Anurag1734/cuda-error-resolution-analysis
07
1[2 {3 "post_stream": {4 "posts": [5 {6 "id": 369644,7 "name": "Rami Ismael",8 "username": "Rami_Ismael",9 "avatar_template": "/user_avatar/discuss.pytorch.org/rami_ismael/{size}/32633_2.png",10 "created_at": "2022-10-09T16:38:29.776Z",11 "cooked": "<p>My goal is to quantize the weight of the model and not to quantize the activation function in QAT. I went through the Pytorch GitHub. However, it gives me an error<br>\nAttributeError: ‘Identity’ object has no attribute ‘calculate_qparams’</p>",12 "post_number": 1,13 "post_type": 1,14 "posts_count": 2,15 "updated_at": "2022-10-09T16:38:29.776Z",16 "reply_count": 0,17 "reply_to_post_number": null,18 "quote_count": 0,19 "incoming_link_count": 333,20 "reads": 9,21 "readers_count": 8,22 "score": 1656.8,23 "yours": false,24 "topic_id": 163141,25 "topic_slug": "torch-quantization-convert-does-not-work-if-my-model-has-qconfig-activation-is-set-to-torch-nn-identity",26 "display_username": "Rami Ismael",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": 40341,48 "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/torch-quantization-convert-does-not-work-if-my-model-has-qconfig-activation-is-set-to-torch-nn-identity/163141/1",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 "id": 369767,64 "name": "Hd Charles",65 "username": "HDCharles",66 "avatar_template": "/user_avatar/discuss.pytorch.org/hdcharles/{size}/38408_2.png",67 "created_at": "2022-10-10T16:05:08.634Z",68 "cooked": "<p>you can’t just use identity because the quantization flow uses information from the qconfig to determine a number of things and makes assumptions about the types of objects in the qconfig, you should use the actual weight only qconfig instead</p>\n<p><a href=\"https://pytorch.org/docs/stable/generated/torch.quantization.qconfig.default_weight_only_qconfig.html?highlight=weight_only_qconfig#torch.quantization.qconfig.default_weight_only_qconfig\" rel=\"noopener nofollow ugc\">float_qparams_weight_only_qconfig</a></p>",69 "post_number": 2,70 "post_type": 1,71 "posts_count": 2,72 "updated_at": "2022-10-10T16:06:43.941Z",73 "reply_count": 0,74 "reply_to_post_number": null,75 "quote_count": 0,76 "incoming_link_count": 17,77 "reads": 9,78 "readers_count": 8,79 "score": 86.8,80 "yours": false,81 "topic_id": 163141,82 "topic_slug": "torch-quantization-convert-does-not-work-if-my-model-has-qconfig-activation-is-set-to-torch-nn-identity",83 "display_username": "Hd Charles",84 "primary_group_name": null,85 "flair_name": null,86 "flair_url": null,87 "flair_bg_color": null,88 "flair_color": null,89 "flair_group_id": null,90 "badges_granted": [],91 "version": 1,92 "can_edit": false,93 "can_delete": false,94 "can_recover": false,95 "can_see_hidden_post": false,96 "can_wiki": false,97 "link_counts": [98 {99 "url": "https://pytorch.org/docs/stable/generated/torch.quantization.qconfig.default_weight_only_qconfig.html?highlight=weight_only_qconfig#torch.quantization.qconfig.default_weight_only_qconfig",100 "internal": false,101 "reflection": false,102 "title": "default_weight_only_qconfig — PyTorch 1.12 documentation",103 "clicks": 65104 }105 ],106 "read": true,107 "user_title": null,108 "bookmarked": false,109 "actions_summary": [],110 "moderator": false,111 "admin": false,112 "staff": false,113 "user_id": 45475,114 "hidden": false,115 "trust_level": 2,116 "deleted_at": null,117 "user_deleted": false,118 "edit_reason": null,119 "can_view_edit_history": true,120 "wiki": false,121 "post_url": "/t/torch-quantization-convert-does-not-work-if-my-model-has-qconfig-activation-is-set-to-torch-nn-identity/163141/2",122 "can_accept_answer": false,123 "can_unaccept_answer": false,124 "accepted_answer": false,125 "topic_accepted_answer": null126 }127 ],128 "stream": [129 369644,130 369767131 ]132 },133 "timeline_lookup": [134 [135 1,136 1112137 ],138 [139 2,140 1111141 ]142 ],143 "suggested_topics": [144 {145 "fancy_title": "pytorch quantized linear function gives shape invalid error",146 "id": 212411,147 "title": "pytorch quantized linear function gives shape invalid error",148 "slug": "pytorch-quantized-linear-function-gives-shape-invalid-error",149 "posts_count": 4,150 "reply_count": 1,151 "highest_post_number": 4,152 "image_url": null,153 "created_at": "2024-11-01T14:02:14.096Z",154 "last_posted_at": "2024-11-01T16:11:37.322Z",155 "bumped": true,156 "bumped_at": "2024-11-01T16:11:37.322Z",157 "archetype": "regular",158 "unseen": false,159 "pinned": false,160 "unpinned": null,161 "visible": true,162 "closed": false,163 "archived": false,164 "bookmarked": null,165 "liked": null,166 "tags_descriptions": {},167 "like_count": 1,168 "views": 242,169 "category_id": 17,170 "featured_link": null,171 "has_accepted_answer": false,172 "posters": [173 {174 "extras": null,175 "description": "Original Poster",176 "user": {177 "id": 80593,178 "username": "hafezmg48",179 "name": "Hafez",180 "avatar_template": "/letter_avatar_proxy/v4/letter/h/b77776/{size}.png",181 "trust_level": 1182 }183 },184 {185 "extras": "latest",186 "description": "Most Recent Poster",187 "user": {188 "id": 31938,189 "username": "Vasiliy_Kuznetsov",190 "name": "Vasiliy Kuznetsov",191 "avatar_template": "/user_avatar/discuss.pytorch.org/vasiliy_kuznetsov/{size}/24530_2.png",192 "trust_level": 2193 }194 }195 ]196 },197 {198 "fancy_title": "Pytorch 2 Export QAT is training",199 "id": 214389,200 "title": "Pytorch 2 Export QAT is training",201 "slug": "pytorch-2-export-qat-is-training",202 "posts_count": 1,203 "reply_count": 0,204 "highest_post_number": 1,205 "image_url": null,206 "created_at": "2024-12-19T05:17:39.134Z",207 "last_posted_at": "2024-12-19T05:17:39.196Z",208 "bumped": true,209 "bumped_at": "2024-12-19T05:17:39.196Z",210 "archetype": "regular",211 "unseen": false,212 "pinned": false,213 "unpinned": null,214 "visible": true,215 "closed": false,216 "archived": false,217 "bookmarked": null,218 "liked": null,219 "tags_descriptions": {},220 "like_count": 0,221 "views": 163,222 "category_id": 17,223 "featured_link": null,224 "has_accepted_answer": false,225 "posters": [226 {227 "extras": "latest single",228 "description": "Original Poster, Most Recent Poster",229 "user": {230 "id": 34929,231 "username": "PavanMV",232 "name": "Pavan MV",233 "avatar_template": "/user_avatar/discuss.pytorch.org/pavanmv/{size}/27801_2.png",234 "trust_level": 1235 }236 }237 ]238 },239 {240 "fancy_title": "Inserting Unnecessary Fake Quants during QAT?",241 "id": 212750,242 "title": "Inserting Unnecessary Fake Quants during QAT?",243 "slug": "inserting-unnecessary-fake-quants-during-qat",244 "posts_count": 3,245 "reply_count": 0,246 "highest_post_number": 3,247 "image_url": "https://discuss.pytorch.org/uploads/default/optimized/3X/4/1/4119adbe4a32a307a9900af174eb1f1d3ecb9957_2_1024x283.png",248 "created_at": "2024-11-09T16:30:00.060Z",249 "last_posted_at": "2024-11-12T23:59:29.531Z",250 "bumped": true,251 "bumped_at": "2024-11-12T23:59:29.531Z",252 "archetype": "regular",253 "unseen": false,254 "pinned": false,255 "unpinned": null,256 "visible": true,257 "closed": false,258 "archived": false,259 "bookmarked": null,260 "liked": null,261 "tags_descriptions": {},262 "like_count": 1,263 "views": 311,264 "category_id": 17,265 "featured_link": null,266 "has_accepted_answer": true,267 "posters": [268 {269 "extras": "latest",270 "description": "Original Poster, Most Recent Poster",271 "user": {272 "id": 80791,273 "username": "jsenn",274 "name": null,275 "avatar_template": "/letter_avatar_proxy/v4/letter/j/f475e1/{size}.png",276 "trust_level": 1277 }278 },279 {280 "extras": null,281 "description": "Frequent Poster, Accepted Answer",282 "user": {283 "id": 45475,284 "username": "HDCharles",285 "name": "Hd Charles",286 "avatar_template": "/user_avatar/discuss.pytorch.org/hdcharles/{size}/38408_2.png",287 "trust_level": 2288 }289 }290 ]291 },292 {293 "fancy_title": "Exclude specific layers from quantization",294 "id": 217097,295 "title": "Exclude specific layers from quantization",296 "slug": "exclude-specific-layers-from-quantization",297 "posts_count": 2,298 "reply_count": 0,299 "highest_post_number": 2,300 "image_url": null,301 "created_at": "2025-02-24T13:57:23.441Z",302 "last_posted_at": "2025-02-24T21:36:30.974Z",303 "bumped": true,304 "bumped_at": "2025-02-24T21:36:30.974Z",305 "archetype": "regular",306 "unseen": false,307 "pinned": false,308 "unpinned": null,309 "visible": true,310 "closed": false,311 "archived": false,312 "bookmarked": null,313 "liked": null,314 "tags_descriptions": {},315 "like_count": 0,316 "views": 133,317 "category_id": 17,318 "featured_link": null,319 "has_accepted_answer": false,320 "posters": [321 {322 "extras": null,323 "description": "Original Poster",324 "user": {325 "id": 82734,326 "username": "jpiat",327 "name": "Jonathan Piat",328 "avatar_template": "/user_avatar/discuss.pytorch.org/jpiat/{size}/75700_2.png",329 "trust_level": 1330 }331 },332 {333 "extras": "latest",334 "description": "Most Recent Poster",335 "user": {336 "id": 21770,337 "username": "jerryzh168",338 "name": "Jerry Zhang",339 "avatar_template": "/user_avatar/discuss.pytorch.org/jerryzh168/{size}/15217_2.png",340 "trust_level": 2341 }342 }343 ]344 },345 {346 "fancy_title": "Quantized LLM inference vs quantized matrix multiplication speed in CPU",347 "id": 217166,348 "title": "Quantized LLM inference vs quantized matrix multiplication speed in CPU",349 "slug": "quantized-llm-inference-vs-quantized-matrix-multiplication-speed-in-cpu",350 "posts_count": 3,351 "reply_count": 0,352 "highest_post_number": 3,353 "image_url": null,354 "created_at": "2025-02-26T03:29:32.009Z",355 "last_posted_at": "2025-04-06T00:11:53.424Z",356 "bumped": true,357 "bumped_at": "2025-04-06T00:11:53.424Z",358 "archetype": "regular",359 "unseen": false,360 "pinned": false,361 "unpinned": null,362 "visible": true,363 "closed": false,364 "archived": false,365 "bookmarked": null,366 "liked": null,367 "tags_descriptions": {},368 "like_count": 0,369 "views": 110,370 "category_id": 17,371 "featured_link": null,372 "has_accepted_answer": false,373 "posters": [374 {375 "extras": null,376 "description": "Original Poster",377 "user": {378 "id": 82918,379 "username": "luentong",380 "name": "Entong Lu",381 "avatar_template": "/user_avatar/discuss.pytorch.org/luentong/{size}/75863_2.png",382 "trust_level": 1383 }384 },385 {386 "extras": "latest",387 "description": "Most Recent Poster",388 "user": {389 "id": 21770,390 "username": "jerryzh168",391 "name": "Jerry Zhang",392 "avatar_template": "/user_avatar/discuss.pytorch.org/jerryzh168/{size}/15217_2.png",393 "trust_level": 2394 }395 }396 ]397 }398 ],399 "tags_descriptions": {},400 "fancy_title": "Torch.quantization.convert does not work if my model has QConfig activation is set to torch.nn.Identity",401 "id": 163141,402 "title": "Torch.quantization.convert does not work if my model has QConfig activation is set to torch.nn.Identity",403 "posts_count": 2,404 "created_at": "2022-10-09T16:38:29.701Z",405 "views": 866,406 "reply_count": 0,407 "like_count": 0,408 "last_posted_at": "2022-10-10T16:05:08.634Z",409 "visible": true,410 "closed": false,411 "archived": false,412 "has_summary": false,413 "archetype": "regular",414 "slug": "torch-quantization-convert-does-not-work-if-my-model-has-qconfig-activation-is-set-to-torch-nn-identity",415 "category_id": 17,416 "word_count": 97,417 "deleted_at": null,418 "user_id": 40341,419 "featured_link": null,420 "pinned_globally": false,421 "pinned_at": null,422 "pinned_until": null,423 "image_url": null,424 "slow_mode_seconds": 0,425 "draft": null,426 "draft_key": "topic_163141",427 "draft_sequence": null,428 "unpinned": null,429 "pinned": false,430 "current_post_number": 1,431 "highest_post_number": 2,432 "deleted_by": null,433 "actions_summary": [434 {435 "id": 4,436 "count": 0,437 "hidden": false,438 "can_act": false439 },440 {441 "id": 8,442 "count": 0,443 "hidden": false,444 "can_act": false445 },446 {447 "id": 10,448 "count": 0,449 "hidden": false,450 "can_act": false451 },452 {453 "id": 7,454 "count": 0,455 "hidden": false,456 "can_act": false457 }458 ],459 "chunk_size": 20,460 "bookmarked": false,461 "topic_timer": null,462 "message_bus_last_id": 0,463 "participant_count": 2,464 "show_read_indicator": false,465 "thumbnails": null,466 "slow_mode_enabled_until": null,467 "can_vote": false,468 "vote_count": 0,469 "user_voted": false,470 "discourse_zendesk_plugin_zendesk_id": null,471 "discourse_zendesk_plugin_zendesk_url": "https://your-url.zendesk.com/agent/tickets/",472 "details": {473 "can_edit": false,474 "notification_level": 1,475 "participants": [476 {477 "id": 40341,478 "username": "Rami_Ismael",479 "name": "Rami Ismael",480 "avatar_template": "/user_avatar/discuss.pytorch.org/rami_ismael/{size}/32633_2.png",481 "post_count": 1,482 "primary_group_name": null,483 "flair_name": null,484 "flair_url": null,485 "flair_color": null,486 "flair_bg_color": null,487 "flair_group_id": null,488 "trust_level": 1489 },490 {491 "id": 45475,492 "username": "HDCharles",493 "name": "Hd Charles",494 "avatar_template": "/user_avatar/discuss.pytorch.org/hdcharles/{size}/38408_2.png",495 "post_count": 1,496 "primary_group_name": null,497 "flair_name": null,498 "flair_url": null,499 "flair_color": null,500 "flair_bg_color": null,501 "flair_group_id": null,502 "trust_level": 2503 }504 ],505 "created_by": {506 "id": 40341,507 "username": "Rami_Ismael",508 "name": "Rami Ismael",509 "avatar_template": "/user_avatar/discuss.pytorch.org/rami_ismael/{size}/32633_2.png"510 },511 "last_poster": {512 "id": 45475,513 "username": "HDCharles",514 "name": "Hd Charles",515 "avatar_template": "/user_avatar/discuss.pytorch.org/hdcharles/{size}/38408_2.png"516 },517 "links": [518 {519 "url": "https://pytorch.org/docs/stable/generated/torch.quantization.qconfig.default_weight_only_qconfig.html?highlight=weight_only_qconfig#torch.quantization.qconfig.default_weight_only_qconfig",520 "title": "default_weight_only_qconfig — PyTorch 1.12 documentation",521 "internal": false,522 "attachment": false,523 "reflection": false,524 "clicks": 65,525 "user_id": 45475,526 "domain": "pytorch.org",527 "root_domain": "pytorch.org"528 }529 ]530 },531 "bookmarks": []532 },533 {534 "post_stream": {535 "posts": [536 {537 "id": 368425,538 "name": "yuri",539 "username": "yuri123",540 "avatar_template": "/user_avatar/discuss.pytorch.org/yuri123/{size}/53013_2.png",541 "created_at": "2022-10-01T06:19:42.358Z",542 "cooked": "<p>Hi,</p>\n<p>I’m studying distribted training by using pytorchDDP.<br>\nI have a question about dataparallel. when GPUs communicate with each other(all_reduce), do they exchange gradients or parameters? I thought It was gradients.</p>\n<p>But, I set the bucket size to 25mb when I use pytorchDDP, I can check the Avg size is 92mb, not 25mb. I think that 92mb is the transferred model parameters size that is divided by buckets. what is right?</p>\n<p>(I used VGG16 model that is about 528mb, and I profiled 10 step. so Calls per step of all_reduce is 6)</p>\n<p><div class=\"lightbox-wrapper\"><a class=\"lightbox\" href=\"https://discuss.pytorch.org/uploads/default/original/3X/f/4/f40fd0dfa84863068d621de217b8fb73bfeb89a5.png\" data-download-href=\"https://discuss.pytorch.org/uploads/default/f40fd0dfa84863068d621de217b8fb73bfeb89a5\" title=\"25\"><img src=\"https://discuss.pytorch.org/uploads/default/optimized/3X/f/4/f40fd0dfa84863068d621de217b8fb73bfeb89a5_2_690x66.png\" alt=\"25\" data-base62-sha1=\"yP4o4lcNMZ6GXTRmRIAUk3hxHhP\" width=\"690\" height=\"66\" srcset=\"https://discuss.pytorch.org/uploads/default/optimized/3X/f/4/f40fd0dfa84863068d621de217b8fb73bfeb89a5_2_690x66.png, https://discuss.pytorch.org/uploads/default/optimized/3X/f/4/f40fd0dfa84863068d621de217b8fb73bfeb89a5_2_1035x99.png 1.5x, https://discuss.pytorch.org/uploads/default/optimized/3X/f/4/f40fd0dfa84863068d621de217b8fb73bfeb89a5_2_1380x132.png 2x\" data-dominant-color=\"F2F3F5\"><div class=\"meta\"><svg class=\"fa d-icon d-icon-far-image svg-icon\" aria-hidden=\"true\"><use href=\"#far-image\"></use></svg><span class=\"filename\">25</span><span class=\"informations\">1462×140 10.8 KB</span><svg class=\"fa d-icon d-icon-discourse-expand svg-icon\" aria-hidden=\"true\"><use href=\"#discourse-expand\"></use></svg></div></a></div></p>\n<p>and I want to know the relationship of model parameter size and gradient size.<br>\nThank you.</p>",543 "post_number": 1,544 "post_type": 1,545 "posts_count": 4,546 "updated_at": "2022-10-01T06:36:40.778Z",547 "reply_count": 0,548 "reply_to_post_number": null,549 "quote_count": 0,550 "incoming_link_count": 52,551 "reads": 8,552 "readers_count": 7,553 "score": 261.6,554 "yours": false,555 "topic_id": 162575,556 "topic_slug": "gradients-or-parameters-exchange",557 "display_username": "yuri",558 "primary_group_name": null,559 "flair_name": null,560 "flair_url": null,561 "flair_bg_color": null,562 "flair_color": null,563 "flair_group_id": null,564 "badges_granted": [],565 "version": 2,566 "can_edit": false,567 "can_delete": false,568 "can_recover": false,569 "can_see_hidden_post": false,570 "can_wiki": false,571 "link_counts": [572 {573 "url": "https://discuss.pytorch.org/uploads/default/original/3X/f/4/f40fd0dfa84863068d621de217b8fb73bfeb89a5.png",574 "internal": true,575 "reflection": false,576 "clicks": 0577 }578 ],579 "read": true,580 "user_title": null,581 "bookmarked": false,582 "actions_summary": [],583 "moderator": false,584 "admin": false,585 "staff": false,586 "user_id": 59131,587 "hidden": false,588 "trust_level": 1,589 "deleted_at": null,590 "user_deleted": false,591 "edit_reason": null,592 "can_view_edit_history": true,593 "wiki": false,594 "post_url": "/t/gradients-or-parameters-exchange/162575/1",595 "can_accept_answer": false,596 "can_unaccept_answer": false,597 "accepted_answer": false,598 "topic_accepted_answer": null,599 "can_vote": false600 },601 {602 "id": 368903,603 "name": "Andrew Gu",604 "username": "agu",605 "avatar_template": "/user_avatar/discuss.pytorch.org/agu/{size}/49913_2.png",606 "created_at": "2022-10-04T19:52:36.994Z",607 "cooked": "<p>I saw your <a href=\"https://discuss.pytorch.org/t/pytorch-ddp-bucket-size-does-not-match-model-size/162420/3\">comment</a> in the other thread, but could you provide a minimal script to reproduce the behavior?</p>",608 "post_number": 2,609 "post_type": 1,610 "posts_count": 4,611 "updated_at": "2022-10-04T19:52:36.994Z",612 "reply_count": 1,613 "reply_to_post_number": null,614 "quote_count": 0,615 "incoming_link_count": 4,616 "reads": 5,617 "readers_count": 4,618 "score": 26.0,619 "yours": false,620 "topic_id": 162575,621 "topic_slug": "gradients-or-parameters-exchange",622 "display_username": "Andrew Gu",623 "primary_group_name": null,624 "flair_name": null,625 "flair_url": null,626 "flair_bg_color": null,627 "flair_color": null,628 "flair_group_id": null,629 "badges_granted": [],630 "version": 1,631 "can_edit": false,632 "can_delete": false,633 "can_recover": false,634 "can_see_hidden_post": false,635 "can_wiki": false,636 "link_counts": [637 {638 "url": "https://discuss.pytorch.org/t/pytorch-ddp-bucket-size-does-not-match-model-size/162420/3",639 "internal": true,640 "reflection": false,641 "title": "Pytorch DDP bucket size does not match model size",642 "clicks": 3643 }644 ],645 "read": true,646 "user_title": null,647 "bookmarked": false,648 "actions_summary": [],649 "moderator": false,650 "admin": false,651 "staff": false,652 "user_id": 49515,653 "hidden": false,654 "trust_level": 2,655 "deleted_at": null,656 "user_deleted": false,657 "edit_reason": null,658 "can_view_edit_history": true,659 "wiki": false,660 "post_url": "/t/gradients-or-parameters-exchange/162575/2",661 "can_accept_answer": false,662 "can_unaccept_answer": false,663 "accepted_answer": false,664 "topic_accepted_answer": null665 },666 {667 "id": 369748,668 "name": "yuri",669 "username": "yuri123",670 "avatar_template": "/user_avatar/discuss.pytorch.org/yuri123/{size}/53013_2.png",671 "created_at": "2022-10-10T13:54:33.868Z",672 "cooked": "<p>Hello Andrew,</p>\n<p>I’m sorry but I can’t give you the script. but I’m just wondering why the bucket size and Avg size(of transferred data) are different. I thought the bucket size and Avg size would be the same. I want to know what point I thought wrong.<br>\nThank you.</p>",673 "post_number": 3,674 "post_type": 1,675 "posts_count": 4,676 "updated_at": "2022-10-10T13:54:33.868Z",677 "reply_count": 0,678 "reply_to_post_number": 2,679 "quote_count": 0,680 "incoming_link_count": 1,681 "reads": 5,682 "readers_count": 4,683 "score": 6.0,684 "yours": false,685 "topic_id": 162575,686 "topic_slug": "gradients-or-parameters-exchange",687 "display_username": "yuri",688 "primary_group_name": null,689 "flair_name": null,690 "flair_url": null,691 "flair_bg_color": null,692 "flair_color": null,693 "flair_group_id": null,694 "badges_granted": [],695 "version": 1,696 "can_edit": false,697 "can_delete": false,698 "can_recover": false,699 "can_see_hidden_post": false,700 "can_wiki": false,701 "read": true,702 "user_title": null,703 "reply_to_user": {704 "id": 49515,705 "username": "agu",706 "name": "Andrew Gu",707 "avatar_template": "/user_avatar/discuss.pytorch.org/agu/{size}/49913_2.png"708 },709 "bookmarked": false,710 "actions_summary": [],711 "moderator": false,712 "admin": false,713 "staff": false,714 "user_id": 59131,715 "hidden": false,716 "trust_level": 1,717 "deleted_at": null,718 "user_deleted": false,719 "edit_reason": null,720 "can_view_edit_history": true,721 "wiki": false,722 "post_url": "/t/gradients-or-parameters-exchange/162575/3",723 "can_accept_answer": false,724 "can_unaccept_answer": false,725 "accepted_answer": false,726 "topic_accepted_answer": null727 },728 {729 "id": 369755,730 "name": "Andrew Gu",731 "username": "agu",732 "avatar_template": "/user_avatar/discuss.pytorch.org/agu/{size}/49913_2.png",733 "created_at": "2022-10-10T14:19:02.507Z",734 "cooked": "<p>Unfortunately, it is hard for me to say for sure without seeing the code or at least the full model definition. Would it be possible to print out the model and provide the parameter sizes?</p>\n<p>One possibility off the top of my head is that there are some parameters that individually are large, which may cause a bucket to exceed the 25 MB limit. For your model, do you think this could be happening?</p>",735 "post_number": 4,736 "post_type": 1,737 "posts_count": 4,738 "updated_at": "2022-10-10T14:19:02.507Z",739 "reply_count": 0,740 "reply_to_post_number": null,741 "quote_count": 0,742 "incoming_link_count": 0,743 "reads": 5,744 "readers_count": 4,745 "score": 1.0,746 "yours": false,747 "topic_id": 162575,748 "topic_slug": "gradients-or-parameters-exchange",749 "display_username": "Andrew Gu",750 "primary_group_name": null,751 "flair_name": null,752 "flair_url": null,753 "flair_bg_color": null,754 "flair_color": null,755 "flair_group_id": null,756 "badges_granted": [],757 "version": 1,758 "can_edit": false,759 "can_delete": false,760 "can_recover": false,761 "can_see_hidden_post": false,762 "can_wiki": false,763 "read": true,764 "user_title": null,765 "bookmarked": false,766 "actions_summary": [],767 "moderator": false,768 "admin": false,769 "staff": false,770 "user_id": 49515,771 "hidden": false,772 "trust_level": 2,773 "deleted_at": null,774 "user_deleted": false,775 "edit_reason": null,776 "can_view_edit_history": true,777 "wiki": false,778 "post_url": "/t/gradients-or-parameters-exchange/162575/4",779 "can_accept_answer": false,780 "can_unaccept_answer": false,781 "accepted_answer": false,782 "topic_accepted_answer": null783 }784 ],785 "stream": [786 368425,787 368903,788 369748,789 369755790 ]791 },792 "timeline_lookup": [793 [794 1,795 1121796 ],797 [798 2,799 1117800 ],801 [802 3,803 1111804 ]805 ],806 "suggested_topics": [807 {808 "fancy_title": "InfiniBand Vs TCP",809 "id": 215204,810 "title": "InfiniBand Vs TCP",811 "slug": "infiniband-vs-tcp",812 "posts_count": 2,813 "reply_count": 0,814 "highest_post_number": 2,815 "image_url": null,816 "created_at": "2025-01-10T10:27:01.563Z",817 "last_posted_at": "2025-01-29T12:49:37.101Z",818 "bumped": true,819 "bumped_at": "2025-01-29T12:49:37.101Z",820 "archetype": "regular",821 "unseen": false,822 "pinned": false,823 "unpinned": null,824 "visible": true,825 "closed": false,826 "archived": false,827 "bookmarked": null,828 "liked": null,829 "tags_descriptions": {},830 "like_count": 0,831 "views": 186,832 "category_id": 12,833 "featured_link": null,834 "has_accepted_answer": false,835 "posters": [836 {837 "extras": "latest single",838 "description": "Original Poster, Most Recent Poster",839 "user": {840 "id": 81002,841 "username": "FGeraci",842 "name": "Fabio Geraci",843 "avatar_template": "/letter_avatar_proxy/v4/letter/f/278dde/{size}.png",844 "trust_level": 1845 }846 }847 ]848 },849 {850 "fancy_title": "How do I run ddp on Windows?",851 "id": 212215,852 "title": "How do I run ddp on Windows?",853 "slug": "how-do-i-run-ddp-on-windows",854 "posts_count": 1,855 "reply_count": 0,856 "highest_post_number": 1,857 "image_url": null,858 "created_at": "2024-10-28T16:51:45.559Z",859 "last_posted_at": "2024-10-28T16:51:45.606Z",860 "bumped": true,861 "bumped_at": "2024-10-28T16:51:45.606Z",862 "archetype": "regular",863 "unseen": false,864 "pinned": false,865 "unpinned": null,866 "visible": true,867 "closed": false,868 "archived": false,869 "bookmarked": null,870 "liked": null,871 "tags_descriptions": {},872 "like_count": 0,873 "views": 27,874 "category_id": 12,875 "featured_link": null,876 "has_accepted_answer": false,877 "posters": [878 {879 "extras": "latest single",880 "description": "Original Poster, Most Recent Poster",881 "user": {882 "id": 80542,883 "username": "pardazesh_2050",884 "name": "pardazesh 2050",885 "avatar_template": "/user_avatar/discuss.pytorch.org/pardazesh_2050/{size}/73634_2.png",886 "trust_level": 0887 }888 }889 ]890 },891 {892 "fancy_title": "How to inference LLM with Multi-GPU",893 "id": 213022,894 "title": "How to inference LLM with Multi-GPU",895 "slug": "how-to-inference-llm-with-multi-gpu",896 "posts_count": 2,897 "reply_count": 0,898 "highest_post_number": 2,899 "image_url": null,900 "created_at": "2024-11-15T19:04:32.869Z",901 "last_posted_at": "2024-11-25T21:58:21.443Z",902 "bumped": true,903 "bumped_at": "2024-11-25T21:58:21.443Z",904 "archetype": "regular",905 "unseen": false,906 "pinned": false,907 "unpinned": null,908 "visible": true,909 "closed": false,910 "archived": false,911 "bookmarked": null,912 "liked": null,913 "tags_descriptions": {},914 "like_count": 0,915 "views": 571,916 "category_id": 12,917 "featured_link": null,918 "has_accepted_answer": false,919 "posters": [920 {921 "extras": null,922 "description": "Original Poster",923 "user": {924 "id": 80831,925 "username": "James_Jang",926 "name": "James Jang",927 "avatar_template": "/user_avatar/discuss.pytorch.org/james_jang/{size}/73925_2.png",928 "trust_level": 1929 }930 },931 {932 "extras": "latest",933 "description": "Most Recent Poster",934 "user": {935 "id": 78505,936 "username": "tianyu",937 "name": "",938 "avatar_template": "/user_avatar/discuss.pytorch.org/tianyu/{size}/72368_2.png",939 "trust_level": 2940 }941 }942 ]943 },944 {945 "fancy_title": "P2P blocking when pipeline schedule communication",946 "id": 212854,947 "title": "P2P blocking when pipeline schedule communication",948 "slug": "p2p-blocking-when-pipeline-schedule-communication",949 "posts_count": 1,950 "reply_count": 0,951 "highest_post_number": 1,952 "image_url": null,953 "created_at": "2024-11-12T09:52:56.645Z",954 "last_posted_at": "2024-11-12T09:52:56.707Z",955 "bumped": true,956 "bumped_at": "2024-11-12T09:52:56.707Z",957 "archetype": "regular",958 "unseen": false,959 "pinned": false,960 "unpinned": null,961 "visible": true,962 "closed": false,963 "archived": false,964 "bookmarked": null,965 "liked": null,966 "tags_descriptions": {},967 "like_count": 0,968 "views": 121,969 "category_id": 12,970 "featured_link": null,971 "has_accepted_answer": false,972 "posters": [973 {974 "extras": "latest single",975 "description": "Original Poster, Most Recent Poster",976 "user": {977 "id": 80849,978 "username": "FlyHorse24",979 "name": "Fly Horse24",980 "avatar_template": "/user_avatar/discuss.pytorch.org/flyhorse24/{size}/73948_2.png",981 "trust_level": 0982 }983 }984 ]985 },986 {987 "fancy_title": "Torch.distributed.checkpoint cannot load checkpoint files in multiple node environment",988 "id": 217025,989 "title": "Torch.distributed.checkpoint cannot load checkpoint files in multiple node environment",990 "slug": "torch-distributed-checkpoint-cannot-load-checkpoint-files-in-multiple-node-environment",991 "posts_count": 1,992 "reply_count": 0,993 "highest_post_number": 1,994 "image_url": null,995 "created_at": "2025-02-22T11:16:14.163Z",996 "last_posted_at": "2025-02-22T11:16:14.206Z",997 "bumped": true,998 "bumped_at": "2025-02-22T11:16:14.206Z",999 "archetype": "regular",1000 "unseen": false,1001 "pinned": false,1002 "unpinned": null,1003 "visible": true,1004 "closed": false,1005 "archived": false,1006 "bookmarked": null,1007 "liked": null,1008 "tags_descriptions": {},1009 "like_count": 0,1010 "views": 184,1011 "category_id": 12,1012 "featured_link": null,1013 "has_accepted_answer": false,1014 "posters": [1015 {1016 "extras": "latest single",1017 "description": "Original Poster, Most Recent Poster",1018 "user": {1019 "id": 70172,1020 "username": "gateofsoul",1021 "name": "",1022 "avatar_template": "/user_avatar/discuss.pytorch.org/gateofsoul/{size}/64657_2.png",1023 "trust_level": 11024 }1025 }1026 ]1027 }1028 ],1029 "tags_descriptions": {},1030 "fancy_title": "Gradients or parameters exchange",1031 "id": 162575,1032 "title": "Gradients or parameters exchange",1033 "posts_count": 4,1034 "created_at": "2022-10-01T06:19:42.273Z",1035 "views": 560,1036 "reply_count": 1,1037 "like_count": 0,1038 "last_posted_at": "2022-10-10T14:19:02.507Z",1039 "visible": true,1040 "closed": false,1041 "archived": false,1042 "has_summary": false,1043 "archetype": "regular",1044 "slug": "gradients-or-parameters-exchange",1045 "category_id": 12,1046 "word_count": 277,1047 "deleted_at": null,1048 "user_id": 59131,1049 "featured_link": null,1050 "pinned_globally": false,1051 "pinned_at": null,1052 "pinned_until": null,1053 "image_url": "https://discuss.pytorch.org/uploads/default/optimized/3X/f/4/f40fd0dfa84863068d621de217b8fb73bfeb89a5_2_1024x98.png",1054 "slow_mode_seconds": 0,1055 "draft": null,1056 "draft_key": "topic_162575",1057 "draft_sequence": null,1058 "unpinned": null,1059 "pinned": false,1060 "current_post_number": 1,1061 "highest_post_number": 4,1062 "deleted_by": null,1063 "actions_summary": [1064 {1065 "id": 4,1066 "count": 0,1067 "hidden": false,1068 "can_act": false1069 },1070 {1071 "id": 8,1072 "count": 0,1073 "hidden": false,1074 "can_act": false1075 },1076 {1077 "id": 10,1078 "count": 0,1079 "hidden": false,1080 "can_act": false1081 },1082 {1083 "id": 7,1084 "count": 0,1085 "hidden": false,1086 "can_act": false1087 }1088 ],1089 "chunk_size": 20,1090 "bookmarked": false,1091 "topic_timer": null,1092 "message_bus_last_id": 0,1093 "participant_count": 2,1094 "show_read_indicator": false,1095 "thumbnails": [1096 {1097 "max_width": null,1098 "max_height": null,1099 "width": 1462,1100 "height": 140,1101 "url": "https://discuss.pytorch.org/uploads/default/original/3X/f/4/f40fd0dfa84863068d621de217b8fb73bfeb89a5.png"1102 },1103 {1104 "max_width": 1024,1105 "max_height": 1024,1106 "width": 1024,1107 "height": 98,1108 "url": "https://discuss.pytorch.org/uploads/default/optimized/3X/f/4/f40fd0dfa84863068d621de217b8fb73bfeb89a5_2_1024x98.png"1109 }1110 ],1111 "slow_mode_enabled_until": null,1112 "can_vote": false,1113 "vote_count": 0,1114 "user_voted": false,1115 "discourse_zendesk_plugin_zendesk_id": null,1116 "discourse_zendesk_plugin_zendesk_url": "https://your-url.zendesk.com/agent/tickets/",1117 "details": {1118 "can_edit": false,1119 "notification_level": 1,1120 "participants": [1121 {1122 "id": 49515,1123 "username": "agu",1124 "name": "Andrew Gu",1125 "avatar_template": "/user_avatar/discuss.pytorch.org/agu/{size}/49913_2.png",1126 "post_count": 2,1127 "primary_group_name": null,1128 "flair_name": null,1129 "flair_url": null,1130 "flair_color": null,1131 "flair_bg_color": null,1132 "flair_group_id": null,1133 "trust_level": 21134 },1135 {1136 "id": 59131,1137 "username": "yuri123",1138 "name": "yuri",1139 "avatar_template": "/user_avatar/discuss.pytorch.org/yuri123/{size}/53013_2.png",1140 "post_count": 2,1141 "primary_group_name": null,1142 "flair_name": null,1143 "flair_url": null,1144 "flair_color": null,1145 "flair_bg_color": null,1146 "flair_group_id": null,1147 "trust_level": 11148 }1149 ],1150 "created_by": {1151 "id": 59131,1152 "username": "yuri123",1153 "name": "yuri",1154 "avatar_template": "/user_avatar/discuss.pytorch.org/yuri123/{size}/53013_2.png"1155 },1156 "last_poster": {1157 "id": 49515,1158 "username": "agu",1159 "name": "Andrew Gu",1160 "avatar_template": "/user_avatar/discuss.pytorch.org/agu/{size}/49913_2.png"1161 },1162 "links": [1163 {1164 "url": "https://discuss.pytorch.org/t/pytorch-ddp-bucket-size-does-not-match-model-size/162420/3",1165 "title": "Pytorch DDP bucket size does not match model size",1166 "internal": true,1167 "attachment": false,1168 "reflection": false,1169 "clicks": 3,1170 "user_id": 49515,1171 "domain": "discuss.pytorch.org",1172 "root_domain": "pytorch.org"1173 }1174 ]1175 },1176 "bookmarks": []1177 },1178 {1179 "post_stream": {1180 "posts": [1181 {1182 "id": 218778,1183 "name": "Dalisson Figueiredo",1184 "username": "Dalisson_Figueiredo",1185 "avatar_template": "/user_avatar/discuss.pytorch.org/dalisson_figueiredo/{size}/27561_2.png",1186 "created_at": "2020-08-06T22:59:17.752Z",1187 "cooked": "<p>Hi, I have an audio processing problem, first I need to chunk and put through and rnn, i start with a tensor of shape<br>\n(64, 16308) then I apply fold:</p>\n<p>tensor.fold(-1, 178, 89)</p>\n<p>so I get the tensor with shape (64, 182, 178) with nice overlapping sections</p>\n<p>now I can pass it through my rnns. Problem is, now I would like to apply the overlap and add method to get the original tensor back, I thought nn.Fold would be a good match to try and do that, but I dont understand its use well enough to know if I can even use it to do that.</p>",1188 "post_number": 1,1189 "post_type": 1,1190 "posts_count": 5,1191 "updated_at": "2020-08-06T22:59:17.752Z",1192 "reply_count": 0,1193 "reply_to_post_number": null,1194 "quote_count": 0,1195 "incoming_link_count": 1521,1196 "reads": 30,1197 "readers_count": 29,1198 "score": 7601.0,1199 "yours": false,1200 "topic_id": 91892,