Matches in SemOpenAlex for { <https://semopenalex.org/work/W4384705429> ?p ?o ?g. }
Showing items 1 to 98 of
98
with 100 items per page.
- W4384705429 abstract "Larger deep learning models usually lead to higher model quality, however with an ever-increasing GPU memory footprint. Although several tensor checkpointing techniques have been proposed to enable training under a restricted GPU memory budget, they fail to exploit the input tensor dynamics due to diverse datasets and subsequent data augmentation, and thus leave the training optimization on table. In this paper, we propose Mimose, an input-aware tensor checkpointing planner respecting the memory budget while enabling efficient model training on GPU. Mimose builds a lightweight but accurate prediction model of GPU memory usage online, without pre-analyzing the model. It generates a tensor checkpointing plan based on per-layer memory prediction and applies it to the training process on the fly. Our experiments show that Mimose achieves superior training throughput compared to state-of-the-art checkpointing frameworks under the same GPU memory budgets." @default.
- W4384705429 created "2023-07-20" @default.
- W4384705429 creator A5001560763 @default.
- W4384705429 creator A5010232454 @default.
- W4384705429 creator A5018705589 @default.
- W4384705429 creator A5046015937 @default.
- W4384705429 creator A5046708261 @default.
- W4384705429 creator A5058313495 @default.
- W4384705429 creator A5060999547 @default.
- W4384705429 creator A5064154151 @default.
- W4384705429 creator A5069632578 @default.
- W4384705429 creator A5074183877 @default.
- W4384705429 creator A5075958056 @default.
- W4384705429 creator A5079362609 @default.
- W4384705429 creator A5086694700 @default.
- W4384705429 date "2023-05-01" @default.
- W4384705429 modified "2023-10-18" @default.
- W4384705429 title "Exploiting Input Tensor Dynamics in Activation Checkpointing for Efficient Training on GPU" @default.
- W4384705429 cites W2285660444 @default.
- W4384705429 cites W2330958039 @default.
- W4384705429 cites W2489529491 @default.
- W4384705429 cites W2883283076 @default.
- W4384705429 cites W2963351448 @default.
- W4384705429 cites W2979816092 @default.
- W4384705429 cites W3012479151 @default.
- W4384705429 cites W3096609285 @default.
- W4384705429 cites W3102015031 @default.
- W4384705429 cites W3102476541 @default.
- W4384705429 cites W3138516171 @default.
- W4384705429 cites W3157864729 @default.
- W4384705429 cites W3167436278 @default.
- W4384705429 cites W3172752666 @default.
- W4384705429 cites W3179431530 @default.
- W4384705429 cites W3189259198 @default.
- W4384705429 cites W3205803342 @default.
- W4384705429 cites W4282974849 @default.
- W4384705429 cites W4285148475 @default.
- W4384705429 doi "https://doi.org/10.1109/ipdps54959.2023.00025" @default.
- W4384705429 hasPublicationYear "2023" @default.
- W4384705429 type Work @default.
- W4384705429 citedByCount "0" @default.
- W4384705429 crossrefType "proceedings-article" @default.
- W4384705429 hasAuthorship W4384705429A5001560763 @default.
- W4384705429 hasAuthorship W4384705429A5010232454 @default.
- W4384705429 hasAuthorship W4384705429A5018705589 @default.
- W4384705429 hasAuthorship W4384705429A5046015937 @default.
- W4384705429 hasAuthorship W4384705429A5046708261 @default.
- W4384705429 hasAuthorship W4384705429A5058313495 @default.
- W4384705429 hasAuthorship W4384705429A5060999547 @default.
- W4384705429 hasAuthorship W4384705429A5064154151 @default.
- W4384705429 hasAuthorship W4384705429A5069632578 @default.
- W4384705429 hasAuthorship W4384705429A5074183877 @default.
- W4384705429 hasAuthorship W4384705429A5075958056 @default.
- W4384705429 hasAuthorship W4384705429A5079362609 @default.
- W4384705429 hasAuthorship W4384705429A5086694700 @default.
- W4384705429 hasConcept C111919701 @default.
- W4384705429 hasConcept C154945302 @default.
- W4384705429 hasConcept C155281189 @default.
- W4384705429 hasConcept C157764524 @default.
- W4384705429 hasConcept C165696696 @default.
- W4384705429 hasConcept C173608175 @default.
- W4384705429 hasConcept C202444582 @default.
- W4384705429 hasConcept C33923547 @default.
- W4384705429 hasConcept C38652104 @default.
- W4384705429 hasConcept C41008148 @default.
- W4384705429 hasConcept C555944384 @default.
- W4384705429 hasConcept C74912251 @default.
- W4384705429 hasConcept C98045186 @default.
- W4384705429 hasConceptScore W4384705429C111919701 @default.
- W4384705429 hasConceptScore W4384705429C154945302 @default.
- W4384705429 hasConceptScore W4384705429C155281189 @default.
- W4384705429 hasConceptScore W4384705429C157764524 @default.
- W4384705429 hasConceptScore W4384705429C165696696 @default.
- W4384705429 hasConceptScore W4384705429C173608175 @default.
- W4384705429 hasConceptScore W4384705429C202444582 @default.
- W4384705429 hasConceptScore W4384705429C33923547 @default.
- W4384705429 hasConceptScore W4384705429C38652104 @default.
- W4384705429 hasConceptScore W4384705429C41008148 @default.
- W4384705429 hasConceptScore W4384705429C555944384 @default.
- W4384705429 hasConceptScore W4384705429C74912251 @default.
- W4384705429 hasConceptScore W4384705429C98045186 @default.
- W4384705429 hasFunder F4320321001 @default.
- W4384705429 hasLocation W43847054291 @default.
- W4384705429 hasOpenAccess W4384705429 @default.
- W4384705429 hasPrimaryLocation W43847054291 @default.
- W4384705429 hasRelatedWork W1534022569 @default.
- W4384705429 hasRelatedWork W1555721731 @default.
- W4384705429 hasRelatedWork W2047588290 @default.
- W4384705429 hasRelatedWork W2152018389 @default.
- W4384705429 hasRelatedWork W2331043530 @default.
- W4384705429 hasRelatedWork W2374820792 @default.
- W4384705429 hasRelatedWork W2393933887 @default.
- W4384705429 hasRelatedWork W2964604098 @default.
- W4384705429 hasRelatedWork W2997512100 @default.
- W4384705429 hasRelatedWork W3202077128 @default.
- W4384705429 isParatext "false" @default.
- W4384705429 isRetracted "false" @default.
- W4384705429 workType "article" @default.