Robust Parameter-Update Envelope / mlp_comparison_results.json
Unverified
1{
2 "seed": 584,
3 "device": "cuda",
4 "results": [
5 {
6 "mode": "endpoint",
7 "lr": 0.25,
8 "K": 8,
9 "final_loss": 0.16628476977348328,
10 "best_loss": 0.16628476977348328,
11 "accepted_steps": 45,
12 "mean_accepted_fraction": 1.0,
13 "max_recorded_constraint_violation": -0.024117200970649733,
14 "positive_violation_steps": 0,
15 "loss_curve": [
16 0.9056398868560791,
17 0.8137199878692627,
18 0.7905830144882202,
19 0.777437686920166,
20 0.7660845518112183,
21 0.7551835775375366,
22 0.7444102168083191,
23 0.7336043119430542,
24 0.7226400375366211,
25 0.7114050388336182,
26 0.6997959613800049,
27 0.6877187490463257,
28 0.6750890016555786,
29 0.6618338823318481,
30 0.6478936076164246,
31 0.6332239508628845,
32 0.6177984476089478,
33 0.6016103029251099,
34 0.584674596786499,
35 0.5670287609100342,
36 0.5487328767776489,
37 0.5298683643341064,
38 0.5105358958244324,
39 0.49085158109664917,
40 0.47094282507896423,
41 0.45094358921051025,
42 0.43098878860473633,
43 0.4112095236778259,
44 0.3917284905910492,
45 0.37265676259994507,
46 0.35409051179885864,
47 0.3361099362373352,
48 0.3187783658504486,
49 0.30214256048202515,
50 0.28623372316360474,
51 0.2710689306259155,
52 0.25665295124053955,
53 0.24298009276390076,
54 0.23003646731376648,
55 0.21780145168304443,
56 0.20624983310699463,
57 0.1953531950712204,
58 0.18508099019527435,
59 0.175401970744133,
60 0.16628476977348328
61 ]
62 },
63 {
64 "mode": "envelope",
65 "lr": 0.25,
66 "K": 8,
67 "final_loss": 0.16628476977348328,
68 "best_loss": 0.16628476977348328,
69 "accepted_steps": 45,
70 "mean_accepted_fraction": 1.0,
71 "max_recorded_constraint_violation": -0.014999999999999902,
72 "positive_violation_steps": 0,
73 "loss_curve": [
74 0.9056398868560791,
75 0.8137199878692627,
76 0.7905830144882202,
77 0.777437686920166,
78 0.7660845518112183,
79 0.7551835775375366,
80 0.7444102168083191,
81 0.7336043119430542,
82 0.7226400375366211,
83 0.7114050388336182,
84 0.6997959613800049,
85 0.6877187490463257,
86 0.6750890016555786,
87 0.6618338823318481,
88 0.6478936076164246,
89 0.6332239508628845,
90 0.6177984476089478,
91 0.6016103029251099,
92 0.584674596786499,
93 0.5670287609100342,
94 0.5487328767776489,
95 0.5298683643341064,
96 0.5105358958244324,
97 0.49085158109664917,
98 0.47094282507896423,
99 0.45094358921051025,
100 0.43098878860473633,
101 0.4112095236778259,
102 0.3917284905910492,
103 0.37265676259994507,
104 0.35409051179885864,
105 0.3361099362373352,
106 0.3187783658504486,
107 0.30214256048202515,
108 0.28623372316360474,
109 0.2710689306259155,
110 0.25665295124053955,
111 0.24298009276390076,
112 0.23003646731376648,
113 0.21780145168304443,
114 0.20624983310699463,
115 0.1953531950712204,
116 0.18508099019527435,
117 0.175401970744133,
118 0.16628476977348328
119 ]
120 },
121 {
122 "mode": "envelope",
123 "lr": 0.25,
124 "K": 16,
125 "final_loss": 0.16628476977348328,
126 "best_loss": 0.16628476977348328,
127 "accepted_steps": 45,
128 "mean_accepted_fraction": 1.0,
129 "max_recorded_constraint_violation": -0.014999999999999902,
130 "positive_violation_steps": 0,
131 "loss_curve": [
132 0.9056398868560791,
133 0.8137199878692627,
134 0.7905830144882202,
135 0.777437686920166,
136 0.7660845518112183,
137 0.7551835775375366,
138 0.7444102168083191,
139 0.7336043119430542,
140 0.7226400375366211,
141 0.7114050388336182,
142 0.6997959613800049,
143 0.6877187490463257,
144 0.6750890016555786,
145 0.6618338823318481,
146 0.6478936076164246,
147 0.6332239508628845,
148 0.6177984476089478,
149 0.6016103029251099,
150 0.584674596786499,
151 0.5670287609100342,
152 0.5487328767776489,
153 0.5298683643341064,
154 0.5105358958244324,
155 0.49085158109664917,
156 0.47094282507896423,
157 0.45094358921051025,
158 0.43098878860473633,
159 0.4112095236778259,
160 0.3917284905910492,
161 0.37265676259994507,
162 0.35409051179885864,
163 0.3361099362373352,
164 0.3187783658504486,
165 0.30214256048202515,
166 0.28623372316360474,
167 0.2710689306259155,
168 0.25665295124053955,
169 0.24298009276390076,
170 0.23003646731376648,
171 0.21780145168304443,
172 0.20624983310699463,
173 0.1953531950712204,
174 0.18508099019527435,
175 0.175401970744133,
176 0.16628476977348328
177 ]
178 },
179 {
180 "mode": "endpoint",
181 "lr": 0.5,
182 "K": 8,
183 "final_loss": 1.4377998113632202,
184 "best_loss": 1.4377998113632202,
185 "accepted_steps": 0,
186 "mean_accepted_fraction": 0.0,
187 "max_recorded_constraint_violation": 0.4823863363265992,
188 "positive_violation_steps": 45,
189 "loss_curve": [
190 1.4377998113632202,
191 1.4377998113632202,
192 1.4377998113632202,
193 1.4377998113632202,
194 1.4377998113632202,
195 1.4377998113632202,
196 1.4377998113632202,
197 1.4377998113632202,
198 1.4377998113632202,
199 1.4377998113632202,
200 1.4377998113632202,
201 1.4377998113632202,
202 1.4377998113632202,
203 1.4377998113632202,
204 1.4377998113632202,
205 1.4377998113632202,
206 1.4377998113632202,
207 1.4377998113632202,
208 1.4377998113632202,
209 1.4377998113632202,
210 1.4377998113632202,
211 1.4377998113632202,
212 1.4377998113632202,
213 1.4377998113632202,
214 1.4377998113632202,
215 1.4377998113632202,
216 1.4377998113632202,
217 1.4377998113632202,
218 1.4377998113632202,
219 1.4377998113632202,
220 1.4377998113632202,
221 1.4377998113632202,
222 1.4377998113632202,
223 1.4377998113632202,
224 1.4377998113632202,
225 1.4377998113632202,
226 1.4377998113632202,
227 1.4377998113632202,
228 1.4377998113632202,
229 1.4377998113632202,
230 1.4377998113632202,
231 1.4377998113632202,
232 1.4377998113632202,
233 1.4377998113632202,
234 1.4377998113632202
235 ]
236 },
237 {
238 "mode": "envelope",
239 "lr": 0.5,
240 "K": 8,
241 "final_loss": 0.15636661648750305,
242 "best_loss": 0.1463879644870758,
243 "accepted_steps": 45,
244 "mean_accepted_fraction": 0.6444444444444445,
245 "max_recorded_constraint_violation": -0.0025947833061218395,
246 "positive_violation_steps": 0,
247 "loss_curve": [
248 0.9056398272514343,
249 0.8137199282646179,
250 0.804550290107727,
251 0.7699558734893799,
252 0.7571280002593994,
253 0.7630926370620728,
254 0.7182933688163757,
255 0.7086610198020935,
256 0.6801978945732117,
257 0.6612212061882019,
258 0.6570027470588684,
259 0.6130310893058777,
260 0.5966678857803345,
261 0.5606674551963806,
262 0.5387527942657471,
263 0.5045109987258911,
264 0.48371022939682007,
265 0.44780808687210083,
266 0.43914589285850525,
267 0.39692848920822144,
268 0.3698751926422119,
269 0.36417150497436523,
270 0.3302154541015625,
271 0.3049224317073822,
272 0.2851577699184418,
273 0.28795167803764343,
274 0.26266154646873474,
275 0.24184635281562805,
276 0.2253611981868744,
277 0.210927814245224,
278 0.1984177529811859,
279 0.1870478391647339,
280 0.17677772045135498,
281 0.16728174686431885,
282 0.17968696355819702,
283 0.17082151770591736,
284 0.1642211675643921,
285 0.15781565010547638,
286 0.15353994071483612,
287 0.14948216080665588,
288 0.14778533577919006,
289 0.1463879644870758,
290 0.14800810813903809,
291 0.15005329251289368,
292 0.15636661648750305
293 ]
294 },
295 {
296 "mode": "envelope",
297 "lr": 0.5,
298 "K": 16,
299 "final_loss": 0.15636661648750305,
300 "best_loss": 0.1463879644870758,
301 "accepted_steps": 45,
302 "mean_accepted_fraction": 0.6444444444444445,
303 "max_recorded_constraint_violation": -0.0025947833061218395,
304 "positive_violation_steps": 0,
305 "loss_curve": [
306 0.9056398272514343,
307 0.8137199282646179,
308 0.804550290107727,
309 0.7699558734893799,
310 0.7571280002593994,
311 0.7630926370620728,
312 0.7182933688163757,
313 0.7086610198020935,
314 0.6801978945732117,
315 0.6612212061882019,
316 0.6570027470588684,
317 0.6130310893058777,
318 0.5966678857803345,
319 0.5606674551963806,
320 0.5387527942657471,
321 0.5045109987258911,
322 0.48371022939682007,
323 0.44780808687210083,
324 0.43914589285850525,
325 0.39692848920822144,
326 0.3698751926422119,
327 0.36417150497436523,
328 0.3302154541015625,
329 0.3049224317073822,
330 0.2851577699184418,
331 0.28795167803764343,
332 0.26266154646873474,
333 0.24184635281562805,
334 0.2253611981868744,
335 0.210927814245224,
336 0.1984177529811859,
337 0.1870478391647339,
338 0.17677772045135498,
339 0.16728174686431885,
340 0.17968696355819702,
341 0.17082151770591736,
342 0.1642211675643921,
343 0.15781565010547638,
344 0.15353994071483612,
345 0.14948216080665588,
346 0.14778533577919006,
347 0.1463879644870758,
348 0.14800810813903809,
349 0.15005329251289368,
350 0.15636661648750305
351 ]
352 },
353 {
354 "mode": "endpoint",
355 "lr": 1.0,
356 "K": 8,
357 "final_loss": 1.4377998113632202,
358 "best_loss": 1.4377998113632202,
359 "accepted_steps": 0,
360 "mean_accepted_fraction": 0.0,
361 "max_recorded_constraint_violation": 6.733340249061585,
362 "positive_violation_steps": 45,
363 "loss_curve": [
364 1.4377998113632202,
365 1.4377998113632202,
366 1.4377998113632202,
367 1.4377998113632202,
368 1.4377998113632202,
369 1.4377998113632202,
370 1.4377998113632202,
371 1.4377998113632202,
372 1.4377998113632202,
373 1.4377998113632202,
374 1.4377998113632202,
375 1.4377998113632202,
376 1.4377998113632202,
377 1.4377998113632202,
378 1.4377998113632202,
379 1.4377998113632202,
380 1.4377998113632202,
381 1.4377998113632202,
382 1.4377998113632202,
383 1.4377998113632202,
384 1.4377998113632202,
385 1.4377998113632202,
386 1.4377998113632202,
387 1.4377998113632202,
388 1.4377998113632202,
389 1.4377998113632202,
390 1.4377998113632202,
391 1.4377998113632202,
392 1.4377998113632202,
393 1.4377998113632202,
394 1.4377998113632202,
395 1.4377998113632202,
396 1.4377998113632202,
397 1.4377998113632202,
398 1.4377998113632202,
399 1.4377998113632202,
400 1.4377998113632202,
401 1.4377998113632202,
402 1.4377998113632202,
403 1.4377998113632202,
404 1.4377998113632202,
405 1.4377998113632202,
406 1.4377998113632202,
407 1.4377998113632202,
408 1.4377998113632202
409 ]
410 },
411 {
412 "mode": "envelope",
413 "lr": 1.0,
414 "K": 8,
415 "final_loss": 0.12245418131351471,
416 "best_loss": 0.12089726328849792,
417 "accepted_steps": 45,
418 "mean_accepted_fraction": 0.3277777777777778,
419 "max_recorded_constraint_violation": -0.002444579601287855,
420 "positive_violation_steps": 0,
421 "loss_curve": [
422 0.9056398868560791,
423 0.8137199282646179,
424 0.8045501708984375,
425 0.7699559926986694,
426 0.7571279406547546,
427 0.7630926966667175,
428 0.7182933688163757,
429 0.7086610198020935,
430 0.6801978349685669,
431 0.6924211382865906,
432 0.6229782104492188,
433 0.6263723373413086,
434 0.5706595778465271,
435 0.5668681859970093,
436 0.5157341957092285,
437 0.5153146982192993,
438 0.46186357736587524,
439 0.43095406889915466,
440 0.41149473190307617,
441 0.37760329246520996,
442 0.383009135723114,
443 0.33855152130126953,
444 0.31084898114204407,
445 0.28943121433258057,
446 0.2920459508895874,
447 0.2634541392326355,
448 0.2429017722606659,
449 0.22581033408641815,
450 0.21157485246658325,
451 0.1989167183637619,
452 0.21147213876247406,
453 0.19552230834960938,
454 0.1831485480070114,
455 0.17169731855392456,
456 0.1624072790145874,
457 0.15374551713466644,
458 0.14661294221878052,
459 0.13996952772140503,
460 0.13461840152740479,
461 0.1297232061624527,
462 0.12614433467388153,
463 0.12308579683303833,
464 0.12163537740707397,
465 0.12089726328849792,
466 0.12245418131351471
467 ]
468 },
469 {
470 "mode": "envelope",
471 "lr": 1.0,
472 "K": 16,
473 "final_loss": 0.12245418131351471,
474 "best_loss": 0.12089726328849792,
475 "accepted_steps": 45,
476 "mean_accepted_fraction": 0.3277777777777778,
477 "max_recorded_constraint_violation": -0.002444579601287855,
478 "positive_violation_steps": 0,
479 "loss_curve": [
480 0.9056398868560791,
481 0.8137199282646179,
482 0.8045501708984375,
483 0.7699559926986694,
484 0.7571279406547546,
485 0.7630926966667175,
486 0.7182933688163757,
487 0.7086610198020935,
488 0.6801978349685669,
489 0.6924211382865906,
490 0.6229782104492188,
491 0.6263723373413086,
492 0.5706595778465271,
493 0.5668681859970093,
494 0.5157341957092285,
495 0.5153146982192993,
496 0.46186357736587524,
497 0.43095406889915466,
498 0.41149473190307617,
499 0.37760329246520996,
500 0.383009135723114,
501 0.33855152130126953,
502 0.31084898114204407,
503 0.28943121433258057,
504 0.2920459508895874,
505 0.2634541392326355,
506 0.2429017722606659,
507 0.22581033408641815,
508 0.21157485246658325,
509 0.1989167183637619,
510 0.21147213876247406,
511 0.19552230834960938,
512 0.1831485480070114,
513 0.17169731855392456,
514 0.1624072790145874,
515 0.15374551713466644,
516 0.14661294221878052,
517 0.13996952772140503,
518 0.13461840152740479,
519 0.1297232061624527,
520 0.12614433467388153,
521 0.12308579683303833,
522 0.12163537740707397,
523 0.12089726328849792,
524 0.12245418131351471
525 ]
526 },
527 {
528 "mode": "endpoint",
529 "lr": 2.0,
530 "K": 8,
531 "final_loss": 1.4377998113632202,
532 "best_loss": 1.4377998113632202,
533 "accepted_steps": 0,
534 "mean_accepted_fraction": 0.0,
535 "max_recorded_constraint_violation": 32.38039873600006,
536 "positive_violation_steps": 45,
537 "loss_curve": [
538 1.4377998113632202,
539 1.4377998113632202,
540 1.4377998113632202,
541 1.4377998113632202,
542 1.4377998113632202,
543 1.4377998113632202,
544 1.4377998113632202,
545 1.4377998113632202,
546 1.4377998113632202,
547 1.4377998113632202,
548 1.4377998113632202,
549 1.4377998113632202,
550 1.4377998113632202,
551 1.4377998113632202,
552 1.4377998113632202,
553 1.4377998113632202,
554 1.4377998113632202,
555 1.4377998113632202,
556 1.4377998113632202,
557 1.4377998113632202,
558 1.4377998113632202,
559 1.4377998113632202,
560 1.4377998113632202,
561 1.4377998113632202,
562 1.4377998113632202,
563 1.4377998113632202,
564 1.4377998113632202,
565 1.4377998113632202,
566 1.4377998113632202,
567 1.4377998113632202,
568 1.4377998113632202,
569 1.4377998113632202,
570 1.4377998113632202,
571 1.4377998113632202,
572 1.4377998113632202,
573 1.4377998113632202,
574 1.4377998113632202,
575 1.4377998113632202,
576 1.4377998113632202,
577 1.4377998113632202,
578 1.4377998113632202,
579 1.4377998113632202,
580 1.4377998113632202,
581 1.4377998113632202,
582 1.4377998113632202
583 ]
584 },
585 {
586 "mode": "envelope",
587 "lr": 2.0,
588 "K": 8,
589 "final_loss": 0.12245430797338486,
590 "best_loss": 0.12089737504720688,
591 "accepted_steps": 45,
592 "mean_accepted_fraction": 0.1638888888888889,
593 "max_recorded_constraint_violation": -0.002444475293159498,
594 "positive_violation_steps": 0,
595 "loss_curve": [
596 0.9056398868560791,
597 0.8137199282646179,
598 0.8045501708984375,
599 0.7699559926986694,
600 0.7571279406547546,
601 0.7630926370620728,
602 0.7182933688163757,
603 0.7086610198020935,
604 0.6801978945732117,
605 0.6924210786819458,
606 0.6229782104492188,
607 0.6263723373413086,
608 0.5706595182418823,
609 0.566868245601654,
610 0.5157342553138733,
611 0.5153146386146545,
612 0.46186351776123047,
613 0.43095409870147705,
614 0.41149479150772095,
615 0.37760329246520996,
616 0.3830091953277588,
617 0.3385516107082367,
618 0.31084901094436646,
619 0.28943130373954773,
620 0.2920461893081665,
621 0.26345428824424744,
622 0.24290180206298828,
623 0.22581037878990173,
624 0.21157491207122803,
625 0.1989167332649231,
626 0.2114722579717636,
627 0.19552242755889893,
628 0.18314872682094574,
629 0.1716974824666977,
630 0.16240742802619934,
631 0.153745636343956,
632 0.14661306142807007,
633 0.1399696320295334,
634 0.13461850583553314,
635 0.12972331047058105,
636 0.1261444240808487,
637 0.12308590859174728,
638 0.12163546681404114,
639 0.12089737504720688,
640 0.12245430797338486
641 ]
642 },
643 {
644 "mode": "envelope",
645 "lr": 2.0,
646 "K": 16,
647 "final_loss": 0.12245430797338486,
648 "best_loss": 0.12089737504720688,
649 "accepted_steps": 45,
650 "mean_accepted_fraction": 0.1638888888888889,
651 "max_recorded_constraint_violation": -0.002444475293159498,
652 "positive_violation_steps": 0,
653 "loss_curve": [
654 0.9056398868560791,
655 0.8137199282646179,
656 0.8045501708984375,
657 0.7699559926986694,
658 0.7571279406547546,
659 0.7630926370620728,
660 0.7182933688163757,
661 0.7086610198020935,
662 0.6801978945732117,
663 0.6924210786819458,
664 0.6229782104492188,
665 0.6263723373413086,
666 0.5706595182418823,
667 0.566868245601654,
668 0.5157342553138733,
669 0.5153146386146545,
670 0.46186351776123047,
671 0.43095409870147705,
672 0.41149479150772095,
673 0.37760329246520996,
674 0.3830091953277588,
675 0.3385516107082367,
676 0.31084901094436646,
677 0.28943130373954773,
678 0.2920461893081665,
679 0.26345428824424744,
680 0.24290180206298828,
681 0.22581037878990173,
682 0.21157491207122803,
683 0.1989167332649231,
684 0.2114722579717636,
685 0.19552242755889893,
686 0.18314872682094574,
687 0.1716974824666977,
688 0.16240742802619934,
689 0.153745636343956,
690 0.14661306142807007,
691 0.1399696320295334,
692 0.13461850583553314,
693 0.12972331047058105,
694 0.1261444240808487,
695 0.12308590859174728,
696 0.12163546681404114,
697 0.12089737504720688,
698 0.12245430797338486
699 ]
700 }
701 ]
702}