Anurag1734/cuda-error-resolution-analysis
07
1[2 {3 "post_stream": {4 "posts": [5 {6 "id": 147194,7 "name": "Leo",8 "username": "leolcling",9 "avatar_template": "/user_avatar/discuss.pytorch.org/leolcling/{size}/17880_2.png",10 "created_at": "2019-11-19T08:46:28.865Z",11 "cooked": "<p>Hi,</p>\n<p>When processing long input sequences in seq2seq model, RNN/ transformer type models often perform worse. I want to implement a remedy by splitting the input sequences into smaller sequences (divide and conquer). And I want to find the splitting position by the loss of the seq2seq task.</p>\n<p>So I want:<br>\nInput -> splitting layer -> Output -> encoder -> decoder -> loss</p>\n<p>e.g.<br>\n[[ [1,1,1], [2,2,2], [3,3,3], [4,4,4] ]] (1 x 4 x 3) [(batch x len x feature)] -> splitting layer -> [ [[1,1,1] , [2,2,2], [3,3,3]], [[4,4,4]] ] (2x3x3 after padding)</p>\n<p>My idea is to use BiLSTM to find the splitting position, and optimize BiLSTM model by seq2seq model output, but I am having trouble understanding</p>\n<ol>\n<li>How the loss back-propagate to the splitting layer?</li>\n<li>How to reshape and pad after finding the split position?</li>\n</ol>\n<p>Thanks!</p>",12 "post_number": 1,13 "post_type": 1,14 "posts_count": 1,15 "updated_at": "2019-11-19T08:46:28.865Z",16 "reply_count": 0,17 "reply_to_post_number": null,18 "quote_count": 0,19 "incoming_link_count": 33,20 "reads": 7,21 "readers_count": 6,22 "score": 166.4,23 "yours": false,24 "topic_id": 61460,25 "topic_slug": "split-input-sequence-based-on-model-training-loss",26 "display_username": "Leo",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": "",42 "bookmarked": false,43 "actions_summary": [],44 "moderator": false,45 "admin": false,46 "staff": false,47 "user_id": 19071,48 "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/split-input-sequence-based-on-model-training-loss/61460/1",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 14719465 ]66 },67 "timeline_lookup": [68 [69 1,70 216871 ]72 ],73 "suggested_topics": [74 {75 "fancy_title": "APL support in pytorch",76 "id": 218060,77 "title": "APL support in pytorch",78 "slug": "apl-support-in-pytorch",79 "posts_count": 4,80 "reply_count": 1,81 "highest_post_number": 4,82 "image_url": null,83 "created_at": "2025-03-20T08:36:58.282Z",84 "last_posted_at": "2025-04-30T19:39:28.503Z",85 "bumped": true,86 "bumped_at": "2025-04-30T19:39:28.503Z",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": 107,99 "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": 83375,108 "username": "Meghana_R_Prakash",109 "name": "Meghana R Prakash",110 "avatar_template": "/user_avatar/discuss.pytorch.org/meghana_r_prakash/{size}/74876_2.png",111 "trust_level": 0112 }113 },114 {115 "extras": "latest",116 "description": "Most Recent Poster",117 "user": {118 "id": 77605,119 "username": "mhall119",120 "name": "Michael Hall",121 "avatar_template": "/user_avatar/discuss.pytorch.org/mhall119/{size}/71599_2.png",122 "trust_level": 1123 }124 }125 ]126 },127 {128 "fancy_title": "Lazy.h produces many compiler errors in UE project",129 "id": 212921,130 "title": "Lazy.h produces many compiler errors in UE project",131 "slug": "lazy-h-produces-many-compiler-errors-in-ue-project",132 "posts_count": 2,133 "reply_count": 0,134 "highest_post_number": 3,135 "image_url": null,136 "created_at": "2024-11-13T11:29:12.008Z",137 "last_posted_at": "2024-11-21T15:12:42.241Z",138 "bumped": true,139 "bumped_at": "2024-11-21T15:12:42.241Z",140 "archetype": "regular",141 "unseen": false,142 "pinned": false,143 "unpinned": null,144 "visible": true,145 "closed": false,146 "archived": false,147 "bookmarked": null,148 "liked": null,149 "tags_descriptions": {},150 "like_count": 1,151 "views": 137,152 "category_id": 1,153 "featured_link": null,154 "has_accepted_answer": false,155 "posters": [156 {157 "extras": null,158 "description": "Original Poster",159 "user": {160 "id": 74486,161 "username": "lavoos",162 "name": "lavoos",163 "avatar_template": "/letter_avatar_proxy/v4/letter/l/85f322/{size}.png",164 "trust_level": 1165 }166 },167 {168 "extras": "latest",169 "description": "Most Recent Poster",170 "user": {171 "id": 32206,172 "username": "sachithpd",173 "name": "Sachith Dickwella",174 "avatar_template": "/user_avatar/discuss.pytorch.org/sachithpd/{size}/25053_2.png",175 "trust_level": 1176 }177 }178 ]179 },180 {181 "fancy_title": "Why is there such a large difference using XLA in this snippet?",182 "id": 213684,183 "title": "Why is there such a large difference using XLA in this snippet?",184 "slug": "why-is-there-such-a-large-difference-using-xla-in-this-snippet",185 "posts_count": 5,186 "reply_count": 3,187 "highest_post_number": 5,188 "image_url": null,189 "created_at": "2024-12-02T07:42:16.130Z",190 "last_posted_at": "2024-12-03T12:32:12.523Z",191 "bumped": true,192 "bumped_at": "2024-12-03T12:32:12.523Z",193 "archetype": "regular",194 "unseen": false,195 "pinned": false,196 "unpinned": null,197 "visible": true,198 "closed": false,199 "archived": false,200 "bookmarked": null,201 "liked": null,202 "tags_descriptions": {},203 "like_count": 2,204 "views": 297,205 "category_id": 1,206 "featured_link": null,207 "has_accepted_answer": true,208 "posters": [209 {210 "extras": "latest",211 "description": "Original Poster, Most Recent Poster",212 "user": {213 "id": 81089,214 "username": "Aknw_Fen",215 "name": "Aknw Fen",216 "avatar_template": "/user_avatar/discuss.pytorch.org/aknw_fen/{size}/74156_2.png",217 "trust_level": 2218 }219 },220 {221 "extras": null,222 "description": "Frequent Poster, Accepted Answer",223 "user": {224 "id": 3534,225 "username": "ptrblck",226 "name": "",227 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",228 "admin": true,229 "moderator": true,230 "trust_level": 2231 }232 }233 ]234 },235 {236 "fancy_title": "Conda Pytorch keeps downgrading itself from cuda -> cpu",237 "id": 216525,238 "title": "Conda Pytorch keeps downgrading itself from cuda -> cpu",239 "slug": "conda-pytorch-keeps-downgrading-itself-from-cuda-cpu",240 "posts_count": 4,241 "reply_count": 1,242 "highest_post_number": 4,243 "image_url": null,244 "created_at": "2025-02-11T10:46:19.656Z",245 "last_posted_at": "2025-02-11T20:21:44.809Z",246 "bumped": true,247 "bumped_at": "2025-02-11T20:21:44.809Z",248 "archetype": "regular",249 "unseen": false,250 "pinned": false,251 "unpinned": null,252 "visible": true,253 "closed": false,254 "archived": false,255 "bookmarked": null,256 "liked": null,257 "tags_descriptions": {},258 "like_count": 0,259 "views": 251,260 "category_id": 1,261 "featured_link": null,262 "has_accepted_answer": false,263 "posters": [264 {265 "extras": null,266 "description": "Original Poster",267 "user": {268 "id": 71072,269 "username": "wilhelm",270 "name": "Dave",271 "avatar_template": "/user_avatar/discuss.pytorch.org/wilhelm/{size}/70280_2.png",272 "trust_level": 2273 }274 },275 {276 "extras": "latest",277 "description": "Most Recent Poster",278 "user": {279 "id": 3534,280 "username": "ptrblck",281 "name": "",282 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",283 "admin": true,284 "moderator": true,285 "trust_level": 2286 }287 }288 ]289 },290 {291 "fancy_title": "unstable WGAN-GP gradients",292 "id": 218537,293 "title": "unstable WGAN-GP gradients",294 "slug": "unstable-wgan-gp-gradients",295 "posts_count": 2,296 "reply_count": 0,297 "highest_post_number": 2,298 "image_url": null,299 "created_at": "2025-04-02T11:45:37.807Z",300 "last_posted_at": "2025-04-06T13:14:34.976Z",301 "bumped": true,302 "bumped_at": "2025-04-06T13:14:34.976Z",303 "archetype": "regular",304 "unseen": false,305 "pinned": false,306 "unpinned": null,307 "visible": true,308 "closed": false,309 "archived": false,310 "bookmarked": null,311 "liked": null,312 "tags_descriptions": {},313 "like_count": 1,314 "views": 79,315 "category_id": 1,316 "featured_link": null,317 "has_accepted_answer": false,318 "posters": [319 {320 "extras": null,321 "description": "Original Poster",322 "user": {323 "id": 83591,324 "username": "Szczerku",325 "name": "Szczerku",326 "avatar_template": "/user_avatar/discuss.pytorch.org/szczerku/{size}/76456_2.png",327 "trust_level": 0328 }329 },330 {331 "extras": "latest",332 "description": "Most Recent Poster",333 "user": {334 "id": 41458,335 "username": "J_Johnson",336 "name": "J Johnson",337 "avatar_template": "/user_avatar/discuss.pytorch.org/j_johnson/{size}/55494_2.png",338 "trust_level": 2339 }340 }341 ]342 }343 ],344 "tags_descriptions": {},345 "fancy_title": "Split input sequence based on model training loss",346 "id": 61460,347 "title": "Split input sequence based on model training loss",348 "posts_count": 1,349 "created_at": "2019-11-19T08:46:28.804Z",350 "views": 323,351 "reply_count": 0,352 "like_count": 0,353 "last_posted_at": "2019-11-19T08:46:28.865Z",354 "visible": true,355 "closed": false,356 "archived": false,357 "has_summary": false,358 "archetype": "regular",359 "slug": "split-input-sequence-based-on-model-training-loss",360 "category_id": 1,361 "word_count": 146,362 "deleted_at": null,363 "user_id": 19071,364 "featured_link": null,365 "pinned_globally": false,366 "pinned_at": null,367 "pinned_until": null,368 "image_url": null,369 "slow_mode_seconds": 0,370 "draft": null,371 "draft_key": "topic_61460",372 "draft_sequence": null,373 "unpinned": null,374 "pinned": false,375 "current_post_number": 1,376 "highest_post_number": 1,377 "deleted_by": null,378 "actions_summary": [379 {380 "id": 4,381 "count": 0,382 "hidden": false,383 "can_act": false384 },385 {386 "id": 8,387 "count": 0,388 "hidden": false,389 "can_act": false390 },391 {392 "id": 10,393 "count": 0,394 "hidden": false,395 "can_act": false396 },397 {398 "id": 7,399 "count": 0,400 "hidden": false,401 "can_act": false402 }403 ],404 "chunk_size": 20,405 "bookmarked": false,406 "topic_timer": null,407 "message_bus_last_id": 0,408 "participant_count": 1,409 "show_read_indicator": false,410 "thumbnails": null,411 "slow_mode_enabled_until": null,412 "can_vote": false,413 "vote_count": 0,414 "user_voted": false,415 "discourse_zendesk_plugin_zendesk_id": null,416 "discourse_zendesk_plugin_zendesk_url": "https://your-url.zendesk.com/agent/tickets/",417 "details": {418 "can_edit": false,419 "notification_level": 1,420 "participants": [421 {422 "id": 19071,423 "username": "leolcling",424 "name": "Leo",425 "avatar_template": "/user_avatar/discuss.pytorch.org/leolcling/{size}/17880_2.png",426 "post_count": 1,427 "primary_group_name": null,428 "flair_name": null,429 "flair_url": null,430 "flair_color": null,431 "flair_bg_color": null,432 "flair_group_id": null,433 "trust_level": 1434 }435 ],436 "created_by": {437 "id": 19071,438 "username": "leolcling",439 "name": "Leo",440 "avatar_template": "/user_avatar/discuss.pytorch.org/leolcling/{size}/17880_2.png"441 },442 "last_poster": {443 "id": 19071,444 "username": "leolcling",445 "name": "Leo",446 "avatar_template": "/user_avatar/discuss.pytorch.org/leolcling/{size}/17880_2.png"447 }448 },449 "bookmarks": []450 },451 {452 "post_stream": {453 "posts": [454 {455 "id": 147101,456 "name": "TG",457 "username": "t.g",458 "avatar_template": "/user_avatar/discuss.pytorch.org/t.g/{size}/74955_2.png",459 "created_at": "2019-11-18T23:30:45.296Z",460 "cooked": "<p>Hi,</p>\n<p>There are already a ton of posts on this topic of \" torch is not recognizing GPU backend\".<br>\nEach of the solutions mentioned is per case (such as upgrade or downgrade cuda or driver to some magic number for a specific torch version; see <a href=\"https://devtalk.nvidia.com/default/topic/1047223/torch-cuda-is_available-returns-false/\" rel=\"nofollow noopener\">1</a>, <a href=\"https://stackoverflow.com/questions/55717751/torch-cuda-is-available-keeps-switching-to-false\" rel=\"nofollow noopener\">2</a>, <a href=\"https://discuss.pytorch.org/t/why-torch-cuda-is-available-still-returned-false-even-i-installed-cuda-9-0-176-help/33364\">3</a>, <a href=\"https://discuss.pytorch.org/t/torch-cuda-is-available-returns-false-nvidia-smi-is-working/20614\">4</a>, <a href=\"https://github.com/pytorch/pytorch/issues/15612\" rel=\"nofollow noopener\">5</a>, <a href=\"https://discuss.pytorch.org/t/pytorch-0-4-get-cuda-version/19058/2\">6</a> …), and none of those answers seem generic enough to address most people, so they keep coming with new versions of torch.</p>\n<p>So, here is what I learned</p>\n<p>Currently</p>\n<pre><code class=\"lang-auto\">$ python -c 'import torch; print(torch.cuda.is_available())'\nFalse\n</code></pre>\n<p>But why? I got <code>nvidia-smi</code> showing GPU correctly.</p>\n<p>One sure-shot way of fixing my cuda lib compatibility problems (not the desirable, though!) is asking tensorflow</p>\n<pre><code class=\"lang-auto\">$ python -c 'import tensorflow as tf; print(tf.test.is_gpu_available())'\n\n2019-11-18 22:56:28.982050: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:\nname: TITAN X (Pascal) major: 6 minor: 1 memoryClockRate(GHz): 1.531\npciBusID: 0000:00:06.0\n2019-11-18 22:56:28.986522: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcudart.so.10.0'; dlerror: libcudart.so.10.0: cannot open shared object file: No such file or directory\n2019-11-18 22:56:28.989327: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcublas.so.10.0'; dlerror: libcublas.so.10.0: cannot open shared object file: No such file or directory\n2019-11-18 22:56:28.992141: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcufft.so.10.0'; dlerror: libcufft.so.10.0: cannot open shared object file: No such file or directory\n2019-11-18 22:56:28.994862: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcurand.so.10.0'; dlerror: libcurand.so.10.0: cannot open shared object file: No such file or directory\n2019-11-18 22:56:28.997474: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcusolver.so.10.0'; dlerror: libcusolver.so.10.0: cannot open shared object file: No such file or directory\n2019-11-18 22:56:29.000329: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcusparse.so.10.0'; dlerror: libcusparse.so.10.0: cannot open shared object file: No such file or directory\n2019-11-18 22:56:29.003020: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcudnn.so.7'; dlerror: libcudnn.so.7: cannot open shared object file: No such file or directory\n2019-11-18 22:56:29.003104: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1641] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.\nSkipping registering GPU devices...\n2019-11-18 22:56:29.003232: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1159] Device interconnect StreamExecutor with strength 1 edge matrix:\n2019-11-18 22:56:29.003284: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1165] 0\n2019-11-18 22:56:29.003372: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1178] 0: N\nFalse\n</code></pre>\n<p>TF precisely states <code>Could not load dynamic library <namehere></code>, example <code>libcudnn.so.7</code> or <code>'libcublas.so.10.0'</code> which is a crucial info to know what cuda libs and versions are missing.</p>\n<p>If there is a way to get such debug info from torch? please let me know. (we would be pleased to not depend on TF to fix it). <strong>Should the <code>torch.cuda.is_available()</code> have <code>debug=True</code> argument to print which of the missing libraries are causing it to return <code>False</code>?</strong></p>\n<hr>\n<p>With this, if we figure out the missing libs and version are:<br>\n<code>Could not load dynamic library 'libcudnn.so.7';</code> or <code>'libcublas.so.10.0'</code><br>\nWe can do</p>\n<pre><code class=\"lang-auto\">conda install cudnn=7 cudatoolkit=10.0 -c anaconda\n</code></pre>\n<p>then for sure torch recognizes GPU backend:</p>\n<pre><code class=\"lang-auto\">$ python -c 'import torch; print(torch.cuda.is_available())'\nTrue\n</code></pre>\n<p>But we have to know first that the missing libs are <code>'libcudnn.so.7'</code> and <code>'libcublas.so.10.0'</code> to install <code>cudnn=7 cudatoolkit=10.0</code> (otherwise it goes to trial-error and magic numbers)</p>",461 "post_number": 1,462 "post_type": 1,463 "posts_count": 4,464 "updated_at": "2019-11-18T23:35:08.633Z",465 "reply_count": 0,466 "reply_to_post_number": null,467 "quote_count": 0,468 "incoming_link_count": 2211,469 "reads": 33,470 "readers_count": 32,471 "score": 11046.6,472 "yours": false,473 "topic_id": 61402,474 "topic_slug": "how-to-get-debug-info-from-torch-cuda-is-available-about-missing-libs",475 "display_username": "TG",476 "primary_group_name": null,477 "flair_name": null,478 "flair_url": null,479 "flair_bg_color": null,480 "flair_color": null,481 "flair_group_id": null,482 "badges_granted": [],483 "version": 1,484 "can_edit": false,485 "can_delete": false,486 "can_recover": false,487 "can_see_hidden_post": false,488 "can_wiki": false,489 "link_counts": [490 {491 "url": "https://devtalk.nvidia.com/default/topic/1047223/torch-cuda-is_available-returns-false/",492 "internal": false,493 "reflection": false,494 "clicks": 2495 },496 {497 "url": "https://discuss.pytorch.org/t/pytorch-0-4-get-cuda-version/19058/2",498 "internal": true,499 "reflection": false,500 "title": "Pytorch 0.4 get cuda version",501 "clicks": 1502 },503 {504 "url": "https://discuss.pytorch.org/t/why-torch-cuda-is-available-still-returned-false-even-i-installed-cuda-9-0-176-help/33364",505 "internal": true,506 "reflection": false,507 "title": "Why 'torch.cuda.is_available()' still returned 'False' even I installed CUDA 9.0.176.....Help",508 "clicks": 1509 },510 {511 "url": "https://stackoverflow.com/questions/55717751/torch-cuda-is-available-keeps-switching-to-false",512 "internal": false,513 "reflection": false,514 "title": "pytorch - Torch.cuda.is_available() keeps switching to False - Stack Overflow",515 "clicks": 0516 },517 {518 "url": "https://discuss.pytorch.org/t/torch-cuda-is-available-returns-false-nvidia-smi-is-working/20614",519 "internal": true,520 "reflection": false,521 "title": "Torch.cuda.is_available() returns False, nvidia-smi is working",522 "clicks": 0523 },524 {525 "url": "https://github.com/pytorch/pytorch/issues/15612",526 "internal": false,527 "reflection": false,528 "title": "torch.cuda.is_available() is false after CUDA 9.0.176 installed. Could anyone help me with this? Thanks! · Issue #15612 · pytorch/pytorch · GitHub",529 "clicks": 0530 }531 ],532 "read": true,533 "user_title": "",534 "bookmarked": false,535 "actions_summary": [],536 "moderator": false,537 "admin": false,538 "staff": false,539 "user_id": 9465,540 "hidden": false,541 "trust_level": 1,542 "deleted_at": null,543 "user_deleted": false,544 "edit_reason": null,545 "can_view_edit_history": true,546 "wiki": false,547 "post_url": "/t/how-to-get-debug-info-from-torch-cuda-is-available-about-missing-libs/61402/1",548 "can_accept_answer": false,549 "can_unaccept_answer": false,550 "accepted_answer": false,551 "topic_accepted_answer": true,552 "can_vote": false553 },554 {555 "id": 147102,556 "name": "TG",557 "username": "t.g",558 "avatar_template": "/user_avatar/discuss.pytorch.org/t.g/{size}/74955_2.png",559 "created_at": "2019-11-18T23:49:05.808Z",560 "cooked": "<p>Also, another topic related to getting extra info from <code>debug=True</code>:</p>\n<pre><code class=\"lang-auto\">$ python -c 'import torch; print(torch.cuda.is_available(), torch.version.cuda)'\nFalse 10.0.130\n</code></pre>\n<pre><code class=\"lang-auto\">$ python -c 'import tensorflow as tf; print(tf.test.is_gpu_available())'\n...\nE tensorflow/stream_executor/cuda/cuda_driver.cc:318] failed call to cuInit: CUDA_ERROR_UNKNOWN: unknown error\n...\nFalse\n</code></pre>\n<p>I guess <code>cuInit: CUDA_ERROR_UNKNOWN: unknown error</code> means I have to restart the machine. Knowing this errors would help as well.</p>",561 "post_number": 2,562 "post_type": 1,563 "posts_count": 4,564 "updated_at": "2019-11-18T23:49:05.808Z",565 "reply_count": 1,566 "reply_to_post_number": null,567 "quote_count": 0,568 "incoming_link_count": 31,569 "reads": 27,570 "readers_count": 26,571 "score": 165.4,572 "yours": false,573 "topic_id": 61402,574 "topic_slug": "how-to-get-debug-info-from-torch-cuda-is-available-about-missing-libs",575 "display_username": "TG",576 "primary_group_name": null,577 "flair_name": null,578 "flair_url": null,579 "flair_bg_color": null,580 "flair_color": null,581 "flair_group_id": null,582 "badges_granted": [],583 "version": 1,584 "can_edit": false,585 "can_delete": false,586 "can_recover": false,587 "can_see_hidden_post": false,588 "can_wiki": false,589 "read": true,590 "user_title": "",591 "bookmarked": false,592 "actions_summary": [],593 "moderator": false,594 "admin": false,595 "staff": false,596 "user_id": 9465,597 "hidden": false,598 "trust_level": 1,599 "deleted_at": null,600 "user_deleted": false,601 "edit_reason": null,602 "can_view_edit_history": true,603 "wiki": false,604 "post_url": "/t/how-to-get-debug-info-from-torch-cuda-is-available-about-missing-libs/61402/2",605 "can_accept_answer": false,606 "can_unaccept_answer": false,607 "accepted_answer": false,608 "topic_accepted_answer": true609 },610 {611 "id": 147105,612 "name": "Alban D",613 "username": "albanD",614 "avatar_template": "/user_avatar/discuss.pytorch.org/alband/{size}/215_2.png",615 "created_at": "2019-11-19T00:18:17.934Z",616 "cooked": "<p>Hi,</p>\n<p>You can try to actually use a cuda element to see such errors: <code>torch.rand(1, device=\"cuda\")</code>.<br>\nDoes that give you the informations you want?</p>",617 "post_number": 3,618 "post_type": 1,619 "posts_count": 4,620 "updated_at": "2019-11-19T07:49:24.986Z",621 "reply_count": 1,622 "reply_to_post_number": 2,623 "quote_count": 0,624 "incoming_link_count": 3,625 "reads": 26,626 "readers_count": 25,627 "score": 40.2,628 "yours": false,629 "topic_id": 61402,630 "topic_slug": "how-to-get-debug-info-from-torch-cuda-is-available-about-missing-libs",631 "display_username": "Alban D",632 "primary_group_name": null,633 "flair_name": null,634 "flair_url": null,635 "flair_bg_color": null,636 "flair_color": null,637 "flair_group_id": null,638 "badges_granted": [],639 "version": 1,640 "can_edit": false,641 "can_delete": false,642 "can_recover": false,643 "can_see_hidden_post": false,644 "can_wiki": false,645 "read": true,646 "user_title": "",647 "reply_to_user": {648 "id": 9465,649 "username": "t.g",650 "name": "TG",651 "avatar_template": "/user_avatar/discuss.pytorch.org/t.g/{size}/74955_2.png"652 },653 "bookmarked": false,654 "actions_summary": [655 {656 "id": 2,657 "count": 1658 }659 ],660 "moderator": true,661 "admin": true,662 "staff": true,663 "user_id": 211,664 "hidden": false,665 "trust_level": 4,666 "deleted_at": null,667 "user_deleted": false,668 "edit_reason": null,669 "can_view_edit_history": true,670 "wiki": false,671 "post_url": "/t/how-to-get-debug-info-from-torch-cuda-is-available-about-missing-libs/61402/3",672 "can_accept_answer": false,673 "can_unaccept_answer": false,674 "accepted_answer": true,675 "topic_accepted_answer": true676 },677 {678 "id": 147189,679 "name": "TG",680 "username": "t.g",681 "avatar_template": "/user_avatar/discuss.pytorch.org/t.g/{size}/74955_2.png",682 "created_at": "2019-11-19T07:53:44.968Z",683 "cooked": "<p>Yes that helps. It prints useful info. Thanks.<br>\nWish it was a documented feature!</p>\n<p>My code branching has been</p>\n<pre><code class=\"lang-auto\">if torch.cuda.is_available():\n use cuda\nelse:\n use cpu, dont attempt to use of cuda\n</code></pre>\n<p>so I never had chance to see that message</p>",684 "post_number": 4,685 "post_type": 1,686 "posts_count": 4,687 "updated_at": "2019-11-19T07:53:44.968Z",688 "reply_count": 0,689 "reply_to_post_number": 3,690 "quote_count": 0,691 "incoming_link_count": 12,692 "reads": 23,693 "readers_count": 22,694 "score": 64.6,695 "yours": false,696 "topic_id": 61402,697 "topic_slug": "how-to-get-debug-info-from-torch-cuda-is-available-about-missing-libs",698 "display_username": "TG",699 "primary_group_name": null,700 "flair_name": null,701 "flair_url": null,702 "flair_bg_color": null,703 "flair_color": null,704 "flair_group_id": null,705 "badges_granted": [],706 "version": 1,707 "can_edit": false,708 "can_delete": false,709 "can_recover": false,710 "can_see_hidden_post": false,711 "can_wiki": false,712 "read": true,713 "user_title": "",714 "reply_to_user": {715 "id": 211,716 "username": "albanD",717 "name": "Alban D",718 "avatar_template": "/user_avatar/discuss.pytorch.org/alband/{size}/215_2.png"719 },720 "bookmarked": false,721 "actions_summary": [],722 "moderator": false,723 "admin": false,724 "staff": false,725 "user_id": 9465,726 "hidden": false,727 "trust_level": 1,728 "deleted_at": null,729 "user_deleted": false,730 "edit_reason": null,731 "can_view_edit_history": true,732 "wiki": false,733 "post_url": "/t/how-to-get-debug-info-from-torch-cuda-is-available-about-missing-libs/61402/4",734 "can_accept_answer": false,735 "can_unaccept_answer": false,736 "accepted_answer": false,737 "topic_accepted_answer": true738 }739 ],740 "stream": [741 147101,742 147102,743 147105,744 147189745 ]746 },747 "timeline_lookup": [748 [749 1,750 2168751 ]752 ],753 "suggested_topics": [754 {755 "fancy_title": "Trying to backward through the graph a second time error",756 "id": 213322,757 "title": "Trying to backward through the graph a second time error",758 "slug": "trying-to-backward-through-the-graph-a-second-time-error",759 "posts_count": 1,760 "reply_count": 0,761 "highest_post_number": 1,762 "image_url": null,763 "created_at": "2024-11-22T19:50:38.013Z",764 "last_posted_at": "2024-11-22T19:50:38.070Z",765 "bumped": true,766 "bumped_at": "2024-11-22T19:50:38.070Z",767 "archetype": "regular",768 "unseen": false,769 "pinned": false,770 "unpinned": null,771 "visible": true,772 "closed": false,773 "archived": false,774 "bookmarked": null,775 "liked": null,776 "tags_descriptions": {},777 "like_count": 0,778 "views": 33,779 "category_id": 1,780 "featured_link": null,781 "has_accepted_answer": false,782 "posters": [783 {784 "extras": "latest single",785 "description": "Original Poster, Most Recent Poster",786 "user": {787 "id": 64607,788 "username": "Selvam_Dharma",789 "name": "Selvam Dharma",790 "avatar_template": "/user_avatar/discuss.pytorch.org/selvam_dharma/{size}/58785_2.png",791 "trust_level": 1792 }793 }794 ]795 },796 {797 "fancy_title": "Cross-entropy loss ignores labels of a specific index but throws an error",798 "id": 214178,799 "title": "Cross-entropy loss ignores labels of a specific index but throws an error",800 "slug": "cross-entropy-loss-ignores-labels-of-a-specific-index-but-throws-an-error",801 "posts_count": 2,802 "reply_count": 0,803 "highest_post_number": 2,804 "image_url": null,805 "created_at": "2024-12-13T02:33:20.769Z",806 "last_posted_at": "2024-12-13T21:10:31.897Z",807 "bumped": true,808 "bumped_at": "2024-12-13T21:10:31.897Z",809 "archetype": "regular",810 "unseen": false,811 "pinned": false,812 "unpinned": null,813 "visible": true,814 "closed": false,815 "archived": false,816 "bookmarked": null,817 "liked": null,818 "tags_descriptions": {},819 "like_count": 0,820 "views": 42,821 "category_id": 1,822 "featured_link": null,823 "has_accepted_answer": false,824 "posters": [825 {826 "extras": null,827 "description": "Original Poster",828 "user": {829 "id": 70205,830 "username": "aohenuo",831 "name": "aohenuo's github account",832 "avatar_template": "/user_avatar/discuss.pytorch.org/aohenuo/{size}/64688_2.png",833 "trust_level": 0834 }835 },836 {837 "extras": "latest",838 "description": "Most Recent Poster",839 "user": {840 "id": 3534,841 "username": "ptrblck",842 "name": "",843 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",844 "admin": true,845 "moderator": true,846 "trust_level": 2847 }848 }849 ]850 },851 {852 "fancy_title": "Explicit casting inside autocast context",853 "id": 216145,854 "title": "Explicit casting inside autocast context",855 "slug": "explicit-casting-inside-autocast-context",856 "posts_count": 3,857 "reply_count": 1,858 "highest_post_number": 3,859 "image_url": null,860 "created_at": "2025-02-02T08:45:15.104Z",861 "last_posted_at": "2025-02-03T02:31:53.809Z",862 "bumped": true,863 "bumped_at": "2025-02-03T02:31:53.809Z",864 "archetype": "regular",865 "unseen": false,866 "pinned": false,867 "unpinned": null,868 "visible": true,869 "closed": false,870 "archived": false,871 "bookmarked": null,872 "liked": null,873 "tags_descriptions": {},874 "like_count": 0,875 "views": 155,876 "category_id": 1,877 "featured_link": null,878 "has_accepted_answer": false,879 "posters": [880 {881 "extras": "latest",882 "description": "Original Poster, Most Recent Poster",883 "user": {884 "id": 82442,885 "username": "avi_singhal",886 "name": "avi singhal",887 "avatar_template": "/user_avatar/discuss.pytorch.org/avi_singhal/{size}/75428_2.png",888 "trust_level": 1889 }890 },891 {892 "extras": null,893 "description": "Frequent Poster",894 "user": {895 "id": 3534,896 "username": "ptrblck",897 "name": "",898 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",899 "admin": true,900 "moderator": true,901 "trust_level": 2902 }903 }904 ]905 },906 {907 "fancy_title": "1D convolutions in parallel",908 "id": 216495,909 "title": "1D convolutions in parallel",910 "slug": "1d-convolutions-in-parallel",911 "posts_count": 2,912 "reply_count": 0,913 "highest_post_number": 2,914 "image_url": null,915 "created_at": "2025-02-10T22:31:39.944Z",916 "last_posted_at": "2025-02-10T22:33:50.762Z",917 "bumped": true,918 "bumped_at": "2025-02-10T22:33:50.762Z",919 "archetype": "regular",920 "unseen": false,921 "pinned": false,922 "unpinned": null,923 "visible": true,924 "closed": false,925 "archived": false,926 "bookmarked": null,927 "liked": null,928 "tags_descriptions": {},929 "like_count": 0,930 "views": 101,931 "category_id": 1,932 "featured_link": null,933 "has_accepted_answer": false,934 "posters": [935 {936 "extras": null,937 "description": "Original Poster",938 "user": {939 "id": 24250,940 "username": "Andrew_Ulmer",941 "name": "Andrew Ulmer",942 "avatar_template": "/letter_avatar_proxy/v4/letter/a/977dab/{size}.png",943 "trust_level": 1944 }945 },946 {947 "extras": "latest",948 "description": "Most Recent Poster",949 "user": {950 "id": 3534,951 "username": "ptrblck",952 "name": "",953 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",954 "admin": true,955 "moderator": true,956 "trust_level": 2957 }958 }959 ]960 },961 {962 "fancy_title": "GradScaler: TypeError: Cannot convert a MPS Tensor to float64 dtype as the MPS framework doesn’t support float64. Please use float32 instead",963 "id": 213206,964 "title": "GradScaler: TypeError: Cannot convert a MPS Tensor to float64 dtype as the MPS framework doesn't support float64. Please use float32 instead",965 "slug": "gradscaler-typeerror-cannot-convert-a-mps-tensor-to-float64-dtype-as-the-mps-framework-doesnt-support-float64-please-use-float32-instead",966 "posts_count": 2,967 "reply_count": 0,968 "highest_post_number": 2,969 "image_url": null,970 "created_at": "2024-11-20T12:09:55.629Z",971 "last_posted_at": "2024-12-23T15:29:26.138Z",972 "bumped": true,973 "bumped_at": "2024-12-23T15:29:26.138Z",974 "archetype": "regular",975 "unseen": false,976 "pinned": false,977 "unpinned": null,978 "visible": true,979 "closed": false,980 "archived": false,981 "bookmarked": null,982 "liked": null,983 "tags_descriptions": {},984 "like_count": 0,985 "views": 658,986 "category_id": 1,987 "featured_link": null,988 "has_accepted_answer": false,989 "posters": [990 {991 "extras": null,992 "description": "Original Poster",993 "user": {994 "id": 80423,995 "username": "Mauro_Sciancalepore",996 "name": "Mauro Sciancalepore",997 "avatar_template": "/user_avatar/discuss.pytorch.org/mauro_sciancalepore/{size}/73510_2.png",998 "trust_level": 1999 }1000 },1001 {1002 "extras": "latest",1003 "description": "Most Recent Poster",1004 "user": {1005 "id": 81681,1006 "username": "starksm64",1007 "name": "Scott M Stark",1008 "avatar_template": "/user_avatar/discuss.pytorch.org/starksm64/{size}/74694_2.png",1009 "trust_level": 01010 }1011 }1012 ]1013 }1014 ],1015 "tags_descriptions": {},1016 "fancy_title": "How to get debug info from `torch.cuda.is_available()` about missing libs",1017 "id": 61402,1018 "title": "How to get debug info from `torch.cuda.is_available()` about missing libs",1019 "posts_count": 4,1020 "created_at": "2019-11-18T23:30:45.203Z",1021 "views": 3962,1022 "reply_count": 2,1023 "like_count": 1,1024 "last_posted_at": "2019-11-19T07:53:44.968Z",1025 "visible": true,1026 "closed": false,1027 "archived": false,1028 "has_summary": false,1029 "archetype": "regular",1030 "slug": "how-to-get-debug-info-from-torch-cuda-is-available-about-missing-libs",1031 "category_id": 1,1032 "word_count": 947,1033 "deleted_at": null,1034 "user_id": 9465,1035 "featured_link": null,1036 "pinned_globally": false,1037 "pinned_at": null,1038 "pinned_until": null,1039 "image_url": null,1040 "slow_mode_seconds": 0,1041 "draft": null,1042 "draft_key": "topic_61402",1043 "draft_sequence": null,1044 "unpinned": null,1045 "pinned": false,1046 "current_post_number": 1,1047 "highest_post_number": 4,1048 "deleted_by": null,1049 "actions_summary": [1050 {1051 "id": 4,1052 "count": 0,1053 "hidden": false,1054 "can_act": false1055 },1056 {1057 "id": 8,1058 "count": 0,1059 "hidden": false,1060 "can_act": false1061 },1062 {1063 "id": 10,1064 "count": 0,1065 "hidden": false,1066 "can_act": false1067 },1068 {1069 "id": 7,1070 "count": 0,1071 "hidden": false,1072 "can_act": false1073 }1074 ],1075 "chunk_size": 20,1076 "bookmarked": false,1077 "topic_timer": null,1078 "message_bus_last_id": 0,1079 "participant_count": 2,1080 "show_read_indicator": false,1081 "thumbnails": null,1082 "slow_mode_enabled_until": null,1083 "accepted_answer": {1084 "post_number": 3,1085 "username": "albanD",1086 "name": "Alban D",1087 "excerpt": "Hi, \nYou can try to actually use a cuda element to see such errors: torch.rand(1, device="cuda"). \nDoes that give you the informations you want?"1088 },1089 "can_vote": false,1090 "vote_count": 0,1091 "user_voted": false,1092 "discourse_zendesk_plugin_zendesk_id": null,1093 "discourse_zendesk_plugin_zendesk_url": "https://your-url.zendesk.com/agent/tickets/",1094 "details": {1095 "can_edit": false,1096 "notification_level": 1,1097 "participants": [1098 {1099 "id": 9465,1100 "username": "t.g",1101 "name": "TG",1102 "avatar_template": "/user_avatar/discuss.pytorch.org/t.g/{size}/74955_2.png",1103 "post_count": 3,1104 "primary_group_name": null,1105 "flair_name": null,1106 "flair_url": null,1107 "flair_color": null,1108 "flair_bg_color": null,1109 "flair_group_id": null,1110 "trust_level": 11111 },1112 {1113 "id": 211,1114 "username": "albanD",1115 "name": "Alban D",1116 "avatar_template": "/user_avatar/discuss.pytorch.org/alband/{size}/215_2.png",1117 "post_count": 1,1118 "primary_group_name": null,1119 "flair_name": null,1120 "flair_url": null,1121 "flair_color": null,1122 "flair_bg_color": null,1123 "flair_group_id": null,1124 "admin": true,1125 "moderator": true,1126 "trust_level": 41127 }1128 ],1129 "created_by": {1130 "id": 9465,1131 "username": "t.g",1132 "name": "TG",1133 "avatar_template": "/user_avatar/discuss.pytorch.org/t.g/{size}/74955_2.png"1134 },1135 "last_poster": {1136 "id": 9465,1137 "username": "t.g",1138 "name": "TG",1139 "avatar_template": "/user_avatar/discuss.pytorch.org/t.g/{size}/74955_2.png"1140 },1141 "links": [1142 {1143 "url": "https://devtalk.nvidia.com/default/topic/1047223/torch-cuda-is_available-returns-false/",1144 "title": null,1145 "internal": false,1146 "attachment": false,1147 "reflection": false,1148 "clicks": 2,1149 "user_id": 9465,1150 "domain": "devtalk.nvidia.com",1151 "root_domain": "nvidia.com"1152 },1153 {1154 "url": "https://discuss.pytorch.org/t/pytorch-0-4-get-cuda-version/19058/2",1155 "title": "Pytorch 0.4 get cuda version",1156 "internal": true,1157 "attachment": false,1158 "reflection": false,1159 "clicks": 1,1160 "user_id": 9465,1161 "domain": "discuss.pytorch.org",1162 "root_domain": "pytorch.org"1163 },1164 {1165 "url": "https://discuss.pytorch.org/t/why-torch-cuda-is-available-still-returned-false-even-i-installed-cuda-9-0-176-help/33364",1166 "title": "Why 'torch.cuda.is_available()' still returned 'False' even I installed CUDA 9.0.176.....Help",1167 "internal": true,1168 "attachment": false,1169 "reflection": false,1170 "clicks": 1,1171 "user_id": 9465,1172 "domain": "discuss.pytorch.org",1173 "root_domain": "pytorch.org"1174 }1175 ]1176 },1177 "bookmarks": []1178 },1179 {1180 "post_stream": {1181 "posts": [1182 {1183 "id": 101578,1184 "name": "lavenderxx",1185 "username": "lavender99",1186 "avatar_template": "/user_avatar/discuss.pytorch.org/lavender99/{size}/10608_2.png",1187 "created_at": "2019-03-25T13:54:25.127Z",1188 "cooked": "<p>Hello<br>\ni am trying to plot Nifty format in pytorch after passing them to the loader but i am getting wrong image , code and image shown below</p>\n<pre><code class=\"lang-auto\">class Dataloder_img(data.Dataset):\n def __init__(self,root_dir,seg_dir,transforms ):\n self.root_dir = root_dir\n self.seg_dir = seg_dir\n self.transforms = transforms\n self.files = os.listdir(self.root_dir)\n self.lables = os.listdir(self.seg_dir)\n print(self.files)\n \n def __len__(self):\n return len(self.files)\n \n def __getitem__(self,idx):\n img_name = self.files[idx]\n label_name = self.lables[idx]\n img = nib.load(os.path.join(self.root_dir,img_name)) #!Image.open(os.path.join(self.root_dir,img_name))\n #change to numpy\n img = np.array(img.dataobj)\n #change to PIL \n img = Image.fromarray(img.astype('uint8'), 'RGB')\n \n print(img.size)\n \n label = nib.load(os.path.join(self.seg_dir,label_name))#!Image.open(os.path.join(self.seg_dir,label_name))\n #change to numpy\n label = np.array(label.dataobj)\n #change to PIL \n label = Image.fromarray(label.astype('uint8'), 'RGB')\n \n print(label.size)\n \n if self.transforms:\n img = self.transforms(img)\n label = self.transforms(label)\n return img,label\n else:\n return img, label\nfull_dataset = Dataloder_img('/``imageTr',\n '/``/labelTr',tfms.Compose([tfms.RandomRotation(0),tfms.Resize((256,256)),tfms.ToTensor()\n ]))#\n \n\ntrain_size = int(0.8 * len(full_dataset))\nval_size = len(full_dataset) - train_size\ntrain_dataset, val_dataset = torch.utils.data.random_split(full_dataset, [train_size, val_size])\ntrain_loader = data.DataLoader(train_dataset,shuffle=False,batch_size=bs)\nval_loader = data.DataLoader(val_dataset,shuffle=False,batch_size=bs)\nimport matplotlib as mpl\nmpl.rcParams['axes.grid'] = False\ntest_img, test_lb = next(iter(full_dataset))\nprint(test_img[0].shape)\nplt.imshow(test_lb[0])\n</code></pre>\n<p><img src=\"https://discuss.pytorch.org/uploads/default/original/2X/e/e527cdbb77b1a3ff47afb1138d94f83836a87a21.png\" alt=\"download\" data-base62-sha1=\"wHcCpU2Z0aEQbW5l05AtQYRoCgp\" width=\"339\" height=\"333\"></p>",1189 "post_number": 1,1190 "post_type": 1,1191 "posts_count": 2,1192 "updated_at": "2019-03-25T13:54:25.127Z",1193 "reply_count": 0,1194 "reply_to_post_number": null,1195 "quote_count": 0,1196 "incoming_link_count": 280,1197 "reads": 17,1198 "readers_count": 16,1199 "score": 1393.4,1200 "yours": false,