisher_w_mean": 0.0,
623
- "has_ewc_reference": true
624
- },
625
- "hypothesis": {
626
- "classifier_trained": true,
627
- "punishment_count": 0,
628
- "success_count": 5,
629
- "training_ready": true,
630
- "buffer_size": 66,
631
- "required_new_samples": 10
632
- }
633
- },
634
- {
635
- "step": 18,
636
- "epoch": 0,
637
- "dataset": "nvidia/OpenMathInstruct-2",
638
- "quality": {
639
- "1.1_train_loss": -0.9938837346736189,
640
- "1.2_train_acc": 0.9878048780487805,
641
- "1.3_val_loss": -0.9938837346736189,
642
- "1.4_val_acc": 0.9878048780487805,
643
- "1.5_perplexity": 0.37013638450004166
644
- },
645
- "speed": {
646
- "2.1_throughput_sps": 4.879517425318076,
647
- "2.2_step_time_ms": 3279.012780440481,
648
- "2.3_epoch_progress": 0.5,
649
- "2.4_rss_mb": 1164.43359375,
650
- "2.5_xeon_tflops": 1.408584045428496
651
- },
652
- "kohonen": {
653
- "sigma_t": 0.9954753752044792,
654
- "alpha_t": 0.08146473164114146,
655
- "t": 410,
656
- "n_neurons": 8,
657
- "fisher_w_mean": 0.0,
658
- "has_ewc_reference": true
659
- },
660
- "hypothesis": {
661
- "classifier_trained": true,
662
- "punishment_count": 1,
663
- "success_count": 0,
664
- "training_ready": true,
665
- "buffer_size": 82,
666
- "required_new_samples": 10
667
- }
668
- },
669
- {
670
- "step": 19,
671
- "epoch": 0,
672
- "dataset": "nvidia/OpenMathInstruct-2",
673
- "quality": {
674
- "1.1_train_loss": -1.0,
675
- "1.2_train_acc": 1.0,
676
- "1.3_val_loss": -1.0,
677
- "1.4_val_acc": 1.0,
678
- "1.5_perplexity": 0.36787944123356736
679
- },
680
- "speed": {
681
- "2.1_throughput_sps": 5.133065463246482,
682
- "2.2_step_time_ms": 3117.045772075653,
683
- "2.3_epoch_progress": 0.5,
684
- "2.4_rss_mb": 1164.43359375,
685
- "2.5_xeon_tflops": 1.408584045428496
686
- },
687
- "kohonen": {
688
- "sigma_t": 0.9954753752044792,
689
- "alpha_t": 0.08146473164114146,
690
- "t": 410,
691
- "n_neurons": 8,
692
- "fisher_w_mean": 0.0,
693
- "has_ewc_reference": true
694
- },
695
- "hypothesis": {
696
- "classifier_trained": true,
697
- "punishment_count": 0,
698
- "success_count": 0,
699
- "training_ready": false,
700
- "buffer_size": 0,
701
- "required_new_samples": 10
702
- }
703
- },
704
- {
705
- "step": 20,
706
- "epoch": 1,
707
- "dataset": "TucanoBR/GigaVerbo",
708
- "quality": {
709
- "1.1_train_loss": -1.0,
710
- "1.2_train_acc": 1.0,
711
- "1.3_val_loss": -1.0,
712
- "1.4_val_acc": 1.0,
713
- "1.5_perplexity": 0.36787944123356736
714
- },
715
- "speed": {
716
- "2.1_throughput_sps": 4.736881395121652,
717
- "2.2_step_time_ms": 3377.749760945638,
718
- "2.3_epoch_progress": 1.0,
719
- "2.4_rss_mb": 1164.43359375,
720
- "2.5_xeon_tflops": 1.408584045428496
721
- },
722
- "kohonen": {
723
- "sigma_t": 0.9189395912766242,
724
- "alpha_t": 0.07827045382418682,
725
- "t": 490,
726
- "n_neurons": 8,
727
- "fisher_w_mean": 0.0,
728
- "has_ewc_reference": true
729
- },
730
- "hypothesis": {
731
- "classifier_trained": true,
732
- "punishment_count": 0,
733
- "success_count": 1,
734
- "training_ready": true,
735
- "buffer_size": 16,
736
- "required_new_samples": 10
737
- }
738
- },
739
- {
740
- "step": 21,
741
- "epoch": 1,
742
- "dataset": "TucanoBR/GigaVerbo",
743
- "quality": {
744
- "1.1_train_loss": -0.9842509842514764,
745
- "1.2_train_acc": 0.96875,
746
- "1.3_val_loss": -0.9842509842514764,
747
- "1.4_val_acc": 0.96875,
748
- "1.5_perplexity": 0.3737190436413242
749
- },
750
- "speed": {
751
- "2.1_throughput_sps": 4.952858178183789,
752
- "2.2_step_time_ms": 3230.457934466275,
753
- "2.3_epoch_progress": 1.0,
754
- "2.4_rss_mb": 1164.43359375,
755
- "2.5_xeon_tflops": 1.408584045428496
756
- },
757
- "kohonen": {
758
- "sigma_t": 0.9189395912766242,
759
- "alpha_t": 0.07827045382418682,
760
- "t": 490,
761
- "n_neurons": 8,
762
- "fisher_w_mean": 0.0,
763
- "has_ewc_reference": true
764
- },
765
- "hypothesis": {
766
- "classifier_trained": true,
767
- "punishment_count": 1,
768
- "success_count": 0,
769
- "training_ready": true,
770
- "buffer_size": 32,
771
- "required_new_samples": 10
772
- }
773
- },
774
- {
775
- "step": 22,
776
- "epoch": 1,
777
- "dataset": "TucanoBR/GigaVerbo",
778
- "quality": {
779
- "1.1_train_loss": -1.0,
780
- "1.2_train_acc": 1.0,
781
- "1.3_val_loss": -1.0,
782
- "1.4_val_acc": 1.0,
783
- "1.5_perplexity": 0.36787944123356736
784
- },
785
- "speed": {
786
- "2.1_throughput_sps": 5.177268879481627,
787
- "2.2_step_time_ms": 3090.4324987660284,
788
- "2.3_epoch_progress": 1.0,
789
- "2.4_rss_mb": 1164.43359375,
790
- "2.5_xeon_tflops": 1.408584045428496
791
- },
792
- "kohonen": {
793
- "sigma_t": 0.9189395912766242,
794
- "alpha_t": 0.07827045382418682,
795
- "t": 490,
796
- "n_neurons": 8,
797
- "fisher_w_mean": 0.0,
798
- "has_ewc_reference": true
799
- },
800
- "hypothesis": {
801
- "classifier_trained": true,
802
- "punishment_count": 0,
803
- "success_count": 0,
804
- "training_ready": false,
805
- "buffer_size": 0,
806
- "required_new_samples": 10
807
- }
808
- },
809
- {
810
- "step": 23,
811
- "epoch": 1,
812
- "dataset": "TucanoBR/GigaVerbo",
813
- "quality": {
814
- "1.1_train_loss": -1.0,
815
- "1.2_train_acc": 1.0,
816
- "1.3_val_loss": -1.0,
817
- "1.4_val_acc": 1.0,
818
- "1.5_perplexity": 0.36787944123356736
819
- },
820
- "speed": {
821
- "2.1_throughput_sps": 5.402250378049719,
822
- "2.2_step_time_ms": 2961.728701988856,
823
- "2.3_epoch_progress": 1.0,
824
- "2.4_rss_mb": 1164.43359375,
825
- "2.5_xeon_tflops": 1.408584045428496
826
- },
827
- "kohonen": {
828
- "sigma_t": 0.9189395912766242,
829
- "alpha_t": 0.07827045382418682,
830
- "t": 490,
831
- "n_neurons": 8,
832
- "fisher_w_mean": 0.0,
833
- "has_ewc_reference": true
834
- },
835
- "hypothesis": {
836
- "classifier_trained": true,
837
- "punishment_count": 0,
838
- "success_count": 0,
839
- "training_ready": false,
840
- "buffer_size": 2,
841
- "required_new_samples": 10
842
- }
843
- },
844
- {
845
- "step": 24,
846
- "epoch": 1,
847
- "dataset": "dominguesm/restore-punctuation-ptbr-dataset",
848
- "quality": {
849
- "1.1_train_loss": -0.97182531580755,
850
- "1.2_train_acc": 0.9444444444444444,
851
- "1.3_val_loss": -0.97182531580755,
852
- "1.4_val_acc": 0.9444444444444444,
853
- "1.5_perplexity": 0.3783917230298394
854
- },
855
- "speed": {
856
- "2.1_throughput_sps": 5.059870370787907,
857
- "2.2_step_time_ms": 3162.136344909668,
858
- "2.3_epoch_progress": 1.0,
859
- "2.4_rss_mb": 1164.43359375,
860
- "2.5_xeon_tflops": 1.408584045428496
861
- },
862
- "kohonen": {
863
- "sigma_t": 0.8398475498481031,
864
- "alpha_t": 0.07482635675785652,
865
- "t": 580,
866
- "n_neurons": 8,
867
- "fisher_w_mean": 0.0,
868
- "has_ewc_reference": true
869
- },
870
- "hypothesis": {
871
- "classifier_trained": true,
872
- "punishment_count": 1,
873
- "success_count": 0,
874
- "training_ready": true,
875
- "buffer_size": 18,
876
- "required_new_samples": 10
877
- }
878
- },
879
- {
880
- "step": 25,
881
- "epoch": 1,
882
- "dataset": "dominguesm/restore-punctuation-ptbr-dataset",
883
- "quality": {
884
- "1.1_train_loss": -1.0,
885
- "1.2_train_acc": 1.0,
886
- "1.3_val_loss": -1.0,
887
- "1.4_val_acc": 1.0,
888
- "1.5_perplexity": 0.36787944123356736
889
- },
890
- "speed": {
891
- "2.1_throughput_sps": 5.26170412596044,
892
- "2.2_step_time_ms": 3040.839928847093,
893
- "2.3_epoch_progress": 1.0,
894
- "2.4_rss_mb": 1164.43359375,
895
- "2.5_xeon_tflops": 1.408584045428496
896
- },
897
- "kohonen": {
898
- "sigma_t": 0.8398475498481031,
899
- "alpha_t": 0.07482635675785652,
900
- "t": 580,
901
- "n_neurons": 8,
902
- "fisher_w_mean": 0.0,
903
- "has_ewc_reference": true
904
- },
905
- "hypothesis": {
906
- "classifier_trained": true,
907
- "punishment_count": 0,
908
- "success_count": 0,
909
- "training_ready": false,
910
- "buffer_size": 0,
911
- "required_new_samples": 10
912
- }
913
- },
914
- {
915
- "step": 26,
916
- "epoch": 1,
917
- "dataset": "dominguesm/restore-punctuation-ptbr-dataset",
918
- "quality": {
919
- "1.1_train_loss": -1.0,
920
- "1.2_train_acc": 1.0,
921
- "1.3_val_loss": -1.0,
922
- "1.4_val_acc": 1.0,
923
- "1.5_perplexity": 0.36787944123356736
924
- },
925
- "speed": {
926
- "2.1_throughput_sps": 5.462896677842805,
927
- "2.2_step_time_ms": 2928.8490966514305,
928
- "2.3_epoch_progress": 1.0,
929
- "2.4_rss_mb": 1164.43359375,
930
- "2.5_xeon_tflops": 1.408584045428496
931
- },
932
- "kohonen": {
933
- "sigma_t": 0.7752770017375488,
934
- "alpha_t": 0.07189237334319262,
935
- "t": 660,
936
- "n_neurons": 8,
937
- "fisher_w_mean": 0.0,
938
- "has_ewc_reference": true
939
- },
940
- "hypothesis": {
941
- "classifier_trained": true,
942
- "punishment_count": 0,
943
- "success_count": 1,
944
- "training_ready": true,
945
- "buffer_size": 16,
946
- "required_new_samples": 10
947
- }
948
- },
949
- {
950
- "step": 27,
951
- "epoch": 1,
952
- "dataset": "dominguesm/restore-punctuation-ptbr-dataset",
953
- "quality": {
954
- "1.1_train_loss": -1.0,
955
- "1.2_train_acc": 1.0,
956
- "1.3_val_loss": -1.0,
957
- "1.4_val_acc": 1.0,
958
- "1.5_perplexity": 0.36787944123356736
959
- },
960
- "speed": {
961
- "2.1_throughput_sps": 5.664832629426341,
962
- "2.2_step_time_ms": 2824.443553175245,
963
- "2.3_epoch_progress": 1.0,
964
- "2.4_rss_mb": 1164.43359375,
965
- "2.5_xeon_tflops": 1.408584045428496
966
- },
967
- "kohonen": {
968
- "sigma_t": 0.7752770017375488,
969
- "alpha_t": 0.07189237334319262,
970
- "t": 660,
971
- "n_neurons": 8,
972
- "fisher_w_mean": 0.0,
973
- "has_ewc_reference": true
974
- },
975
- "hypothesis": {
976
- "classifier_trained": true,
977
- "punishment_count": 0,
978
- "success_count": 2,
979
- "training_ready": true,
980
- "buffer_size": 18,
981
- "required_new_samples": 10
982
- }
983
- },
984
- {
985
- "step": 28,
986
- "epoch": 1,
987
- "dataset": "Madras1/corpus-ptbr-v2",
988
- "quality": {
989
- "1.1_train_loss": -0.985184366143778,
990
- "1.2_train_acc": 0.9705882352941176,
991
- "1.3_val_loss": -0.985184366143778,
992
- "1.4_val_acc": 0.9705882352941176,
993
- "1.5_perplexity": 0.3733703837949463
994
- },
995
- "speed": {
996
- "2.1_throughput_sps": 5.262945305074037,
997
- "2.2_step_time_ms": 3040.1227967492464,
998
- "2.3_epoch_progress": 1.0,
999
- "2.4_rss_mb": 1164.43359375,
1000
- "2.5_xeon_tflops": 1.408584045428496
1001
- },
1002
- "kohonen": {
1003
- "sigma_t": 0.7752770017375488,
1004
- "alpha_t": 0.07189237334319262,
1005
- "t": 660,
1006
- "n_neurons": 8,
1007
- "fisher_w_mean": 0.0,
1008
- "has_ewc_reference": true
1009
- },
1010
- "hypothesis": {
1011
- "classifier_trained": true,
1012
- "punishment_count": 1,
1013
- "success_count": 0,
1014
- "training_ready": true,
1015
- "buffer_size": 34,
1016
- "required_new_samples": 10
1017
- }
1018
- },
1019
- {
1020
- "step": 29,
1021
- "epoch": 1,
1022
- "dataset": "Madras1/corpus-ptbr-v2",
1023
- "quality": {
1024
- "1.1_train_loss": -1.0,
1025
- "1.2_train_acc": 1.0,
1026
- "1.3_val_loss": -1.0,
1027
- "1.4_val_acc": 1.0,
1028
- "1.5_perplexity": 0.36787944123356736
1029
- },
1030
- "speed": {
1031
- "2.1_throughput_sps": 5.443794300545346,
1032
- "2.2_step_time_ms": 2939.1264835993447,
1033
- "2.3_epoch_progress": 1.0,
1034
- "2.4_rss_mb": 1164.43359375,
1035
- "2.5_xeon_tflops": 1.408584045428496
1036
- },
1037
- "kohonen": {
1038
- "sigma_t": 0.7752770017375488,
1039
- "alpha_t": 0.07189237334319262,
1040
- "t": 660,
1041
- "n_neurons": 8,
1042
- "fisher_w_mean": 0.0,
1043
- "has_ewc_reference": true
1044
- },
1045
- "hypothesis": {
1046
- "classifier_trained": true,
1047
- "punishment_count": 0,
1048
- "success_count": 0,
1049
- "training_ready": false,
1050
- "buffer_size": 0,
1051
- "required_new_samples": 10
1052
- }
1053
- },
1054
- {
1055
- "step": 30,
1056
- "epoch": 1,
1057
- "dataset": "Madras1/corpus-ptbr-v2",
1058
- "quality": {
1059
- "1.1_train_loss": -1.0,
1060
- "1.2_train_acc": 1.0,
1061
- "1.3_val_loss": -1.0,
1062
- "1.4_val_acc": 1.0,
1063
- "1.5_perplexity": 0.36787944123356736
1064
- },
1065
- "speed": {
1066
- "2.1_throughput_sps": 5.624202472596881,
1067
- "2.2_step_time_ms": 2844.847794502012,
1068
- "2.3_epoch_progress": 1.0,
1069
- "2.4_rss_mb": 1164.43359375,
1070
- "2.5_xeon_tflops": 1.408584045428496
1071
- },
1072
- "kohonen": {
1073
- "sigma_t": 0.7156708732815515,
1074
- "alpha_t": 0.06907343306373546,
1075
- "t": 740,
1076
- "n_neurons": 8,
1077
- "fisher_w_mean": 0.0,
1078
- "has_ewc_reference": true
1079
- },
1080
- "hypothesis": {
1081
- "classifier_trained": true,
1082
- "punishment_count": 0,
1083
- "success_count": 1,
1084
- "training_ready": true,
1085
- "buffer_size": 16,
1086
- "required_new_samples": 10
1087
- }
1088
- },
1089
- {
1090
- "step": 31,
1091
- "epoch": 1,
1092
- "dataset": "Madras1/corpus-ptbr-v2",
1093
- "quality": {
1094
- "1.1_train_loss": -0.97182531580755,
1095
- "1.2_train_acc": 0.9444444444444444,
1096
- "1.3_val_loss": -0.97182531580755,
1097
- "1.4_val_acc": 0.9444444444444444,
1098
- "1.5_perplexity": 0.3783917230298394
1099
- },
1100
- "speed": {
1101
- "2.1_throughput_sps": 5.795399133742749,
1102
- "2.2_step_time_ms": 2760.81071048975,
1103
- "2.3_epoch_progress": 1.0,
1104
- "2.4_rss_mb": 1164.43359375,
1105
- "2.5_xeon_tflops": 1.408584045428496
1106
- },
1107
- "kohonen": {
1108
- "sigma_t": 0.7156708732815515,
1109
- "alpha_t": 0.06907343306373546,
1110
- "t": 740,
1111
- "n_neurons": 8,
1112
- "fisher_w_mean": 0.0,
1113
- "has_ewc_reference": true
1114
- },
1115
- "hypothesis": {
1116
- "classifier_trained": true,
1117
- "punishment_count": 1,
1118
- "success_count": 0,
1119
- "training_ready": true,
1120
- "buffer_size": 18,
1121
- "required_new_samples": 10
1122
- }
1123
- },
1124
- {
1125
- "step": 32,
1126
- "epoch": 1,
1127
- "dataset": "CEIA-POSITIVO/ultrachat_br_clustred_balanced_v1",
1128
- "quality": {
1129
- "1.1_train_loss": -1.0,
1130
- "1.2_train_acc": 1.0,
1131
- "1.3_val_loss": -1.0,
1132
- "1.4_val_acc": 1.0,
1133
- "1.5_perplexity": 0.36787944123356736
1134
- },
1135
- "speed": {
1136
- "2.1_throughput_sps": 5.160408763637774,
1137
- "2.2_step_time_ms": 3100.5295767928615,
1138
- "2.3_epoch_progress": 1.0,
1139
- "2.4_rss_mb": 1261.6953125,
1140
- "2.5_xeon_tflops": 1.408584045428496
1141
- },
1142
- "kohonen": {
1143
- "sigma_t": 0.7156708732815515,
1144
- "alpha_t": 0.06907343306373546,
1145
- "t": 740,
1146
- "n_neurons": 8,
1147
- "fisher_w_mean": 0.0,
1148
- "has_ewc_reference": true
1149
- },
1150
- "hypothesis": {
1151
- "classifier_trained": true,
1152
- "punishment_count": 0,
1153
- "success_count": 0,
1154
- "training_ready": false,
1155
- "buffer_size": 0,
1156
- "required_new_samples": 10
1157
- }
1158
- },
1159
- {
1160
- "step": 33,
1161
- "epoch": 1,
1162
- "dataset": "CEIA-POSITIVO/ultrachat_br_clustred_balanced_v1",
1163
- "quality": {
1164
- "1.1_train_loss": -1.0,
1165
- "1.2_train_acc": 1.0,
1166
- "1.3_val_loss": -1.0,
1167
- "1.4_val_acc": 1.0,
1168
- "1.5_perplexity": 0.36787944123356736
1169
- },
1170
- "speed": {
1171
- "2.1_throughput_sps": 5.315961374434798,
1172
- "2.2_step_time_ms": 3009.8036597756777,
1173
- "2.3_epoch_progress": 1.0,
1174
- "2.4_rss_mb": 1261.6953125,
1175
- "2.5_xeon_tflops": 1.408584045428496
1176
- },
1177
- "kohonen": {
1178
- "sigma_t": 0.6606474817589989,
1179
- "alpha_t": 0.06636502501363194,
1180
- "t": 820,
1181
- "n_neurons": 8,
1182
- "fisher_w_mean": 0.0,
1183
- "has_ewc_reference": true
1184
- },
1185
- "hypothesis": {
1186
- "classifier_trained": true,
1187
- "punishment_count": 0,
1188
- "success_count": 1,
1189
- "training_ready": true,
1190
- "buffer_size": 16,
1191
- "required_new_samples": 10
1192
- }
1193
- },
1194
- {
1195
- "step": 34,
1196
- "epoch": 1,
1197
- "dataset": "CEIA-POSITIVO/ultrachat_br_clustred_balanced_v1",
1198
- "quality": {
1199
- "1.1_train_loss": -1.0,
1200
- "1.2_train_acc": 1.0,
1201
- "1.3_val_loss": -1.0,
1202
- "1.4_val_acc": 1.0,
1203
- "1.5_perplexity": 0.36787944123356736
1204
- },
1205
- "speed": {
1206
- "2.1_throughput_sps": 5.47175251296733,
1207
- "2.2_step_time_ms": 2924.108859470912,
1208
- "2.3_epoch_progress": 1.0,
1209
- "2.4_rss_mb": 1261.6953125,
1210
- "2.5_xeon_tflops": 1.408584045428496
1211
- },
1212
- "kohonen": {
1213
- "sigma_t": 0.6606474817589989,
1214
- "alpha_t": 0.06636502501363194,
1215
- "t": 820,
1216
- "n_neurons": 8,
1217
- "fisher_w_mean": 0.0,
1218
- "has_ewc_reference": true
1219
- },
1220
- "hypothesis": {
1221
- "classifier_trained": true,
1222
- "punishment_count": 0,
1223
- "success_count": 2,
1224
- "training_ready": true,
1225
- "buffer_size": 32,
1226
- "required_new_samples": 10
1227
- }
1228
- },
1229
- {
1230
- "step": 35,
1231
- "epoch": 1,
1232
- "dataset": "CEIA-POSITIVO/ultrachat_br_clustred_balanced_v1",
1233
- "quality": {
1234
- "1.1_train_loss": -1.0,
1235
- "1.2_train_acc": 1.0,
1236
- "1.3_val_loss": -1.0,
1237
- "1.4_val_acc": 1.0,
1238
- "1.5_perplexity": 0.36787944123356736
1239
- },
1240
- "speed": {
1241
- "2.1_throughput_sps": 5.627667057550271,
1242
- "2.2_step_time_ms": 2843.0964085790847,
1243
- "2.3_epoch_progress": 1.0,
1244
- "2.4_rss_mb": 1261.6953125,
1245
- "2.5_xeon_tflops": 1.408584045428496
1246
- },
1247
- "kohonen": {
1248
- "sigma_t": 0.6606474817589989,
1249
- "alpha_t": 0.06636502501363194,
1250
- "t": 820,
1251
- "n_neurons": 8,
1252
- "fisher_w_mean": 0.0,
1253
- "has_ewc_reference": true
1254
- },
1255
- "hypothesis": {
1256
- "classifier_trained": true,
1257
- "punishment_count": 0,
1258
- "success_count": 3,
1259
- "training_ready": true,
1260
- "buffer_size": 34,
1261
- "required_new_samples": 10
1262
- }
1263
- },
1264
- {
1265
- "step": 36,
1266
- "epoch": 1,
1267
- "dataset": "nvidia/OpenMathInstruct-2",
1268
- "quality": {
1269
- "1.1_train_loss": -1.0,
1270
- "1.2_train_acc": 1.0,
1271
- "1.3_val_loss": -1.0,
1272
- "1.4_val_acc": 1.0,
1273
- "1.5_perplexity": 0.36787944123356736
1274
- },
1275
- "speed": {
1276
- "2.1_throughput_sps": 5.375498391903022,
1277
- "2.2_step_time_ms": 2976.468195786347,
1278
- "2.3_epoch_progress": 1.0,
1279
- "2.4_rss_mb": 1305.5859375,
1280
- "2.5_xeon_tflops": 1.408584045428496
1281
- },
1282
- "kohonen": {
1283
- "sigma_t": 0.6606474817589989,
1284
- "alpha_t": 0.06636502501363194,
1285
- "t": 820,
1286
- "n_neurons": 8,
1287
- "fisher_w_mean": 0.0,
1288
- "has_ewc_reference": true
1289
- },
1290
- "hypothesis": {
1291
- "classifier_trained": true,
1292
- "punishment_count": 0,
1293
- "success_count": 4,
1294
- "training_ready": true,
1295
- "buffer_size": 50,
1296
- "required_new_samples": 10
1297
- }
1298
- },
1299
- {
1300
- "step": 37,
1301
- "epoch": 1,
1302
- "dataset": "nvidia/OpenMathInstruct-2",
1303
- "quality": {
1304
- "1.1_train_loss": -1.0,
1305
- "1.2_train_acc": 1.0,
1306
- "1.3_val_loss": -1.0,
1307
- "1.4_val_acc": 1.0,
1308
- "1.5_perplexity": 0.36787944123356736
1309
- },
1310
- "speed": {
1311
- "2.1_throughput_sps": 5.519868523180669,
1312
- "2.2_step_time_ms": 2898.61976472955,
1313
- "2.3_epoch_progress": 1.0,
1314
- "2.4_rss_mb": 1305.5859375,
1315
- "2.5_xeon_tflops": 1.408584045428496
1316
- },
1317
- "kohonen": {
1318
- "sigma_t": 0.6606474817589989,
1319
- "alpha_t": 0.06636502501363194,
1320
- "t": 820,
1321
- "n_neurons": 8,
1322
- "fisher_w_mean": 0.0,
1323
- "has_ewc_reference": true
1324
- },
1325
- "hypothesis": {
1326
- "classifier_trained": true,
1327
- "punishment_count": 0,
1328
- "success_count": 5,
1329
- "training_ready": true,
1330
- "buffer_size": 66,
1331
- "required_new_samples": 10
1332
- }
1333
- },
1334
- {
1335
- "step": 38,
1336
- "epoch": 1,
1337
- "dataset": "nvidia/OpenMathInstruct-2",
1338
- "quality": {
1339
- "1.1_train_loss": -0.9938837346736189,
1340
- "1.2_train_acc": 0.9878048780487805,
1341
- "1.3_val_loss": -0.9938837346736189,
1342
- "1.4_val_acc": 0.9878048780487805,
1343
- "1.5_perplexity": 0.37013638450004166
1344
- },
1345
- "speed": {
1346
- "2.1_throughput_sps": 5.654276150669449,
1347
- "2.2_step_time_ms": 2829.716761906942,
1348
- "2.3_epoch_progress": 1.0,
1349
- "2.4_rss_mb": 1305.5859375,
1350
- "2.5_xeon_tflops": 1.408584045428496
1351
- },
1352
- "kohonen": {
1353
- "sigma_t": 0.6606474817589989,
1354
- "alpha_t": 0.06636502501363194,
1355
- "t": 820,
1356
- "n_neurons": 8,
1357
- "fisher_w_mean": 0.0,
1358
- "has_ewc_reference": true
1359
- },
1360
- "hypothesis": {
1361
- "classifier_trained": true,
1362
- "punishment_count": 1,
1363
- "success_count": 0,
1364
- "training_ready": true,
1365
- "buffer_size": 82,
1366
- "required_new_samples": 10
1367
- }
1368
- },
1369
- {
1370
- "step": 39,
1371
- "epoch": 1,
1372
- "dataset": "nvidia/OpenMathInstruct-2",
1373
- "quality": {
1374
- "1.1_train_loss": -1.0,
1375
- "1.2_train_acc": 1.0,
1376
- "1.3_val_loss": -1.0,
1377
- "1.4_val_acc": 1.0,
1378
- "1.5_perplexity": 0.36787944123356736
1379
- },
1380
- "speed": {
1381
- "2.1_throughput_sps": 5.798680614951719,
1382
- "2.2_step_time_ms": 2759.24836397171,
1383
- "2.3_epoch_progress": 1.0,
1384
- "2.4_rss_mb": 1305.5859375,
1385
- "2.5_xeon_tflops": 1.408584045428496
1386
- },
1387
- "kohonen": {
1388
- "sigma_t": 0.6606474817589989,
1389
- "alpha_t": 0.06636502501363194,
1390
- "t": 820,
1391
- "n_neurons": 8,
1392
- "fisher_w_mean": 0.0,
1393
- "has_ewc_reference": true
1394
- },
1395
- "hypothesis": {
1396
- "classifier_trained": true,
1397
- "punishment_count": 0,
1398
- "success_count": 0,
1399
- "training_ready": false,
1400
- "buffer_size": 0,
1401
- "required_new_samples": 10
1402
- }
1403
- }
1404
- ],
1405
- "summary": {
1406
- "n_steps": 40,
1407
- "final_loss": -1.0,
1408
- "mean_loss": -0.9953484858341988,
1409
- "min_loss": -1.0,
1410
- "max_loss": -0.97182531580755,
1411
- "final_acc": 1.0,
1412
- "mean_acc": 0.9908016001115894,
1413
- "sigma_start": 1.3846745195799537,
1414
- "sigma_end": 0.6606474817589989,
1415
- "alpha_start": 0.09607894391523232,
1416
- "alpha_end": 0.06636502501363194,
1417
- "rss_max_mb": 1305.5859375,
1418
- "rss_final_mb": 1305.5859375,
1419
- "rss_trend": "stable",
1420
- "n_alerts": 39,
1421
- "alerts": [
1422
- {
1423
- "type": "3.3_loss_vanishing",
1424
- "step": 1,
1425
- "value": -0.9842509842514764
1426
- },
1427
- {
1428
- "type": "3.3_loss_vanishing",
1429
- "step": 2,
1430
- "value": -1.0
1431
- },
1432
- {
1433
- "type": "3.3_loss_vanishing",
1434
- "step": 3,
1435
- "value": -1.0
1436
- },
1437
- {
1438
- "type": "3.3_loss_vanishing",
1439
- "step": 4,
1440
- "value": -0.97182531580755
1441
- },
1442
- {
1443
- "type": "3.3_loss_vanishing",
1444
- "step": 5,
1445
- "value": -1.0
1446
- },
1447
- {
1448
- "type": "3.3_loss_vanishing",
1449
- "step": 6,
1450
- "value": -1.0
1451
- },
1452
- {
1453
- "type": "3.3_loss_vanishing",
1454
- "step": 7,
1455
- "value": -1.0
1456
- },
1457
- {
1458
- "type": "3.3_loss_vanishing",
1459
- "step": 8,
1460
- "value": -0.985184366143778
1461
- },
1462
- {
1463
- "type": "3.3_loss_vanishing",
1464
- "step": 9,
1465
- "value": -1.0
1466
- },
1467
- {
1468
- "type": "3.3_loss_vanishing",
1469
- "step": 10,
1470
- "value": -1.0
1471
- },
1472
- {
1473
- "type": "3.3_loss_vanishing",
1474
- "step": 11,
1475
- "value": -0.97182531580755
1476
- },
1477
- {
1478
- "type": "3.3_loss_vanishing",
1479
- "step": 12,
1480
- "value": -1.0
1481
- },
1482
- {
1483
- "type": "3.3_loss_vanishing",
1484
- "step": 13,
1485
- "value": -1.0
1486
- },
1487
- {
1488
- "type": "3.3_loss_vanishing",
1489
- "step": 14,
1490
- "value": -1.0
1491
- },
1492
- {
1493
- "type": "3.3_loss_vanishing",
1494
- "step": 15,
1495
- "value": -1.0
1496
- },
1497
- {
1498
- "type": "3.3_loss_vanishing",
1499
- "step": 16,
1500
- "value": -1.0
1501
- },
1502
- {
1503
- "type": "3.3_loss_vanishing",
1504
- "step": 17,
1505
- "value": -1.0
1506
- },
1507
- {
1508
- "type": "3.3_loss_vanishing",
1509
- "step": 18,
1510
- "value": -0.9938837346736189
1511
- },
1512
- {
1513
- "type": "3.3_loss_vanishing",
1514
- "step": 19,
1515
- "value": -1.0
1516
- },
1517
- {
1518
- "type": "3.3_loss_vanishing",
1519
- "step": 20,
1520
- "value": -1.0
1521
- }
1522
- ],
1523
- "kohonen_final": {
1524
- "sigma_t": 0.6606474817589989,
1525
- "alpha_t": 0.06636502501363194,
1526
- "t": 820,
1527
- "n_neurons": 8,
1528
- "fisher_w_mean": 0.0,
1529
- "has_ewc_reference": true
1530
- },
1531
- "hypothesis_final": {
1532
- "classifier_trained": true,
1533
- "punishment_count": 0,
1534
- "success_count": 0,
1535
- "training_ready": false,
1536
- "buffer_size": 0,
1537
- "required_new_samples": 10
1538
- }
1539
- },
1540
- "alerts": [
1541
- {
1542
- "type": "3.3_loss_vanishing",
1543
- "step": 1,
1544
- "value": -0.9842509842514764
1545
- },
1546
- {
1547
- "type": "3.3_loss_vanishing",
1548
- "step": 2,
1549
- "value": -1.0
1550
- },
1551
- {
1552
- "type": "3.3_loss_vanishing",
1553
- "step": 3,
1554
- "value": -1.0
1555
- },
1556
- {
1557
- "type": "3.3_loss_vanishing",
1558
- "step": 4,
1559
- "value": -0.97182531580755
1560
- },
1561
- {
1562
- "type": "3.3_loss_vanishing",
1563
- "step": 5,
1564
- "value": -1.0
1565
- },
1566
- {
1567
- "type": "3.3_loss_vanishing",
1568
- "step": 6,
1569
- "value": -1.0
1570
- },
1571
- {
1572
- "type": "3.3_loss_vanishing",
1573
- "step": 7,
1574
- "value": -1.0
1575
- },
1576
- {
1577
- "type": "3.3_loss_vanishing",
1578
- "step": 8,
1579
- "value": -0.985184366143778
1580
- },
1581
- {
1582
- "type": "3.3_loss_vanishing",
1583
- "step": 9,
1584
- "value": -1.0
1585
- },
1586
- {
1587
- "type": "3.3_loss_vanishing",
1588
- "step": 10,
1589
- "value": -1.0
1590
- },
1591
- {
1592
- "type": "3.3_loss_vanishing",
1593
- "step": 11,
1594
- "value": -0.97182531580755
1595
- },
1596
- {
1597
- "type": "3.3_loss_vanishing",
1598
- "step": 12,
1599
- "value": -1.0
1600
- },
1601
- {
1602
- "type": "3.3_loss_vanishing",
1603
- "step": 13,
1604
- "value": -1.0
1605
- },
1606
- {
1607
- "type": "3.3_loss_vanishing",
1608
- "step": 14,
1609
- "value": -1.0
1610
- },
1611
- {
1612
- "type": "3.3_loss_vanishing",
1613
- "step": 15,
1614
- "value": -1.0
1615
- },
1616
- {
1617
- "type": "3.3_loss_vanishing",
1618
- "step": 16,
1619
- "value": -1.0
1620
- },
1621
- {
1622
- "type": "3.3_loss_vanishing",
1623
- "step": 17,
1624
- "value": -1.0
1625
- },
1626
- {
1627
- "type": "3.3_loss_vanishing",
1628
- "step": 18,
1629
- "value": -0.9938837346736189
1630
- },
1631
- {
1632
- "type": "3.3_loss_vanishing",
1633
- "step": 19,
1634
- "value": -1.0
1635
- },
1636
- {
1637
- "type": "3.3_loss_vanishing",
1638
- "step": 20,
1639
- "value": -1.0
1640
- },
1641
- {
1642
- "type": "3.3_loss_vanishing",
1643
- "step": 21,
1644
- "value": -0.9842509842514764
1645
- },
1646
- {
1647
- "type": "3.3_loss_vanishing",
1648
- "step": 22,
1649
- "value": -1.0
1650
- },
1651
- {
1652
- "type": "3.3_loss_vanishing",
1653
- "step": 23,
1654
- "value": -1.0
1655
- },
1656
- {
1657
- "type": "3.3_loss_vanishing",
1658
- "step": 24,
1659
- "value": -0.97182531580755
1660
- },
1661
- {
1662
- "type": "3.3_loss_vanishing",
1663
- "step": 25,
1664
- "value": -1.0
1665
- },
1666
- {
1667
- "type": "3.3_loss_vanishing",
1668
- "step": 26,
1669
- "value": -1.0
1670
- },
1671
- {
1672
- "type": "3.3_loss_vanishing",
1673
- "step": 27,
1674
- "value": -1.0
1675
- },
1676
- {
1677
- "type": "3.3_loss_vanishing",
1678
- "step": 28,
1679
- "value": -0.985184366143778
1680
- },
1681
- {
1682
- "type": "3.3_loss_vanishing",
1683
- "step": 29,
1684
- "value": -1.0
1685
- },
1686
- {
1687
- "type": "3.3_loss_vanishing",
1688
- "step": 30,
1689
- "value": -1.0
1690
- },
1691
- {
1692
- "type": "3.3_loss_vanishing",
1693
- "step": 31,
1694
- "value": -0.97182531580755
1695
- },
1696
- {
1697
- "type": "3.3_loss_vanishing",
1698
- "step": 32,
1699
- "value": -1.0
1700
- },
1701
- {
1702
- "type": "3.3_loss_vanishing",
1703
- "step": 33,
1704
- "value": -1.0
1705
- },
1706
- {
1707
- "type": "3.3_loss_vanishing",
1708
- "step": 34,
1709
- "value": -1.0
1710
- },
1711
- {
1712
- "type": "3.3_loss_vanishing",
1713
- "step": 35,
1714
- "value": -1.0
1715
- },
1716
- {
1717
- "type": "3.3_loss_vanishing",
1718
- "step": 36,
1719
- "value": -1.0
1720
- },
1721
- {
1722
- "type": "3.3_loss_vanishing",
1723
- "step": 37,
1724
- "value": -1.0
1725
- },
1726
- {
1727
- "type": "3.3_loss_vanishing",
1728
- "step": 38,
1729
- "value": -0.9938837346736189
1730
- },
1731
- {
1732
- "type": "3.3_loss_vanishing",
1733
- "step": 39,
1734
- "value": -1.0
1735
- }
1736
- ]
1737
- }