Anurag1734/cuda-error-resolution-analysis
07
1[2 {3 "post_stream": {4 "posts": [5 {6 "id": 327150,7 "name": "Sean Farrell",8 "username": "seanfarr788",9 "avatar_template": "/user_avatar/discuss.pytorch.org/seanfarr788/{size}/44321_2.png",10 "created_at": "2022-01-18T17:10:20.989Z",11 "cooked": "<p>I am trying to batch iterate over my dataset. So say my dataset is 5000 lines long, I am dividing this into 1000 blocks, and then passing this blocks through a loop, where I am tokenizing the block and then appending the result to the original ‘inputs’.</p>\n<p>The loop then goes back to the top, and gets the next batch however in doing so it is overwriting the previous block, so the resultant product is a tensor with length of 1000 (containing the last 1000 block)</p>\n<pre><code class=\"lang-auto\">def batchReader(Dataset, block_size=1000):\n block = []\n for line in Dataset:\n block.append(line)\n if len(block) == block_size:\n yield block\n block = []\n if block:\n yield block\n\ninput_ids = []\nattention_masks = []\ntoken_type_ids = []\ncount=0\nwith open('5000.txt') as Dataset:\n blocks = batchReader(Dataset)\n for block in blocks:\n inputs = tokenizer(\n block, \n truncation=True,\n max_length=512,\n padding='max_length',\n return_tensors='pt', \n )\n input_ids.append(inputs['input_ids'])\n attention_masks.append(inputs['attention_mask'])\n token_type_ids.append(inputs['token_type_ids'])\n inputs['labels'] = inputs.input_ids.detach().clone()\n</code></pre>\n<p>I am looking for the output of this code to yield a tensor with 5000 length, any help much appreciated!</p>",12 "post_number": 1,13 "post_type": 1,14 "posts_count": 1,15 "updated_at": "2022-01-18T17:10:20.989Z",16 "reply_count": 0,17 "reply_to_post_number": null,18 "quote_count": 0,19 "incoming_link_count": 9,20 "reads": 4,21 "readers_count": 3,22 "score": 45.8,23 "yours": false,24 "topic_id": 141954,25 "topic_slug": "not-able-to-append-multiple-batches-to-single-output-whilst-tokenizing",26 "display_username": "Sean Farrell",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": 51066,48 "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/not-able-to-append-multiple-batches-to-single-output-whilst-tokenizing/141954/1",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 32715065 ]66 },67 "timeline_lookup": [68 [69 1,70 137671 ]72 ],73 "suggested_topics": [74 {75 "fancy_title": "TransformerDecoder masks shape error using model.eval()",76 "id": 215775,77 "title": "TransformerDecoder masks shape error using model.eval()",78 "slug": "transformerdecoder-masks-shape-error-using-model-eval",79 "posts_count": 4,80 "reply_count": 1,81 "highest_post_number": 4,82 "image_url": null,83 "created_at": "2025-01-23T14:27:08.766Z",84 "last_posted_at": "2025-01-27T15:14:26.282Z",85 "bumped": true,86 "bumped_at": "2025-01-27T15:14:26.282Z",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": 267,99 "category_id": 8,100 "featured_link": null,101 "has_accepted_answer": false,102 "posters": [103 {104 "extras": "latest",105 "description": "Original Poster, Most Recent Poster",106 "user": {107 "id": 82267,108 "username": "Mateo-drr",109 "name": "Mateo Rodriguez",110 "avatar_template": "/user_avatar/discuss.pytorch.org/mateo-drr/{size}/75262_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 },129 {130 "fancy_title": "Understanding logits in GPT2",131 "id": 213865,132 "title": "Understanding logits in GPT2",133 "slug": "understanding-logits-in-gpt2",134 "posts_count": 1,135 "reply_count": 0,136 "highest_post_number": 1,137 "image_url": null,138 "created_at": "2024-12-05T15:14:26.136Z",139 "last_posted_at": "2024-12-05T15:14:26.194Z",140 "bumped": true,141 "bumped_at": "2024-12-05T15:14:26.194Z",142 "archetype": "regular",143 "unseen": false,144 "pinned": false,145 "unpinned": null,146 "visible": true,147 "closed": false,148 "archived": false,149 "bookmarked": null,150 "liked": null,151 "tags_descriptions": {},152 "like_count": 0,153 "views": 200,154 "category_id": 8,155 "featured_link": null,156 "has_accepted_answer": false,157 "posters": [158 {159 "extras": "latest single",160 "description": "Original Poster, Most Recent Poster",161 "user": {162 "id": 81335,163 "username": "firolommones3",164 "name": "",165 "avatar_template": "/letter_avatar_proxy/v4/letter/f/b2d939/{size}.png",166 "trust_level": 0167 }168 }169 ]170 },171 {172 "fancy_title": "Flex_attention returning logits",173 "id": 213820,174 "title": "Flex_attention returning logits",175 "slug": "flex-attention-returning-logits",176 "posts_count": 1,177 "reply_count": 0,178 "highest_post_number": 1,179 "image_url": null,180 "created_at": "2024-12-04T23:23:39.405Z",181 "last_posted_at": "2024-12-04T23:23:39.455Z",182 "bumped": true,183 "bumped_at": "2024-12-04T23:23:39.455Z",184 "archetype": "regular",185 "unseen": false,186 "pinned": false,187 "unpinned": null,188 "visible": true,189 "closed": false,190 "archived": false,191 "bookmarked": null,192 "liked": null,193 "tags_descriptions": {},194 "like_count": 0,195 "views": 103,196 "category_id": 8,197 "featured_link": null,198 "has_accepted_answer": false,199 "posters": [200 {201 "extras": "latest single",202 "description": "Original Poster, Most Recent Poster",203 "user": {204 "id": 81310,205 "username": "pikantrop",206 "name": "",207 "avatar_template": "/letter_avatar_proxy/v4/letter/p/b77776/{size}.png",208 "trust_level": 1209 }210 }211 ]212 },213 {214 "fancy_title": "How to Implement Flash Attention in a Pre-Trained BERT Model on custom dataset?",215 "id": 215112,216 "title": "How to Implement Flash Attention in a Pre-Trained BERT Model on custom dataset?",217 "slug": "how-to-implement-flash-attention-in-a-pre-trained-bert-model-on-custom-dataset",218 "posts_count": 1,219 "reply_count": 0,220 "highest_post_number": 1,221 "image_url": null,222 "created_at": "2025-01-08T10:01:45.049Z",223 "last_posted_at": "2025-01-08T10:01:45.095Z",224 "bumped": true,225 "bumped_at": "2025-01-08T11:59:12.564Z",226 "archetype": "regular",227 "unseen": false,228 "pinned": false,229 "unpinned": null,230 "visible": true,231 "closed": false,232 "archived": false,233 "bookmarked": null,234 "liked": null,235 "tags_descriptions": {},236 "like_count": 0,237 "views": 197,238 "category_id": 8,239 "featured_link": null,240 "has_accepted_answer": false,241 "posters": [242 {243 "extras": "latest single",244 "description": "Original Poster, Most Recent Poster",245 "user": {246 "id": 81942,247 "username": "Karam1533",248 "name": "Karam",249 "avatar_template": "/user_avatar/discuss.pytorch.org/karam1533/{size}/74966_2.png",250 "trust_level": 0251 }252 }253 ]254 },255 {256 "fancy_title": "combining TEXT.build_vocab with BERT Embedding",257 "id": 215914,258 "title": "combining TEXT.build_vocab with BERT Embedding",259 "slug": "combining-text-build-vocab-with-bert-embedding",260 "posts_count": 1,261 "reply_count": 0,262 "highest_post_number": 1,263 "image_url": null,264 "created_at": "2025-01-27T03:28:57.212Z",265 "last_posted_at": "2025-01-27T03:28:57.243Z",266 "bumped": true,267 "bumped_at": "2025-01-27T03:28:57.243Z",268 "archetype": "regular",269 "unseen": false,270 "pinned": false,271 "unpinned": null,272 "visible": true,273 "closed": false,274 "archived": false,275 "bookmarked": null,276 "liked": null,277 "tags_descriptions": {},278 "like_count": 0,279 "views": 70,280 "category_id": 8,281 "featured_link": null,282 "has_accepted_answer": false,283 "posters": [284 {285 "extras": "latest single",286 "description": "Original Poster, Most Recent Poster",287 "user": {288 "id": 82334,289 "username": "Muhammad_Alfian",290 "name": "Muhammad Alfian",291 "avatar_template": "/user_avatar/discuss.pytorch.org/muhammad_alfian/{size}/75314_2.png",292 "trust_level": 1293 }294 }295 ]296 }297 ],298 "tags_descriptions": {},299 "fancy_title": "Not able to append multiple batches to single output whilst tokenizing",300 "id": 141954,301 "title": "Not able to append multiple batches to single output whilst tokenizing",302 "posts_count": 1,303 "created_at": "2022-01-18T17:10:20.932Z",304 "views": 520,305 "reply_count": 0,306 "like_count": 0,307 "last_posted_at": "2022-01-18T17:10:20.989Z",308 "visible": true,309 "closed": false,310 "archived": false,311 "has_summary": false,312 "archetype": "regular",313 "slug": "not-able-to-append-multiple-batches-to-single-output-whilst-tokenizing",314 "category_id": 8,315 "word_count": 177,316 "deleted_at": null,317 "user_id": 51066,318 "featured_link": null,319 "pinned_globally": false,320 "pinned_at": null,321 "pinned_until": null,322 "image_url": null,323 "slow_mode_seconds": 0,324 "draft": null,325 "draft_key": "topic_141954",326 "draft_sequence": null,327 "unpinned": null,328 "pinned": false,329 "current_post_number": 1,330 "highest_post_number": 1,331 "deleted_by": null,332 "actions_summary": [333 {334 "id": 4,335 "count": 0,336 "hidden": false,337 "can_act": false338 },339 {340 "id": 8,341 "count": 0,342 "hidden": false,343 "can_act": false344 },345 {346 "id": 10,347 "count": 0,348 "hidden": false,349 "can_act": false350 },351 {352 "id": 7,353 "count": 0,354 "hidden": false,355 "can_act": false356 }357 ],358 "chunk_size": 20,359 "bookmarked": false,360 "topic_timer": null,361 "message_bus_last_id": 0,362 "participant_count": 1,363 "show_read_indicator": false,364 "thumbnails": null,365 "slow_mode_enabled_until": null,366 "can_vote": false,367 "vote_count": 0,368 "user_voted": false,369 "discourse_zendesk_plugin_zendesk_id": null,370 "discourse_zendesk_plugin_zendesk_url": "https://your-url.zendesk.com/agent/tickets/",371 "details": {372 "can_edit": false,373 "notification_level": 1,374 "participants": [375 {376 "id": 51066,377 "username": "seanfarr788",378 "name": "Sean Farrell",379 "avatar_template": "/user_avatar/discuss.pytorch.org/seanfarr788/{size}/44321_2.png",380 "post_count": 1,381 "primary_group_name": null,382 "flair_name": null,383 "flair_url": null,384 "flair_color": null,385 "flair_bg_color": null,386 "flair_group_id": null,387 "trust_level": 1388 }389 ],390 "created_by": {391 "id": 51066,392 "username": "seanfarr788",393 "name": "Sean Farrell",394 "avatar_template": "/user_avatar/discuss.pytorch.org/seanfarr788/{size}/44321_2.png"395 },396 "last_poster": {397 "id": 51066,398 "username": "seanfarr788",399 "name": "Sean Farrell",400 "avatar_template": "/user_avatar/discuss.pytorch.org/seanfarr788/{size}/44321_2.png"401 }402 },403 "bookmarks": []404 },405 {406 "post_stream": {407 "posts": [408 {409 "id": 327148,410 "name": "Khabbab Zakaria",411 "username": "Khabbab_Zakaria",412 "avatar_template": "/user_avatar/discuss.pytorch.org/khabbab_zakaria/{size}/33485_2.png",413 "created_at": "2022-01-18T16:42:55.789Z",414 "cooked": "<p>Hi,<br>\nI am working on an Actor Critic Method where there are infinite numbers of possible actions (let’s say movement of a robot, where the possible movement is not just left-right-up-down but every possible angle).<br>\nMy obvious idea was to have output mu and sigma^2 in the policy:</p>\n<pre><code class=\"lang-auto\">class Policy(nn.Module):\n \"\"\"\n implements both actor and critic in one model\n \"\"\"\n def __init__(self):\n super(Policy, self).__init__()\n self.fc1 = nn.Linear(state_size+1, 128)\n\n self.fc2 = nn.Linear(128, 64)\n\n # actor's layer\n self.action_head = nn.Linear(64, 1)\n self.mu = nn.Sigmoid()\n self.var = nn.Softplus()\n\n # critic's layer\n self.value_head = nn.Linear(64, 1)\n\n\n def forward(self, x):\n \"\"\"\n forward of both actor and critic\n \"\"\"\n x = F.relu(self.fc1(x))\n x = F.relu(self.fc2(x))\n\n action_prob = self.action_head(x)\n mu = self.mu(action_prob)\n var = self.var(action_prob)\n\n state_values = self.value_head(x)\n\n return mu, var, state_values\n</code></pre>\n<p>Now, I also need to calculate the action from this (mu, sigma**2=var) pair. I also need to calculate the probability of happening of that action from that Normal Distribution. I am doing these for them:</p>\n<pre><code class=\"lang-auto\">sigma = torch.sqrt(var)\naction = torch.normal(mu, sigma)\naction = torch.clip(action, 0, 1)\npdf_probability = stats.norm.pdf(action.cpu().detach().numpy(), loc=mu.cpu().detach().numpy(), scale=sigma.cpu().detach().numpy())\n</code></pre>\n<p>Now I have some questions. What I am doing, is it ok? It does not feel ok, as I suppose the pdf_probability should be backpropagated as well, but during conversion Tensor->Numpy->Tensor, we lose it.<br>\nI have gone through <a href=\"https://discuss.pytorch.org/t/resolved-actor-critic-with-a-large-amount-of-possible-actions/3933/5\">[resolved] Actor Critic with a large amount of possible actions - reinforcement-learning - PyTorch Forums</a>, where they discussed a similar issue. They are also calculating mu, sigma^2 from the policy, but they never talked about how to calculate Action from these (mu,sigma^2).<br>\nI also read the A3C paper <a href=\"https://arxiv.org/pdf/1602.01783v2.pdf\" rel=\"noopener nofollow ugc\">Asynchronous Methods for Deep Reinforcement Learning (arxiv.org)</a>, where they stated that the mu should be calculated by a Linear Layer. Is it mandatory? In my case, I want to only consider positive angles, so, mu should be followed by Sigmoid or ReLU right?</p>",415 "post_number": 1,416 "post_type": 1,417 "posts_count": 1,418 "updated_at": "2022-01-18T16:48:39.508Z",419 "reply_count": 0,420 "reply_to_post_number": null,421 "quote_count": 0,422 "incoming_link_count": 153,423 "reads": 5,424 "readers_count": 4,425 "score": 766.0,426 "yours": false,427 "topic_id": 141952,428 "topic_slug": "how-to-find-action-in-an-actor-critic-method-with-infinite-possible-actions",429 "display_username": "Khabbab Zakaria",430 "primary_group_name": null,431 "flair_name": null,432 "flair_url": null,433 "flair_bg_color": null,434 "flair_color": null,435 "flair_group_id": null,436 "badges_granted": [],437 "version": 2,438 "can_edit": false,439 "can_delete": false,440 "can_recover": false,441 "can_see_hidden_post": false,442 "can_wiki": false,443 "link_counts": [444 {445 "url": "https://discuss.pytorch.org/t/resolved-actor-critic-with-a-large-amount-of-possible-actions/3933/5",446 "internal": true,447 "reflection": false,448 "title": "[resolved] Actor Critic with a large amount of possible actions",449 "clicks": 2450 },451 {452 "url": "https://arxiv.org/pdf/1602.01783v2.pdf",453 "internal": false,454 "reflection": false,455 "clicks": 0456 }457 ],458 "read": true,459 "user_title": null,460 "bookmarked": false,461 "actions_summary": [],462 "moderator": false,463 "admin": false,464 "staff": false,465 "user_id": 42520,466 "hidden": false,467 "trust_level": 1,468 "deleted_at": null,469 "user_deleted": false,470 "edit_reason": null,471 "can_view_edit_history": true,472 "wiki": false,473 "post_url": "/t/how-to-find-action-in-an-actor-critic-method-with-infinite-possible-actions/141952/1",474 "can_accept_answer": false,475 "can_unaccept_answer": false,476 "accepted_answer": false,477 "topic_accepted_answer": null,478 "can_vote": false479 }480 ],481 "stream": [482 327148483 ]484 },485 "timeline_lookup": [486 [487 1,488 1376489 ]490 ],491 "suggested_topics": [492 {493 "fancy_title": "Understanding the traceback of a compiler error with cache size limit",494 "id": 215669,495 "title": "Understanding the traceback of a compiler error with cache size limit",496 "slug": "understanding-the-traceback-of-a-compiler-error-with-cache-size-limit",497 "posts_count": 1,498 "reply_count": 0,499 "highest_post_number": 1,500 "image_url": null,501 "created_at": "2025-01-21T10:43:52.773Z",502 "last_posted_at": "2025-01-21T10:43:52.812Z",503 "bumped": true,504 "bumped_at": "2025-01-21T10:43:52.812Z",505 "archetype": "regular",506 "unseen": false,507 "pinned": false,508 "unpinned": null,509 "visible": true,510 "closed": false,511 "archived": false,512 "bookmarked": null,513 "liked": null,514 "tags_descriptions": {},515 "like_count": 0,516 "views": 53,517 "category_id": 1,518 "featured_link": null,519 "has_accepted_answer": false,520 "posters": [521 {522 "extras": "latest single",523 "description": "Original Poster, Most Recent Poster",524 "user": {525 "id": 82217,526 "username": "Johannes_Vogt",527 "name": "Johannes Vogt",528 "avatar_template": "/user_avatar/discuss.pytorch.org/johannes_vogt/{size}/75220_2.png",529 "trust_level": 1530 }531 }532 ]533 },534 {535 "fancy_title": "NVIDIA L40S-48Q and “RuntimeError: CUDA error: operation not supported”",536 "id": 212716,537 "title": "NVIDIA L40S-48Q and \"RuntimeError: CUDA error: operation not supported\"",538 "slug": "nvidia-l40s-48q-and-runtimeerror-cuda-error-operation-not-supported",539 "posts_count": 11,540 "reply_count": 9,541 "highest_post_number": 11,542 "image_url": null,543 "created_at": "2024-11-08T19:29:41.292Z",544 "last_posted_at": "2025-02-11T14:25:17.459Z",545 "bumped": true,546 "bumped_at": "2025-02-11T14:26:35.274Z",547 "archetype": "regular",548 "unseen": false,549 "pinned": false,550 "unpinned": null,551 "visible": true,552 "closed": false,553 "archived": false,554 "bookmarked": null,555 "liked": null,556 "tags_descriptions": {},557 "like_count": 1,558 "views": 1360,559 "category_id": 1,560 "featured_link": null,561 "has_accepted_answer": false,562 "posters": [563 {564 "extras": null,565 "description": "Original Poster",566 "user": {567 "id": 7291,568 "username": "Chris_Palmer",569 "name": "Chris Palmer",570 "avatar_template": "/user_avatar/discuss.pytorch.org/chris_palmer/{size}/12322_2.png",571 "trust_level": 1572 }573 },574 {575 "extras": null,576 "description": "Frequent Poster",577 "user": {578 "id": 3534,579 "username": "ptrblck",580 "name": "",581 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",582 "admin": true,583 "moderator": true,584 "trust_level": 2585 }586 },587 {588 "extras": "latest",589 "description": "Most Recent Poster",590 "user": {591 "id": 82625,592 "username": "briskajanis1",593 "name": "Briskajanis1",594 "avatar_template": "/user_avatar/discuss.pytorch.org/briskajanis1/{size}/75599_2.png",595 "trust_level": 0596 }597 }598 ]599 },600 {601 "fancy_title": "Uint64 tensors do not wraparound on overflow",602 "id": 213926,603 "title": "Uint64 tensors do not wraparound on overflow",604 "slug": "uint64-tensors-do-not-wraparound-on-overflow",605 "posts_count": 3,606 "reply_count": 1,607 "highest_post_number": 3,608 "image_url": null,609 "created_at": "2024-12-06T20:49:54.790Z",610 "last_posted_at": "2024-12-10T01:59:00.010Z",611 "bumped": true,612 "bumped_at": "2024-12-10T01:59:00.010Z",613 "archetype": "regular",614 "unseen": false,615 "pinned": false,616 "unpinned": null,617 "visible": true,618 "closed": false,619 "archived": false,620 "bookmarked": null,621 "liked": null,622 "tags_descriptions": {},623 "like_count": 0,624 "views": 204,625 "category_id": 1,626 "featured_link": null,627 "has_accepted_answer": true,628 "posters": [629 {630 "extras": "latest",631 "description": "Original Poster, Most Recent Poster",632 "user": {633 "id": 31035,634 "username": "kmaeng",635 "name": "",636 "avatar_template": "/user_avatar/discuss.pytorch.org/kmaeng/{size}/23707_2.png",637 "trust_level": 1638 }639 },640 {641 "extras": null,642 "description": "Frequent Poster, Accepted Answer",643 "user": {644 "id": 18088,645 "username": "KFrank",646 "name": "K. Frank",647 "avatar_template": "/letter_avatar_proxy/v4/letter/k/ecb155/{size}.png",648 "trust_level": 2649 }650 }651 ]652 },653 {654 "fancy_title": "CUDA not available",655 "id": 215422,656 "title": "CUDA not available",657 "slug": "cuda-not-available",658 "posts_count": 8,659 "reply_count": 6,660 "highest_post_number": 8,661 "image_url": null,662 "created_at": "2025-01-15T13:11:20.130Z",663 "last_posted_at": "2025-01-16T16:46:47.831Z",664 "bumped": true,665 "bumped_at": "2025-01-16T16:46:47.831Z",666 "archetype": "regular",667 "unseen": false,668 "pinned": false,669 "unpinned": null,670 "visible": true,671 "closed": false,672 "archived": false,673 "bookmarked": null,674 "liked": null,675 "tags_descriptions": {},676 "like_count": 0,677 "views": 1418,678 "category_id": 1,679 "featured_link": null,680 "has_accepted_answer": true,681 "posters": [682 {683 "extras": null,684 "description": "Original Poster",685 "user": {686 "id": 46535,687 "username": "MoRoBe",688 "name": "",689 "avatar_template": "/user_avatar/discuss.pytorch.org/morobe/{size}/39577_2.png",690 "trust_level": 1691 }692 },693 {694 "extras": "latest",695 "description": "Most Recent Poster, Accepted Answer",696 "user": {697 "id": 18088,698 "username": "KFrank",699 "name": "K. Frank",700 "avatar_template": "/letter_avatar_proxy/v4/letter/k/ecb155/{size}.png",701 "trust_level": 2702 }703 },704 {705 "extras": null,706 "description": "Frequent Poster",707 "user": {708 "id": 3534,709 "username": "ptrblck",710 "name": "",711 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",712 "admin": true,713 "moderator": true,714 "trust_level": 2715 }716 }717 ]718 },719 {720 "fancy_title": "How to Perform Multiple Inferences on Multiple GPUs in PyTorch?",721 "id": 216826,722 "title": "How to Perform Multiple Inferences on Multiple GPUs in PyTorch?",723 "slug": "how-to-perform-multiple-inferences-on-multiple-gpus-in-pytorch",724 "posts_count": 3,725 "reply_count": 1,726 "highest_post_number": 3,727 "image_url": null,728 "created_at": "2025-02-18T13:31:19.033Z",729 "last_posted_at": "2025-02-19T10:55:51.344Z",730 "bumped": true,731 "bumped_at": "2025-02-19T10:55:51.344Z",732 "archetype": "regular",733 "unseen": false,734 "pinned": false,735 "unpinned": null,736 "visible": true,737 "closed": false,738 "archived": false,739 "bookmarked": null,740 "liked": null,741 "tags_descriptions": {},742 "like_count": 0,743 "views": 128,744 "category_id": 1,745 "featured_link": null,746 "has_accepted_answer": false,747 "posters": [748 {749 "extras": "latest",750 "description": "Original Poster, Most Recent Poster",751 "user": {752 "id": 82761,753 "username": "dmp58",754 "name": "",755 "avatar_template": "/user_avatar/discuss.pytorch.org/dmp58/{size}/75721_2.png",756 "trust_level": 0757 }758 },759 {760 "extras": null,761 "description": "Frequent Poster",762 "user": {763 "id": 3534,764 "username": "ptrblck",765 "name": "",766 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",767 "admin": true,768 "moderator": true,769 "trust_level": 2770 }771 }772 ]773 }774 ],775 "tags_descriptions": {},776 "fancy_title": "How to find Action in an Actor Critic Method with infinite possible actions",777 "id": 141952,778 "title": "How to find Action in an Actor Critic Method with infinite possible actions",779 "posts_count": 1,780 "created_at": "2022-01-18T16:42:55.703Z",781 "views": 803,782 "reply_count": 0,783 "like_count": 0,784 "last_posted_at": "2022-01-18T16:42:55.789Z",785 "visible": true,786 "closed": false,787 "archived": false,788 "has_summary": false,789 "archetype": "regular",790 "slug": "how-to-find-action-in-an-actor-critic-method-with-infinite-possible-actions",791 "category_id": 1,792 "word_count": 381,793 "deleted_at": null,794 "user_id": 42520,795 "featured_link": null,796 "pinned_globally": false,797 "pinned_at": null,798 "pinned_until": null,799 "image_url": null,800 "slow_mode_seconds": 0,801 "draft": null,802 "draft_key": "topic_141952",803 "draft_sequence": null,804 "unpinned": null,805 "pinned": false,806 "current_post_number": 1,807 "highest_post_number": 1,808 "deleted_by": null,809 "actions_summary": [810 {811 "id": 4,812 "count": 0,813 "hidden": false,814 "can_act": false815 },816 {817 "id": 8,818 "count": 0,819 "hidden": false,820 "can_act": false821 },822 {823 "id": 10,824 "count": 0,825 "hidden": false,826 "can_act": false827 },828 {829 "id": 7,830 "count": 0,831 "hidden": false,832 "can_act": false833 }834 ],835 "chunk_size": 20,836 "bookmarked": false,837 "topic_timer": null,838 "message_bus_last_id": 0,839 "participant_count": 1,840 "show_read_indicator": false,841 "thumbnails": null,842 "slow_mode_enabled_until": null,843 "can_vote": false,844 "vote_count": 0,845 "user_voted": false,846 "discourse_zendesk_plugin_zendesk_id": null,847 "discourse_zendesk_plugin_zendesk_url": "https://your-url.zendesk.com/agent/tickets/",848 "details": {849 "can_edit": false,850 "notification_level": 1,851 "participants": [852 {853 "id": 42520,854 "username": "Khabbab_Zakaria",855 "name": "Khabbab Zakaria",856 "avatar_template": "/user_avatar/discuss.pytorch.org/khabbab_zakaria/{size}/33485_2.png",857 "post_count": 1,858 "primary_group_name": null,859 "flair_name": null,860 "flair_url": null,861 "flair_color": null,862 "flair_bg_color": null,863 "flair_group_id": null,864 "trust_level": 1865 }866 ],867 "created_by": {868 "id": 42520,869 "username": "Khabbab_Zakaria",870 "name": "Khabbab Zakaria",871 "avatar_template": "/user_avatar/discuss.pytorch.org/khabbab_zakaria/{size}/33485_2.png"872 },873 "last_poster": {874 "id": 42520,875 "username": "Khabbab_Zakaria",876 "name": "Khabbab Zakaria",877 "avatar_template": "/user_avatar/discuss.pytorch.org/khabbab_zakaria/{size}/33485_2.png"878 },879 "links": [880 {881 "url": "https://discuss.pytorch.org/t/resolved-actor-critic-with-a-large-amount-of-possible-actions/3933/5",882 "title": "[resolved] Actor Critic with a large amount of possible actions",883 "internal": true,884 "attachment": false,885 "reflection": false,886 "clicks": 2,887 "user_id": 42520,888 "domain": "discuss.pytorch.org",889 "root_domain": "pytorch.org"890 }891 ]892 },893 "bookmarks": []894 },895 {896 "post_stream": {897 "posts": [898 {899 "id": 327144,900 "name": "Hao Peng",901 "username": "Hao_Peng",902 "avatar_template": "/user_avatar/discuss.pytorch.org/hao_peng/{size}/45872_2.png",903 "created_at": "2022-01-18T15:38:31.040Z",904 "cooked": "<p>Hi guys,</p>\n<p>Thanks for your excellent job!</p>\n<p>I had some troubles using PyTorch, especially when I needed to solve the sparse linear equations Ax=b (usually by LU decomposition). Unfortunately, PyTorch doesn’t seem to have this feature.<br>\nI wonder if you have any plans to add this feature?<br>\nI found a great library to do LU decomposition based on GPU, but it’s only in c++.<br>\n<a href=\"https://github.com/sheldonucr/GLU_public\" rel=\"noopener nofollow ugc\">sheldonucr/GLU_public: GLU - GLU-accelerated Sparse Parellel LU factorization solver V3.0 (github.com)</a></p>\n<p>Would you like to add a Pytorch wrapper for it?</p>\n<p>Many thanks!</p>",905 "post_number": 1,906 "post_type": 1,907 "posts_count": 1,908 "updated_at": "2022-01-18T15:38:31.040Z",909 "reply_count": 0,910 "reply_to_post_number": null,911 "quote_count": 0,912 "incoming_link_count": 98,913 "reads": 4,914 "readers_count": 3,915 "score": 490.8,916 "yours": false,917 "topic_id": 141948,918 "topic_slug": "addinf-gpu-based-lu-decomposition-of-sparse-matrix",919 "display_username": "Hao Peng",920 "primary_group_name": null,921 "flair_name": null,922 "flair_url": null,923 "flair_bg_color": null,924 "flair_color": null,925 "flair_group_id": null,926 "badges_granted": [],927 "version": 1,928 "can_edit": false,929 "can_delete": false,930 "can_recover": false,931 "can_see_hidden_post": false,932 "can_wiki": false,933 "link_counts": [934 {935 "url": "https://github.com/sheldonucr/GLU_public",936 "internal": false,937 "reflection": false,938 "title": "GitHub - sheldonucr/GLU_public: GLU - GLU-accelerated Sparse Parellel LU factorization solver V3.0",939 "clicks": 19940 }941 ],942 "read": true,943 "user_title": null,944 "bookmarked": false,945 "actions_summary": [],946 "moderator": false,947 "admin": false,948 "staff": false,949 "user_id": 52486,950 "hidden": false,951 "trust_level": 1,952 "deleted_at": null,953 "user_deleted": false,954 "edit_reason": null,955 "can_view_edit_history": true,956 "wiki": false,957 "post_url": "/t/addinf-gpu-based-lu-decomposition-of-sparse-matrix/141948/1",958 "can_accept_answer": false,959 "can_unaccept_answer": false,960 "accepted_answer": false,961 "topic_accepted_answer": null,962 "can_vote": false963 }964 ],965 "stream": [966 327144967 ]968 },969 "timeline_lookup": [970 [971 1,972 1376973 ]974 ],975 "suggested_topics": [976 {977 "fancy_title": "Extracting data from cudnnBackendDescriptor_t",978 "id": 218926,979 "title": "Extracting data from cudnnBackendDescriptor_t",980 "slug": "extracting-data-from-cudnnbackenddescriptor-t",981 "posts_count": 1,982 "reply_count": 0,983 "highest_post_number": 1,984 "image_url": null,985 "created_at": "2025-04-10T05:02:56.937Z",986 "last_posted_at": "2025-04-10T05:02:56.979Z",987 "bumped": true,988 "bumped_at": "2025-04-10T05:02:56.979Z",989 "archetype": "regular",990 "unseen": false,991 "pinned": false,992 "unpinned": null,993 "visible": true,994 "closed": false,995 "archived": false,996 "bookmarked": null,997 "liked": null,998 "tags_descriptions": {},999 "like_count": 0,1000 "views": 42,1001 "category_id": 11,1002 "featured_link": null,1003 "has_accepted_answer": false,1004 "posters": [1005 {1006 "extras": "latest single",1007 "description": "Original Poster, Most Recent Poster",1008 "user": {1009 "id": 83734,1010 "username": "chanheenam",1011 "name": "chanheenam",1012 "avatar_template": "/letter_avatar_proxy/v4/letter/c/6f9a4e/{size}.png",1013 "trust_level": 01014 }1015 }1016 ]1017 },1018 {1019 "fancy_title": "Libtorch: torch::autograd::grad create_graph and retain_graph malfunctioning?",1020 "id": 212368,1021 "title": "Libtorch: torch::autograd::grad create_graph and retain_graph malfunctioning?",1022 "slug": "libtorch-torch-grad-create-graph-and-retain-graph-malfunctioning",1023 "posts_count": 1,1024 "reply_count": 0,1025 "highest_post_number": 1,1026 "image_url": null,1027 "created_at": "2024-10-31T14:04:51.642Z",1028 "last_posted_at": "2024-10-31T14:04:51.705Z",1029 "bumped": true,1030 "bumped_at": "2024-10-31T14:04:51.705Z",1031 "archetype": "regular",1032 "unseen": false,1033 "pinned": false,1034 "unpinned": null,1035 "visible": true,1036 "closed": false,1037 "archived": false,1038 "bookmarked": null,1039 "liked": null,1040 "unicode_title": "Libtorch: torch::autograd::grad create_graph and retain_graph malfunctioning?",1041 "tags_descriptions": {},1042 "like_count": 0,1043 "views": 33,1044 "category_id": 11,1045 "featured_link": null,1046 "has_accepted_answer": false,1047 "posters": [1048 {1049 "extras": "latest single",1050 "description": "Original Poster, Most Recent Poster",1051 "user": {1052 "id": 78568,1053 "username": "Rsmeets",1054 "name": "Robin Smeets",1055 "avatar_template": "/letter_avatar_proxy/v4/letter/r/ed8c4c/{size}.png",1056 "trust_level": 11057 }1058 }1059 ]1060 },1061 {1062 "fancy_title": "LibTorch symbols getting overriden when linking to openBLAS",1063 "id": 212811,1064 "title": "LibTorch symbols getting overriden when linking to openBLAS",1065 "slug": "libtorch-symbols-getting-overriden-when-linking-to-openblas",1066 "posts_count": 1,1067 "reply_count": 0,1068 "highest_post_number": 1,1069 "image_url": null,1070 "created_at": "2024-11-11T16:59:25.913Z",1071 "last_posted_at": "2024-11-11T16:59:25.969Z",1072 "bumped": true,1073 "bumped_at": "2024-11-11T17:58:05.352Z",1074 "archetype": "regular",1075 "unseen": false,1076 "pinned": false,1077 "unpinned": null,1078 "visible": true,1079 "closed": false,1080 "archived": false,1081 "bookmarked": null,1082 "liked": null,1083 "tags_descriptions": {},1084 "like_count": 0,1085 "views": 46,1086 "category_id": 11,1087 "featured_link": null,1088 "has_accepted_answer": false,1089 "posters": [1090 {1091 "extras": "latest single",1092 "description": "Original Poster, Most Recent Poster",1093 "user": {1094 "id": 13551,1095 "username": "dfalbel",1096 "name": "Daniel Falbel",1097 "avatar_template": "/user_avatar/discuss.pytorch.org/dfalbel/{size}/8880_2.png",1098 "trust_level": 21099 }1100 }1101 ]1102 },1103 {1104 "fancy_title": "Quession about including torch/extension",1105 "id": 216127,1106 "title": "Quession about including torch/extension",1107 "slug": "quession-about-including-torch-extension",1108 "posts_count": 1,1109 "reply_count": 0,1110 "highest_post_number": 1,1111 "image_url": "https://discuss.pytorch.org/uploads/default/original/3X/b/6/b6c1d6014497944935fb34ae023787bba5fff013.jpeg",1112 "created_at": "2025-02-01T18:32:17.095Z",1113 "last_posted_at": "2025-02-01T18:32:17.137Z",1114 "bumped": true,1115 "bumped_at": "2025-02-01T18:32:17.137Z",1116 "archetype": "regular",1117 "unseen": false,1118 "pinned": false,1119 "unpinned": null,1120 "visible": true,1121 "closed": false,1122 "archived": false,1123 "bookmarked": null,1124 "liked": null,1125 "tags_descriptions": {},1126 "like_count": 0,1127 "views": 125,1128 "category_id": 11,1129 "featured_link": null,1130 "has_accepted_answer": false,1131 "posters": [1132 {1133 "extras": "latest single",1134 "description": "Original Poster, Most Recent Poster",1135 "user": {1136 "id": 82431,1137 "username": "p3n9u1n",1138 "name": "郭宸瑋",1139 "avatar_template": "/user_avatar/discuss.pytorch.org/p3n9u1n/{size}/75413_2.png",1140 "trust_level": 11141 }1142 }1143 ]1144 },1145 {1146 "fancy_title": "How to use NestedTensor in C++?",1147 "id": 222094,1148 "title": "How to use NestedTensor in C++?",1149 "slug": "how-to-use-nestedtensor-in-c",1150 "posts_count": 3,1151 "reply_count": 1,1152 "highest_post_number": 3,1153 "image_url": null,1154 "created_at": "2025-08-05T14:25:26.752Z",1155 "last_posted_at": "2025-08-05T16:29:56.416Z",1156 "bumped": true,1157 "bumped_at": "2025-08-05T16:29:56.416Z",1158 "archetype": "regular",1159 "unseen": false,1160 "pinned": false,1161 "unpinned": null,1162 "visible": true,1163 "closed": false,1164 "archived": false,1165 "bookmarked": null,1166 "liked": null,1167 "tags_descriptions": {},1168 "like_count": 0,1169 "views": 37,1170 "category_id": 11,1171 "featured_link": null,1172 "has_accepted_answer": false,1173 "posters": [1174 {1175 "extras": "latest single",1176 "description": "Original Poster, Most Recent Poster",1177 "user": {1178 "id": 85373,1179 "username": "byrzhm",1180 "name": "Hongming Zhu",1181 "avatar_template": "/user_avatar/discuss.pytorch.org/byrzhm/{size}/78414_2.png",1182 "trust_level": 11183 }1184 }1185 ]1186 }1187 ],1188 "tags_descriptions": {},1189 "fancy_title": "Addinf GPU-based LU decomposition of sparse matrix",1190 "id": 141948,1191 "title": "Addinf GPU-based LU decomposition of sparse matrix",1192 "posts_count": 1,1193 "created_at": "2022-01-18T15:38:30.968Z",1194 "views": 576,1195 "reply_count": 0,1196 "like_count": 0,1197 "last_posted_at": "2022-01-18T15:38:31.040Z",1198 "visible": true,1199 "closed": false,1200 "archived": false,