FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3041621704> ?p ?o ?g. }

• W3041621704 abstract "The last decade has witnessed growth in the computational requirements for training deep neural networks. Current approaches (e.g., data/model parallelism, pipeline parallelism) parallelize training tasks onto multiple devices. However, these approaches always rely on specific deep learning frameworks and requires elaborate manual design, which make it difficult to maintain and share between different type of models. In this paper, we propose Auto-MAP, a framework for exploring distributed execution plans for DNN workloads, which can automatically discovering fast parallelization strategies through reinforcement learning on IR level of deep learning models. Efficient exploration remains a major challenge for reinforcement learning. We leverage DQN with task-specific pruning strategies to help efficiently explore the search space including optimized strategies. Our evaluation shows that Auto-MAP can find the optimal solution in two hours, while achieving better throughput on several NLP and convolution models." @default.
• W3041621704 created "2020-07-16" @default.
• W3041621704 creator A5017996548 @default.
• W3041621704 creator A5020574416 @default.
• W3041621704 creator A5025904134 @default.
• W3041621704 creator A5046708388 @default.
• W3041621704 creator A5051978977 @default.
• W3041621704 creator A5052795742 @default.
• W3041621704 creator A5053518605 @default.
• W3041621704 creator A5074656322 @default.
• W3041621704 creator A5083147441 @default.
• W3041621704 date "2020-07-08" @default.
• W3041621704 modified "2023-09-25" @default.
• W3041621704 title "Auto-MAP: A DQN Framework for Exploring Distributed Execution Plans for DNN Workloads" @default.
• W3041621704 cites W1522301498 @default.
• W3041621704 cites W1598866093 @default.
• W3041621704 cites W1686810756 @default.
• W3041621704 cites W1757796397 @default.
• W3041621704 cites W2145339207 @default.
• W3041621704 cites W2155968351 @default.
• W3041621704 cites W2173564293 @default.
• W3041621704 cites W2186615578 @default.
• W3041621704 cites W2201581102 @default.
• W3041621704 cites W2724169821 @default.
• W3041621704 cites W2787831171 @default.
• W3041621704 cites W2807147113 @default.
• W3041621704 cites W2884700152 @default.
• W3041621704 cites W2950592884 @default.
• W3041621704 cites W2952926545 @default.
• W3041621704 cites W2963341956 @default.
• W3041621704 cites W2964291307 @default.
• W3041621704 cites W2964308564 @default.
• W3041621704 cites W2969388332 @default.
• W3041621704 cites W2969766737 @default.
• W3041621704 cites W2973727699 @default.
• W3041621704 cites W2977720775 @default.
• W3041621704 cites W2980268386 @default.
• W3041621704 cites W2981852735 @default.
• W3041621704 cites W2983101505 @default.
• W3041621704 cites W2991040477 @default.
• W3041621704 cites W2991615366 @default.
• W3041621704 cites W3010672652 @default.
• W3041621704 cites W3025935268 @default.
• W3041621704 cites W3038581078 @default.
• W3041621704 cites W3040573126 @default.
• W3041621704 doi "https://doi.org/10.48550/arxiv.2007.04069" @default.
• W3041621704 hasPublicationYear "2020" @default.
• W3041621704 type Work @default.
• W3041621704 sameAs 3041621704 @default.
• W3041621704 citedByCount "1" @default.
• W3041621704 countsByYear W30416217042021 @default.
• W3041621704 crossrefType "posted-content" @default.
• W3041621704 hasAuthorship W3041621704A5017996548 @default.
• W3041621704 hasAuthorship W3041621704A5020574416 @default.
• W3041621704 hasAuthorship W3041621704A5025904134 @default.
• W3041621704 hasAuthorship W3041621704A5046708388 @default.
• W3041621704 hasAuthorship W3041621704A5051978977 @default.
• W3041621704 hasAuthorship W3041621704A5052795742 @default.
• W3041621704 hasAuthorship W3041621704A5053518605 @default.
• W3041621704 hasAuthorship W3041621704A5074656322 @default.
• W3041621704 hasAuthorship W3041621704A5083147441 @default.
• W3041621704 hasBestOaLocation W30416217041 @default.
• W3041621704 hasConcept C108010975 @default.
• W3041621704 hasConcept C108583219 @default.
• W3041621704 hasConcept C119857082 @default.
• W3041621704 hasConcept C120314980 @default.
• W3041621704 hasConcept C153083717 @default.
• W3041621704 hasConcept C154945302 @default.
• W3041621704 hasConcept C162324750 @default.
• W3041621704 hasConcept C173608175 @default.
• W3041621704 hasConcept C187736073 @default.
• W3041621704 hasConcept C199360897 @default.
• W3041621704 hasConcept C206729178 @default.
• W3041621704 hasConcept C21547014 @default.
• W3041621704 hasConcept C2780451532 @default.
• W3041621704 hasConcept C2781172179 @default.
• W3041621704 hasConcept C2984842247 @default.
• W3041621704 hasConcept C41008148 @default.
• W3041621704 hasConcept C43521106 @default.
• W3041621704 hasConcept C50644808 @default.
• W3041621704 hasConcept C61483411 @default.
• W3041621704 hasConcept C6557445 @default.
• W3041621704 hasConcept C86803240 @default.
• W3041621704 hasConcept C97541855 @default.
• W3041621704 hasConceptScore W3041621704C108010975 @default.
• W3041621704 hasConceptScore W3041621704C108583219 @default.
• W3041621704 hasConceptScore W3041621704C119857082 @default.
• W3041621704 hasConceptScore W3041621704C120314980 @default.
• W3041621704 hasConceptScore W3041621704C153083717 @default.
• W3041621704 hasConceptScore W3041621704C154945302 @default.
• W3041621704 hasConceptScore W3041621704C162324750 @default.
• W3041621704 hasConceptScore W3041621704C173608175 @default.
• W3041621704 hasConceptScore W3041621704C187736073 @default.
• W3041621704 hasConceptScore W3041621704C199360897 @default.
• W3041621704 hasConceptScore W3041621704C206729178 @default.
• W3041621704 hasConceptScore W3041621704C21547014 @default.
• W3041621704 hasConceptScore W3041621704C2780451532 @default.
• W3041621704 hasConceptScore W3041621704C2781172179 @default.
• W3041621704 hasConceptScore W3041621704C2984842247 @default.
• W3041621704 hasConceptScore W3041621704C41008148 @default.

• next