Anurag1734/cuda-error-resolution-analysis
07
1[2 {3 "post_stream": {4 "posts": [5 {6 "id": 156636,7 "name": "이홍석",8 "username": "lhsICT",9 "avatar_template": "/user_avatar/discuss.pytorch.org/lhsict/{size}/18877_2.png",10 "created_at": "2020-01-05T18:05:52.126Z",11 "cooked": "<p>Which I wanted was a software that return ‘2’ when I give float list’1,2,3,4’as input.<br>\nHowever, it made error message</p>\n<blockquote>\n<p>import torch<br>\nimport torch.nn as nn<br>\nimport torch.nn.functional as F<br>\nimport numpy as np<br>\nimport pandas as pd<br>\nimport matplotlib.pyplot as plt<br>\nimport datetime<br>\nimport torch.optim as optim</p>\n<p>torch.manual_seed(1)</p>\n<p>input_size=1<br>\nhidden_size=1<br>\nlearning_rate=0.1<br>\nx_data=[[1,2,3,4]]<br>\nx_one_hot=[[[1],[2],[3],[4]]]<br>\ny_data=[[2]]</p>\n<p>x=torch.FloatTensor(x_one_hot)<br>\ny=torch.LongTensor(y_data)</p>\n<p><span class=\"hashtag-raw\">#declare</span> RNN<br>\nrnn=torch.nn.RNN(input_size,hidden_size,batch_first=True)</p>\n<p><span class=\"hashtag-raw\">#loss</span>&optimizer setting<br>\ncriterion=torch.nn.CrossEntropyLoss()<br>\noptimizer=optim.Adam(rnn.parameters(),learning_rate)</p>\n<p><span class=\"hashtag-raw\">#start</span> training<br>\nfor i in range(100):<br>\noptimizer.zero_grad()<br>\noutputs,_status=rnn(x)<br>\n<span class=\"hashtag-raw\">#print</span>(outputs.shape)<br>\n<span class=\"hashtag-raw\">#print</span>(outputs.view(-1,input_size).shape)<br>\n<span class=\"hashtag-raw\">#print</span>(y.view(-1).shape)<br>\nprint(outputs[0].shape)<br>\nprint(y[0].shape)<br>\nloss=criterion(outputs[0],y[0])<br>\nloss.backward()<br>\noptimizer.step()<br>\nresult=outputs.data.numpy().argmax(axis=2)<br>\nprint(“prediction:”,result)</p>\n</blockquote>\n<p>and this is an error message<br>\nRuntimeError: Assertion `cur_target >= 0 && cur_target < n_classes’ failed.<br>\nI need some help.Thank you.</p>",12 "post_number": 1,13 "post_type": 1,14 "posts_count": 1,15 "updated_at": "2020-01-05T18:07:48.321Z",16 "reply_count": 0,17 "reply_to_post_number": null,18 "quote_count": 0,19 "incoming_link_count": 15,20 "reads": 7,21 "readers_count": 6,22 "score": 76.4,23 "yours": false,24 "topic_id": 65831,25 "topic_slug": "rnn-assertion-cur-target-0-cur-target-n-classes-failed",26 "display_username": "이홍석",27 "primary_group_name": null,28 "flair_name": null,29 "flair_url": null,30 "flair_bg_color": null,31 "flair_color": null,32 "flair_group_id": null,33 "badges_granted": [],34 "version": 1,35 "can_edit": false,36 "can_delete": false,37 "can_recover": false,38 "can_see_hidden_post": false,39 "can_wiki": false,40 "read": true,41 "user_title": null,42 "bookmarked": false,43 "actions_summary": [],44 "moderator": false,45 "admin": false,46 "staff": false,47 "user_id": 25690,48 "hidden": false,49 "trust_level": 1,50 "deleted_at": null,51 "user_deleted": false,52 "edit_reason": null,53 "can_view_edit_history": true,54 "wiki": false,55 "post_url": "/t/rnn-assertion-cur-target-0-cur-target-n-classes-failed/65831/1",56 "can_accept_answer": false,57 "can_unaccept_answer": false,58 "accepted_answer": false,59 "topic_accepted_answer": null,60 "can_vote": false61 }62 ],63 "stream": [64 15663665 ]66 },67 "timeline_lookup": [68 [69 1,70 212071 ]72 ],73 "suggested_topics": [74 {75 "fancy_title": "Segment Reduce memory problems while building?",76 "id": 220488,77 "title": "Segment Reduce memory problems while building?",78 "slug": "segment-reduce-memory-problems-while-building",79 "posts_count": 3,80 "reply_count": 0,81 "highest_post_number": 3,82 "image_url": null,83 "created_at": "2025-06-01T00:16:26.856Z",84 "last_posted_at": "2025-06-17T13:49:29.491Z",85 "bumped": true,86 "bumped_at": "2025-06-17T13:49:29.491Z",87 "archetype": "regular",88 "unseen": false,89 "pinned": false,90 "unpinned": null,91 "visible": true,92 "closed": false,93 "archived": false,94 "bookmarked": null,95 "liked": null,96 "tags_descriptions": {},97 "like_count": 0,98 "views": 189,99 "category_id": 1,100 "featured_link": null,101 "has_accepted_answer": false,102 "posters": [103 {104 "extras": null,105 "description": "Original Poster",106 "user": {107 "id": 84532,108 "username": "NateTalley",109 "name": "Nate Talley",110 "avatar_template": "/user_avatar/discuss.pytorch.org/natetalley/{size}/77228_2.png",111 "trust_level": 0112 }113 },114 {115 "extras": null,116 "description": "Frequent Poster",117 "user": {118 "id": 3534,119 "username": "ptrblck",120 "name": "",121 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",122 "admin": true,123 "moderator": true,124 "trust_level": 2125 }126 },127 {128 "extras": "latest",129 "description": "Most Recent Poster",130 "user": {131 "id": 67009,132 "username": "atalman",133 "name": "Andrey",134 "avatar_template": "/user_avatar/discuss.pytorch.org/atalman/{size}/72308_2.png",135 "trust_level": 1136 }137 }138 ]139 },140 {141 "fancy_title": "RuntimeError: mat1 and mat2 shapes cannot be multiplied (1x92160 and 9216x120)",142 "id": 212186,143 "title": "RuntimeError: mat1 and mat2 shapes cannot be multiplied (1x92160 and 9216x120)",144 "slug": "runtimeerror-mat1-and-mat2-shapes-cannot-be-multiplied-1x92160-and-9216x120",145 "posts_count": 9,146 "reply_count": 7,147 "highest_post_number": 9,148 "image_url": null,149 "created_at": "2024-10-28T03:24:27.630Z",150 "last_posted_at": "2024-10-28T08:36:54.814Z",151 "bumped": true,152 "bumped_at": "2024-10-28T08:36:54.814Z",153 "archetype": "regular",154 "unseen": false,155 "pinned": false,156 "unpinned": null,157 "visible": true,158 "closed": false,159 "archived": false,160 "bookmarked": null,161 "liked": null,162 "tags_descriptions": {},163 "like_count": 0,164 "views": 109,165 "category_id": 1,166 "featured_link": null,167 "has_accepted_answer": false,168 "posters": [169 {170 "extras": "latest",171 "description": "Original Poster, Most Recent Poster",172 "user": {173 "id": 80524,174 "username": "nassimus",175 "name": "nassim moualek",176 "avatar_template": "/user_avatar/discuss.pytorch.org/nassimus/{size}/73605_2.png",177 "trust_level": 1178 }179 },180 {181 "extras": null,182 "description": "Frequent Poster",183 "user": {184 "id": 12117,185 "username": "Tony-Y",186 "name": "",187 "avatar_template": "/user_avatar/discuss.pytorch.org/tony-y/{size}/8373_2.png",188 "trust_level": 2189 }190 }191 ]192 },193 {194 "fancy_title": "Dropout design choice",195 "id": 214984,196 "title": "Dropout design choice",197 "slug": "dropout-design-choice",198 "posts_count": 2,199 "reply_count": 0,200 "highest_post_number": 2,201 "image_url": null,202 "created_at": "2025-01-05T00:07:45.968Z",203 "last_posted_at": "2025-01-05T19:33:48.026Z",204 "bumped": true,205 "bumped_at": "2025-01-05T19:33:48.026Z",206 "archetype": "regular",207 "unseen": false,208 "pinned": false,209 "unpinned": null,210 "visible": true,211 "closed": false,212 "archived": false,213 "bookmarked": null,214 "liked": null,215 "tags_descriptions": {},216 "like_count": 0,217 "views": 135,218 "category_id": 1,219 "featured_link": null,220 "has_accepted_answer": false,221 "posters": [222 {223 "extras": null,224 "description": "Original Poster",225 "user": {226 "id": 77182,227 "username": "Siddhanth_Ramani",228 "name": "Siddhanth Ramani",229 "avatar_template": "/user_avatar/discuss.pytorch.org/siddhanth_ramani/{size}/62072_2.png",230 "trust_level": 1231 }232 },233 {234 "extras": "latest",235 "description": "Most Recent Poster",236 "user": {237 "id": 3534,238 "username": "ptrblck",239 "name": "",240 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",241 "admin": true,242 "moderator": true,243 "trust_level": 2244 }245 }246 ]247 },248 {249 "fancy_title": "Running Pytorch 1.13 on H100",250 "id": 212333,251 "title": "Running Pytorch 1.13 on H100",252 "slug": "running-pytorch-1-13-on-h100",253 "posts_count": 6,254 "reply_count": 4,255 "highest_post_number": 6,256 "image_url": null,257 "created_at": "2024-10-30T23:18:12.056Z",258 "last_posted_at": "2024-11-16T14:28:04.330Z",259 "bumped": true,260 "bumped_at": "2024-11-16T14:28:04.330Z",261 "archetype": "regular",262 "unseen": false,263 "pinned": false,264 "unpinned": null,265 "visible": true,266 "closed": false,267 "archived": false,268 "bookmarked": null,269 "liked": null,270 "tags_descriptions": {},271 "like_count": 0,272 "views": 577,273 "category_id": 1,274 "featured_link": null,275 "has_accepted_answer": false,276 "posters": [277 {278 "extras": null,279 "description": "Original Poster",280 "user": {281 "id": 2920,282 "username": "Sia_Rezaei",283 "name": "Sia Rezaei",284 "avatar_template": "/user_avatar/discuss.pytorch.org/sia_rezaei/{size}/43436_2.png",285 "trust_level": 2286 }287 },288 {289 "extras": null,290 "description": "Frequent Poster",291 "user": {292 "id": 80949,293 "username": "Yangqi_Long",294 "name": "Yangqi Long",295 "avatar_template": "/user_avatar/discuss.pytorch.org/yangqi_long/{size}/74032_2.png",296 "trust_level": 0297 }298 },299 {300 "extras": "latest",301 "description": "Most Recent Poster",302 "user": {303 "id": 3534,304 "username": "ptrblck",305 "name": "",306 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",307 "admin": true,308 "moderator": true,309 "trust_level": 2310 }311 }312 ]313 },314 {315 "fancy_title": "Create PackedSequence directly",316 "id": 221453,317 "title": "Create PackedSequence directly",318 "slug": "create-packedsequence-directly",319 "posts_count": 1,320 "reply_count": 0,321 "highest_post_number": 1,322 "image_url": null,323 "created_at": "2025-07-12T01:27:02.093Z",324 "last_posted_at": "2025-07-12T01:27:02.143Z",325 "bumped": true,326 "bumped_at": "2025-07-12T01:27:02.143Z",327 "archetype": "regular",328 "unseen": false,329 "pinned": false,330 "unpinned": null,331 "visible": true,332 "closed": false,333 "archived": false,334 "bookmarked": null,335 "liked": null,336 "tags_descriptions": {},337 "like_count": 0,338 "views": 24,339 "category_id": 1,340 "featured_link": null,341 "has_accepted_answer": false,342 "posters": [343 {344 "extras": "latest single",345 "description": "Original Poster, Most Recent Poster",346 "user": {347 "id": 68506,348 "username": "alexj",349 "name": "Alex Jouravlev",350 "avatar_template": "/user_avatar/discuss.pytorch.org/alexj/{size}/70445_2.png",351 "trust_level": 1352 }353 }354 ]355 }356 ],357 "tags_descriptions": {},358 "fancy_title": "RNN:RuntimeError: Assertion `cur_target >= 0 && cur_target < n_classes’ failed",359 "id": 65831,360 "title": "RNN:RuntimeError: Assertion `cur_target >= 0 && cur_target < n_classes' failed",361 "posts_count": 1,362 "created_at": "2020-01-05T18:05:52.042Z",363 "views": 330,364 "reply_count": 0,365 "like_count": 0,366 "last_posted_at": "2020-01-05T18:05:52.126Z",367 "visible": true,368 "closed": false,369 "archived": false,370 "has_summary": false,371 "archetype": "regular",372 "slug": "rnn-assertion-cur-target-0-cur-target-n-classes-failed",373 "category_id": 1,374 "word_count": 185,375 "deleted_at": null,376 "user_id": 25690,377 "featured_link": null,378 "pinned_globally": false,379 "pinned_at": null,380 "pinned_until": null,381 "image_url": null,382 "slow_mode_seconds": 0,383 "draft": null,384 "draft_key": "topic_65831",385 "draft_sequence": null,386 "unpinned": null,387 "pinned": false,388 "current_post_number": 1,389 "highest_post_number": 1,390 "deleted_by": null,391 "actions_summary": [392 {393 "id": 4,394 "count": 0,395 "hidden": false,396 "can_act": false397 },398 {399 "id": 8,400 "count": 0,401 "hidden": false,402 "can_act": false403 },404 {405 "id": 10,406 "count": 0,407 "hidden": false,408 "can_act": false409 },410 {411 "id": 7,412 "count": 0,413 "hidden": false,414 "can_act": false415 }416 ],417 "chunk_size": 20,418 "bookmarked": false,419 "topic_timer": null,420 "unicode_title": "RNN:RuntimeError: Assertion `cur_target >= 0 && cur_target < n_classes' failed",421 "message_bus_last_id": 0,422 "participant_count": 1,423 "show_read_indicator": false,424 "thumbnails": null,425 "slow_mode_enabled_until": null,426 "can_vote": false,427 "vote_count": 0,428 "user_voted": false,429 "discourse_zendesk_plugin_zendesk_id": null,430 "discourse_zendesk_plugin_zendesk_url": "https://your-url.zendesk.com/agent/tickets/",431 "details": {432 "can_edit": false,433 "notification_level": 1,434 "participants": [435 {436 "id": 25690,437 "username": "lhsICT",438 "name": "이홍석",439 "avatar_template": "/user_avatar/discuss.pytorch.org/lhsict/{size}/18877_2.png",440 "post_count": 1,441 "primary_group_name": null,442 "flair_name": null,443 "flair_url": null,444 "flair_color": null,445 "flair_bg_color": null,446 "flair_group_id": null,447 "trust_level": 1448 }449 ],450 "created_by": {451 "id": 25690,452 "username": "lhsICT",453 "name": "이홍석",454 "avatar_template": "/user_avatar/discuss.pytorch.org/lhsict/{size}/18877_2.png"455 },456 "last_poster": {457 "id": 25690,458 "username": "lhsICT",459 "name": "이홍석",460 "avatar_template": "/user_avatar/discuss.pytorch.org/lhsict/{size}/18877_2.png"461 }462 },463 "bookmarks": []464 },465 {466 "post_stream": {467 "posts": [468 {469 "id": 156450,470 "name": "achref",471 "username": "achref",472 "avatar_template": "/user_avatar/discuss.pytorch.org/achref/{size}/51659_2.png",473 "created_at": "2020-01-04T13:21:14.758Z",474 "cooked": "<p>Hello<br>\nnp.unpackbits get 24 bits for Cityscapes we reverse the order (total 19 classes) for SBD datasets we reverse the order (total 20 classes)</p>\n<p>For ADE20K that has 150 classes what i do for getting 150 bits.</p>\n<p>def _sync_transform(self, img, mask):<br>\n<span class=\"hashtag\">#hy</span> modified this function<br>\n# random crop crop_size<br>\ncrop_size = self.crop_size<br>\nw, h = img.size<br>\nx1 = random.randint(0, w - crop_size)<br>\ny1 = random.randint(0, h - crop_size)<br>\nimg = img.crop((x1, y1, x1+crop_size, y1+crop_size))<br>\nmask = mask.crop((x1, y1, x1+crop_size, y1+crop_size))<br>\n<span class=\"hashtag\">#np</span>.unpackbits get 24 bits, we extract [:,:5:] and reverse the order (total 19 classes), i.e. [:,:,-1:-20:-1]<br>\nmask = np.unpackbits(np.array(mask), axis=2)[:,:,-1:-20:-1]<br>\nmask = torch.from_numpy(np.array(mask)).float()<br>\nmask = mask.transpose(0, 1).transpose(0, 2) <span class=\"hashtag\">#channel</span> first<br>\n# return img, self._mask_transform(mask)<br>\nreturn img, mask</p>\n<p>Thank you</p>",475 "post_number": 1,476 "post_type": 1,477 "posts_count": 3,478 "updated_at": "2020-01-04T13:21:14.758Z",479 "reply_count": 0,480 "reply_to_post_number": null,481 "quote_count": 0,482 "incoming_link_count": 39,483 "reads": 3,484 "readers_count": 2,485 "score": 195.6,486 "yours": false,487 "topic_id": 65744,488 "topic_slug": "def-sync-transform-self-img-mask",489 "display_username": "achref",490 "primary_group_name": null,491 "flair_name": null,492 "flair_url": null,493 "flair_bg_color": null,494 "flair_color": null,495 "flair_group_id": null,496 "badges_granted": [],497 "version": 1,498 "can_edit": false,499 "can_delete": false,500 "can_recover": false,501 "can_see_hidden_post": false,502 "can_wiki": false,503 "read": true,504 "user_title": null,505 "bookmarked": false,506 "actions_summary": [],507 "moderator": false,508 "admin": false,509 "staff": false,510 "user_id": 26015,511 "hidden": false,512 "trust_level": 1,513 "deleted_at": null,514 "user_deleted": false,515 "edit_reason": null,516 "can_view_edit_history": true,517 "wiki": false,518 "post_url": "/t/def-sync-transform-self-img-mask/65744/1",519 "can_accept_answer": false,520 "can_unaccept_answer": false,521 "accepted_answer": false,522 "topic_accepted_answer": null,523 "can_vote": false524 },525 {526 "id": 156561,527 "name": "",528 "username": "ptrblck",529 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",530 "created_at": "2020-01-05T06:35:32.561Z",531 "cooked": "<p>As far as I see it in the <a href=\"https://docs.scipy.org/doc/numpy/reference/generated/numpy.unpackbits.html\">docs</a>, only <code>np.uint8</code> is supported by <code>np.unpackbits</code>, so you might need to use <a href=\"https://docs.scipy.org/doc/numpy/user/basics.byteswapping.html\">byte swapping</a> to get the underlying representation.</p>\n<p>What is your use case that you need to hack around the bits?</p>",532 "post_number": 2,533 "post_type": 1,534 "posts_count": 3,535 "updated_at": "2020-01-05T06:35:32.561Z",536 "reply_count": 0,537 "reply_to_post_number": null,538 "quote_count": 0,539 "incoming_link_count": 2,540 "reads": 3,541 "readers_count": 2,542 "score": 25.6,543 "yours": false,544 "topic_id": 65744,545 "topic_slug": "def-sync-transform-self-img-mask",546 "display_username": "",547 "primary_group_name": null,548 "flair_name": null,549 "flair_url": null,550 "flair_bg_color": null,551 "flair_color": null,552 "flair_group_id": null,553 "badges_granted": [],554 "version": 1,555 "can_edit": false,556 "can_delete": false,557 "can_recover": false,558 "can_see_hidden_post": false,559 "can_wiki": false,560 "link_counts": [561 {562 "url": "https://docs.scipy.org/doc/numpy/user/basics.byteswapping.html",563 "internal": false,564 "reflection": false,565 "title": "Byte-swapping — NumPy v1.17 Manual",566 "clicks": 1567 },568 {569 "url": "https://docs.scipy.org/doc/numpy/reference/generated/numpy.unpackbits.html",570 "internal": false,571 "reflection": false,572 "title": "numpy.unpackbits — NumPy v1.17 Manual",573 "clicks": 0574 }575 ],576 "read": true,577 "user_title": "",578 "bookmarked": false,579 "actions_summary": [580 {581 "id": 2,582 "count": 1583 }584 ],585 "moderator": true,586 "admin": true,587 "staff": true,588 "user_id": 3534,589 "hidden": false,590 "trust_level": 2,591 "deleted_at": null,592 "user_deleted": false,593 "edit_reason": null,594 "can_view_edit_history": true,595 "wiki": false,596 "post_url": "/t/def-sync-transform-self-img-mask/65744/2",597 "can_accept_answer": false,598 "can_unaccept_answer": false,599 "accepted_answer": false,600 "topic_accepted_answer": null601 },602 {603 "id": 156619,604 "name": "achref",605 "username": "achref",606 "avatar_template": "/user_avatar/discuss.pytorch.org/achref/{size}/51659_2.png",607 "created_at": "2020-01-05T14:52:23.862Z",608 "cooked": "<p>Hi<br>\nThank you for your response.</p>\n<p>DFF algorithm estimate for each pixel at the boundry the probabilities belongs to each class.<br>\nI’ve understood each bit if equal 1 then this pixel correspond to Class_i<br>\nbut until now it is complicate to modify mask = np.unpackbits(np.array(mask), axis=2) to extract more than 24 bits.</p>",609 "post_number": 3,610 "post_type": 1,611 "posts_count": 3,612 "updated_at": "2020-01-05T14:52:23.862Z",613 "reply_count": 0,614 "reply_to_post_number": null,615 "quote_count": 0,616 "incoming_link_count": 2,617 "reads": 2,618 "readers_count": 1,619 "score": 10.4,620 "yours": false,621 "topic_id": 65744,622 "topic_slug": "def-sync-transform-self-img-mask",623 "display_username": "achref",624 "primary_group_name": null,625 "flair_name": null,626 "flair_url": null,627 "flair_bg_color": null,628 "flair_color": null,629 "flair_group_id": null,630 "badges_granted": [],631 "version": 1,632 "can_edit": false,633 "can_delete": false,634 "can_recover": false,635 "can_see_hidden_post": false,636 "can_wiki": false,637 "read": true,638 "user_title": null,639 "bookmarked": false,640 "actions_summary": [],641 "moderator": false,642 "admin": false,643 "staff": false,644 "user_id": 26015,645 "hidden": false,646 "trust_level": 1,647 "deleted_at": null,648 "user_deleted": false,649 "edit_reason": null,650 "can_view_edit_history": true,651 "wiki": false,652 "post_url": "/t/def-sync-transform-self-img-mask/65744/3",653 "can_accept_answer": false,654 "can_unaccept_answer": false,655 "accepted_answer": false,656 "topic_accepted_answer": null657 }658 ],659 "stream": [660 156450,661 156561,662 156619663 ]664 },665 "timeline_lookup": [666 [667 1,668 2121669 ],670 [671 3,672 2120673 ]674 ],675 "suggested_topics": [676 {677 "fancy_title": "Distributed Tutorial works with mp.spawn but not torchrun",678 "id": 218381,679 "title": "Distributed Tutorial works with mp.spawn but not torchrun",680 "slug": "distributed-tutorial-works-with-mp-spawn-but-not-torchrun",681 "posts_count": 1,682 "reply_count": 0,683 "highest_post_number": 1,684 "image_url": "https://discuss.pytorch.org/uploads/default/original/3X/3/8/38bbe773a664165bad26af95d32fdf354409be61.png",685 "created_at": "2025-03-28T21:31:25.982Z",686 "last_posted_at": "2025-03-28T21:31:26.020Z",687 "bumped": true,688 "bumped_at": "2025-03-28T21:31:26.020Z",689 "archetype": "regular",690 "unseen": false,691 "pinned": false,692 "unpinned": null,693 "visible": true,694 "closed": false,695 "archived": false,696 "bookmarked": null,697 "liked": null,698 "tags_descriptions": {},699 "like_count": 0,700 "views": 26,701 "category_id": 1,702 "featured_link": null,703 "has_accepted_answer": false,704 "posters": [705 {706 "extras": "latest single",707 "description": "Original Poster, Most Recent Poster",708 "user": {709 "id": 83390,710 "username": "Marcelo_Sena",711 "name": "Marcelo Sena",712 "avatar_template": "/user_avatar/discuss.pytorch.org/marcelo_sena/{size}/76267_2.png",713 "trust_level": 0714 }715 }716 ]717 },718 {719 "fancy_title": "Pytorch binaries with cuda version 12.2/12.0",720 "id": 215078,721 "title": "Pytorch binaries with cuda version 12.2/12.0",722 "slug": "pytorch-binaries-with-cuda-version-12-2-12-0",723 "posts_count": 2,724 "reply_count": 0,725 "highest_post_number": 2,726 "image_url": null,727 "created_at": "2025-01-07T15:02:32.838Z",728 "last_posted_at": "2025-01-07T15:43:32.219Z",729 "bumped": true,730 "bumped_at": "2025-01-07T15:43:32.219Z",731 "archetype": "regular",732 "unseen": false,733 "pinned": false,734 "unpinned": null,735 "visible": true,736 "closed": false,737 "archived": false,738 "bookmarked": null,739 "liked": null,740 "tags_descriptions": {},741 "like_count": 0,742 "views": 127,743 "category_id": 1,744 "featured_link": null,745 "has_accepted_answer": false,746 "posters": [747 {748 "extras": null,749 "description": "Original Poster",750 "user": {751 "id": 81925,752 "username": "gal_kesten",753 "name": "gal kesten",754 "avatar_template": "/user_avatar/discuss.pytorch.org/gal_kesten/{size}/74948_2.png",755 "trust_level": 0756 }757 },758 {759 "extras": "latest",760 "description": "Most Recent Poster",761 "user": {762 "id": 3534,763 "username": "ptrblck",764 "name": "",765 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",766 "admin": true,767 "moderator": true,768 "trust_level": 2769 }770 }771 ]772 },773 {774 "fancy_title": "Suggestion for ML approach to mimic FEM results?",775 "id": 213993,776 "title": "Suggestion for ML approach to mimic FEM results?",777 "slug": "suggestion-for-ml-approach-to-mimic-fem-results",778 "posts_count": 1,779 "reply_count": 0,780 "highest_post_number": 1,781 "image_url": null,782 "created_at": "2024-12-09T12:34:40.329Z",783 "last_posted_at": "2024-12-09T12:34:40.382Z",784 "bumped": true,785 "bumped_at": "2024-12-09T13:19:10.610Z",786 "archetype": "regular",787 "unseen": false,788 "pinned": false,789 "unpinned": null,790 "visible": true,791 "closed": false,792 "archived": false,793 "bookmarked": null,794 "liked": null,795 "tags_descriptions": {},796 "like_count": 0,797 "views": 46,798 "category_id": 1,799 "featured_link": null,800 "has_accepted_answer": false,801 "posters": [802 {803 "extras": "latest single",804 "description": "Original Poster, Most Recent Poster",805 "user": {806 "id": 81403,807 "username": "LainuUrdin",808 "name": "Laino Urdin",809 "avatar_template": "/user_avatar/discuss.pytorch.org/lainuurdin/{size}/74437_2.png",810 "trust_level": 0811 }812 }813 ]814 },815 {816 "fancy_title": "[CPU] Train network using float 16?",817 "id": 213486,818 "title": "[CPU] Train network using float 16?",819 "slug": "cpu-train-network-using-float-16",820 "posts_count": 2,821 "reply_count": 2,822 "highest_post_number": 2,823 "image_url": null,824 "created_at": "2024-11-26T19:22:22.622Z",825 "last_posted_at": "2024-11-26T21:13:26.532Z",826 "bumped": true,827 "bumped_at": "2024-11-27T03:07:04.265Z",828 "archetype": "regular",829 "unseen": false,830 "pinned": false,831 "unpinned": null,832 "visible": true,833 "closed": false,834 "archived": false,835 "bookmarked": null,836 "liked": null,837 "tags_descriptions": {},838 "like_count": 1,839 "views": 457,840 "category_id": 1,841 "featured_link": null,842 "has_accepted_answer": true,843 "posters": [844 {845 "extras": null,846 "description": "Original Poster",847 "user": {848 "id": 81089,849 "username": "Aknw_Fen",850 "name": "Aknw Fen",851 "avatar_template": "/user_avatar/discuss.pytorch.org/aknw_fen/{size}/74156_2.png",852 "trust_level": 2853 }854 },855 {856 "extras": "latest",857 "description": "Most Recent Poster, Accepted Answer",858 "user": {859 "id": 3534,860 "username": "ptrblck",861 "name": "",862 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",863 "admin": true,864 "moderator": true,865 "trust_level": 2866 }867 }868 ]869 },870 {871 "fancy_title": "“turing_fp16_s1688gemm_fp16_128x128_ldg8_relu_f2f_tn”",872 "id": 219212,873 "title": "\"turing_fp16_s1688gemm_fp16_128x128_ldg8_relu_f2f_tn\"",874 "slug": "turing-fp16-s1688gemm-fp16-128x128-ldg8-relu-f2f-tn",875 "posts_count": 6,876 "reply_count": 4,877 "highest_post_number": 6,878 "image_url": null,879 "created_at": "2025-04-17T16:50:15.925Z",880 "last_posted_at": "2025-04-21T12:41:41.181Z",881 "bumped": true,882 "bumped_at": "2025-04-21T12:41:41.181Z",883 "archetype": "regular",884 "unseen": false,885 "pinned": false,886 "unpinned": null,887 "visible": true,888 "closed": false,889 "archived": false,890 "bookmarked": null,891 "liked": null,892 "tags_descriptions": {},893 "like_count": 0,894 "views": 132,895 "category_id": 1,896 "featured_link": null,897 "has_accepted_answer": false,898 "posters": [899 {900 "extras": null,901 "description": "Original Poster",902 "user": {903 "id": 83873,904 "username": "JB13",905 "name": null,906 "avatar_template": "/letter_avatar_proxy/v4/letter/j/71c47a/{size}.png",907 "trust_level": 0908 }909 },910 {911 "extras": "latest",912 "description": "Most Recent Poster",913 "user": {914 "id": 3534,915 "username": "ptrblck",916 "name": "",917 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",918 "admin": true,919 "moderator": true,920 "trust_level": 2921 }922 }923 ]924 }925 ],926 "tags_descriptions": {},927 "fancy_title": "Def _sync_transform(self, img, mask):",928 "id": 65744,929 "title": "Def _sync_transform(self, img, mask):",930 "posts_count": 3,931 "created_at": "2020-01-04T13:21:14.664Z",932 "views": 397,933 "reply_count": 0,934 "like_count": 1,935 "last_posted_at": "2020-01-05T14:52:23.862Z",936 "visible": true,937 "closed": false,938 "archived": false,939 "has_summary": false,940 "archetype": "regular",941 "slug": "def-sync-transform-self-img-mask",942 "category_id": 1,943 "word_count": 266,944 "deleted_at": null,945 "user_id": 26015,946 "featured_link": null,947 "pinned_globally": false,948 "pinned_at": null,949 "pinned_until": null,950 "image_url": null,951 "slow_mode_seconds": 0,952 "draft": null,953 "draft_key": "topic_65744",954 "draft_sequence": null,955 "unpinned": null,956 "pinned": false,957 "current_post_number": 1,958 "highest_post_number": 3,959 "deleted_by": null,960 "actions_summary": [961 {962 "id": 4,963 "count": 0,964 "hidden": false,965 "can_act": false966 },967 {968 "id": 8,969 "count": 0,970 "hidden": false,971 "can_act": false972 },973 {974 "id": 10,975 "count": 0,976 "hidden": false,977 "can_act": false978 },979 {980 "id": 7,981 "count": 0,982 "hidden": false,983 "can_act": false984 }985 ],986 "chunk_size": 20,987 "bookmarked": false,988 "topic_timer": null,989 "message_bus_last_id": 0,990 "participant_count": 2,991 "show_read_indicator": false,992 "thumbnails": null,993 "slow_mode_enabled_until": null,994 "can_vote": false,995 "vote_count": 0,996 "user_voted": false,997 "discourse_zendesk_plugin_zendesk_id": null,998 "discourse_zendesk_plugin_zendesk_url": "https://your-url.zendesk.com/agent/tickets/",999 "details": {1000 "can_edit": false,1001 "notification_level": 1,1002 "participants": [1003 {1004 "id": 26015,1005 "username": "achref",1006 "name": "achref",1007 "avatar_template": "/user_avatar/discuss.pytorch.org/achref/{size}/51659_2.png",1008 "post_count": 2,1009 "primary_group_name": null,1010 "flair_name": null,1011 "flair_url": null,1012 "flair_color": null,1013 "flair_bg_color": null,1014 "flair_group_id": null,1015 "trust_level": 11016 },1017 {1018 "id": 3534,1019 "username": "ptrblck",1020 "name": "",1021 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",1022 "post_count": 1,1023 "primary_group_name": null,1024 "flair_name": null,1025 "flair_url": null,1026 "flair_color": null,1027 "flair_bg_color": null,1028 "flair_group_id": null,1029 "admin": true,1030 "moderator": true,1031 "trust_level": 21032 }1033 ],1034 "created_by": {1035 "id": 26015,1036 "username": "achref",1037 "name": "achref",1038 "avatar_template": "/user_avatar/discuss.pytorch.org/achref/{size}/51659_2.png"1039 },1040 "last_poster": {1041 "id": 26015,1042 "username": "achref",1043 "name": "achref",1044 "avatar_template": "/user_avatar/discuss.pytorch.org/achref/{size}/51659_2.png"1045 },1046 "links": [1047 {1048 "url": "https://docs.scipy.org/doc/numpy/user/basics.byteswapping.html",1049 "title": "Byte-swapping — NumPy v1.17 Manual",1050 "internal": false,1051 "attachment": false,1052 "reflection": false,1053 "clicks": 1,1054 "user_id": 3534,1055 "domain": "docs.scipy.org",1056 "root_domain": "scipy.org"1057 }1058 ]1059 },1060 "bookmarks": []1061 },1062 {1063 "post_stream": {1064 "posts": [1065 {1066 "id": 156590,1067 "name": "A curious guy here!",1068 "username": "Shisho_Sama",1069 "avatar_template": "/user_avatar/discuss.pytorch.org/shisho_sama/{size}/6926_2.png",1070 "created_at": "2020-01-05T10:12:07.740Z",1071 "cooked": "<p>Hello everyone, hope you are having a great day.<br>\nI’m curious to know whether Pytorch (as of latest version) have support for <strong>Fused</strong> <strong>BatchNormalization</strong>.<br>\nBasically <code>FusedBatchNormalization</code> is simply the fusion of <code>BatchNormalization</code> into precdeding convolutional neural network since, the parameters after training are fixed and can thus be used as constants.<br>\nBased on tensorflow’s documentation, it provides 12% to 30% boost of performance at inference time which is a considerable gain. <a href=\"https://web.archive.org/web/20180620042504/https://www.tensorflow.org/performance/performance_guide\" rel=\"noopener nofollow ugc\">Link</a></p>\n<blockquote>\n<p>Fused batch norm combines the multiple operations needed to do batch normalization into a single kernel. Batch norm is an expensive process that for some models makes up a large percentage of the operation time. Using fused batch norm can result in a 12%-30% speedup.</p>\n<p>There are two commonly used batch norms and both support fusing. The core <a href=\"https://web.archive.org/web/20180620042504/https://www.tensorflow.org/api_docs/python/tf/layers/batch_normalization\" rel=\"noopener nofollow ugc\"> <code>tf.layers.batch_normalization</code> </a> added fused starting in TensorFlow 1.3.</p>\n</blockquote>\n<p>I know we had a Pr back in 2017 which was rejected. but I dont know if we have it implemented or not!<br>\nAny update in this regard is greatly appreciated.</p>\n<p><strong>Update :</strong><br>\nHere is a Pytorch implementation from <strong>Intel’s</strong> <strong>NervanaSystems</strong> : <a href=\"https://github.com/NervanaSystems/distiller/blob/4717596112b70600bb3ac54c7a23a55abfea113e/distiller/model_transforms.py#L102\" rel=\"noopener nofollow ugc\">Folded_batch_normalization</a></p>",1072 "post_number": 1,1073 "post_type": 1,1074 "posts_count": 2,1075 "updated_at": "2020-01-05T10:55:11.378Z",1076 "reply_count": 0,1077 "reply_to_post_number": null,1078 "quote_count": 0,1079 "incoming_link_count": 2946,1080 "reads": 108,1081 "readers_count": 107,1082 "score": 14756.6,1083 "yours": false,1084 "topic_id": 65801,1085 "topic_slug": "does-pytorch-support-fused-folded-batchnormalization",1086 "display_username": "A curious guy here!",1087 "primary_group_name": null,1088 "flair_name": null,1089 "flair_url": null,1090 "flair_bg_color": null,1091 "flair_color": null,1092 "flair_group_id": null,1093 "badges_granted": [],1094 "version": 2,1095 "can_edit": false,1096 "can_delete": false,1097 "can_recover": false,1098 "can_see_hidden_post": false,1099 "can_wiki": false,1100 "link_counts": [1101 {1102 "url": "https://github.com/NervanaSystems/distiller/blob/4717596112b70600bb3ac54c7a23a55abfea113e/distiller/model_transforms.py#L102",1103 "internal": false,1104 "reflection": false,1105 "title": "distiller/model_transforms.py at 4717596112b70600bb3ac54c7a23a55abfea113e · NervanaSystems/distiller · GitHub",1106 "clicks": 1371107 },1108 {1109 "url": "https://web.archive.org/web/20180620042504/https://www.tensorflow.org/performance/performance_guide",1110 "internal": false,1111 "reflection": false,1112 "title": "Performance Guide | TensorFlow",1113 "clicks": 311114 },1115 {1116 "url": "https://web.archive.org/web/20180620042504/https://www.tensorflow.org/api_docs/python/tf/layers/batch_normalization",1117 "internal": false,1118 "reflection": false,1119 "clicks": 251120 }1121 ],1122 "read": true,1123 "user_title": "",1124 "bookmarked": false,1125 "actions_summary": [1126 {1127 "id": 2,1128 "count": 11129 }1130 ],1131 "moderator": false,1132 "admin": false,1133 "staff": false,1134 "user_id": 7863,1135 "hidden": false,1136 "trust_level": 2,1137 "deleted_at": null,1138 "user_deleted": false,1139 "edit_reason": null,1140 "can_view_edit_history": true,1141 "wiki": false,1142 "post_url": "/t/does-pytorch-support-fused-folded-batchnormalization/65801/1",1143 "can_accept_answer": false,1144 "can_unaccept_answer": false,1145 "accepted_answer": false,1146 "topic_accepted_answer": true,1147 "can_vote": false1148 },1149 {1150 "id": 156618,1151 "name": "Vadim Kantorov",1152 "username": "vadimkantorov",1153 "avatar_template": "/user_avatar/discuss.pytorch.org/vadimkantorov/{size}/365_2.png",1154 "created_at": "2020-01-05T14:52:10.911Z",1155 "cooked": "<p>PyTorch currently has <code>torch.nn.utils.fuse_conv_bn_eval</code> (and I think some ConvBnRelu fusion with kernels only in quantized setting) which you must call manually (for details look at the source code, it’s really simple code), but it’s <a href=\"https://github.com/pytorch/pytorch/issues/28757\" rel=\"nofollow noopener\">not documented yet</a></p>",1156 "post_number": 2,1157 "post_type": 1,1158 "posts_count": 2,1159 "updated_at": "2020-01-07T02:43:28.371Z",1160 "reply_count": 0,1161 "reply_to_post_number": null,1162 "quote_count": 0,1163 "incoming_link_count": 54,1164 "reads": 102,1165 "readers_count": 101,1166 "score": 350.4,1167 "yours": false,1168 "topic_id": 65801,1169 "topic_slug": "does-pytorch-support-fused-folded-batchnormalization",1170 "display_username": "Vadim Kantorov",1171 "primary_group_name": null,1172 "flair_name": null,1173 "flair_url": null,1174 "flair_bg_color": null,1175 "flair_color": null,1176 "flair_group_id": null,1177 "badges_granted": [],1178 "version": 1,1179 "can_edit": false,1180 "can_delete": false,1181 "can_recover": false,1182 "can_see_hidden_post": false,1183 "can_wiki": false,1184 "link_counts": [1185 {1186 "url": "https://github.com/pytorch/pytorch/issues/28757",1187 "internal": false,1188 "reflection": false,1189 "title": "[feature request] Docs for fuse_conv_bn_eval · Issue #28757 · pytorch/pytorch · GitHub",1190 "clicks": 3831191 }1192 ],1193 "read": true,1194 "user_title": null,1195 "bookmarked": false,1196 "actions_summary": [1197 {1198 "id": 2,1199 "count": 21200 }