Matches in SemOpenAlex for { <https://semopenalex.org/work/W3172181153> ?p ?o ?g. }
Showing items 1 to 91 of
91
with 100 items per page.
- W3172181153 abstract "As a language model that integrates traditional symbolic operations and flexible neural representations, recurrent neural network grammars (RNNGs) have attracted great attention from both scientific and engineering perspectives. However, RNNGs are known to be harder to scale due to the difficulty of batched training. In this paper, we propose effective batching for RNNGs, where every operation is computed in parallel with tensors across multiple sentences. Our PyTorch implementation effectively employs a GPU and achieves x6 speedup compared to the existing C++ DyNet implementation with model-independent auto-batching. Moreover, our batched RNNG also accelerates inference and achieves x20-150 speedup for beam search depending on beam sizes. Finally, we evaluate syntactic generalization performance of the scaled RNNG against the LSTM baseline, based on the large training data of 100M tokens from English Wikipedia and the broad-coverage targeted syntactic evaluation benchmark. Our RNNG implementation is available at https://github.com/aistairc/rnng-pytorch/." @default.
- W3172181153 created "2021-06-22" @default.
- W3172181153 creator A5025796706 @default.
- W3172181153 creator A5076043621 @default.
- W3172181153 date "2021-05-31" @default.
- W3172181153 modified "2023-09-25" @default.
- W3172181153 title "Effective Batching for Recurrent Neural Network Grammars" @default.
- W3172181153 cites W2064675550 @default.
- W3172181153 cites W2139621418 @default.
- W3172181153 cites W2554915555 @default.
- W3172181153 cites W2563157576 @default.
- W3172181153 cites W2594047108 @default.
- W3172181153 cites W2798727047 @default.
- W3172181153 cites W2911435132 @default.
- W3172181153 cites W2949952998 @default.
- W3172181153 cites W2962733492 @default.
- W3172181153 cites W2962784628 @default.
- W3172181153 cites W2963073938 @default.
- W3172181153 cites W2963250244 @default.
- W3172181153 cites W2963341956 @default.
- W3172181153 cites W2963643701 @default.
- W3172181153 cites W2964121744 @default.
- W3172181153 cites W3035267217 @default.
- W3172181153 cites W3037115370 @default.
- W3172181153 cites W3092973380 @default.
- W3172181153 cites W3117738520 @default.
- W3172181153 cites W3168987555 @default.
- W3172181153 doi "https://doi.org/10.48550/arxiv.2105.14822" @default.
- W3172181153 hasPublicationYear "2021" @default.
- W3172181153 type Work @default.
- W3172181153 sameAs 3172181153 @default.
- W3172181153 citedByCount "0" @default.
- W3172181153 crossrefType "posted-content" @default.
- W3172181153 hasAuthorship W3172181153A5025796706 @default.
- W3172181153 hasAuthorship W3172181153A5076043621 @default.
- W3172181153 hasBestOaLocation W31721811531 @default.
- W3172181153 hasConcept C125583679 @default.
- W3172181153 hasConcept C13280743 @default.
- W3172181153 hasConcept C134306372 @default.
- W3172181153 hasConcept C137293760 @default.
- W3172181153 hasConcept C147168706 @default.
- W3172181153 hasConcept C154945302 @default.
- W3172181153 hasConcept C173608175 @default.
- W3172181153 hasConcept C177148314 @default.
- W3172181153 hasConcept C185798385 @default.
- W3172181153 hasConcept C19889080 @default.
- W3172181153 hasConcept C199360897 @default.
- W3172181153 hasConcept C204321447 @default.
- W3172181153 hasConcept C205649164 @default.
- W3172181153 hasConcept C2776214188 @default.
- W3172181153 hasConcept C33923547 @default.
- W3172181153 hasConcept C41008148 @default.
- W3172181153 hasConcept C50644808 @default.
- W3172181153 hasConcept C53893814 @default.
- W3172181153 hasConcept C68339613 @default.
- W3172181153 hasConceptScore W3172181153C125583679 @default.
- W3172181153 hasConceptScore W3172181153C13280743 @default.
- W3172181153 hasConceptScore W3172181153C134306372 @default.
- W3172181153 hasConceptScore W3172181153C137293760 @default.
- W3172181153 hasConceptScore W3172181153C147168706 @default.
- W3172181153 hasConceptScore W3172181153C154945302 @default.
- W3172181153 hasConceptScore W3172181153C173608175 @default.
- W3172181153 hasConceptScore W3172181153C177148314 @default.
- W3172181153 hasConceptScore W3172181153C185798385 @default.
- W3172181153 hasConceptScore W3172181153C19889080 @default.
- W3172181153 hasConceptScore W3172181153C199360897 @default.
- W3172181153 hasConceptScore W3172181153C204321447 @default.
- W3172181153 hasConceptScore W3172181153C205649164 @default.
- W3172181153 hasConceptScore W3172181153C2776214188 @default.
- W3172181153 hasConceptScore W3172181153C33923547 @default.
- W3172181153 hasConceptScore W3172181153C41008148 @default.
- W3172181153 hasConceptScore W3172181153C50644808 @default.
- W3172181153 hasConceptScore W3172181153C53893814 @default.
- W3172181153 hasConceptScore W3172181153C68339613 @default.
- W3172181153 hasLocation W31721811531 @default.
- W3172181153 hasOpenAccess W3172181153 @default.
- W3172181153 hasPrimaryLocation W31721811531 @default.
- W3172181153 hasRelatedWork W1569841287 @default.
- W3172181153 hasRelatedWork W1697423248 @default.
- W3172181153 hasRelatedWork W1985007624 @default.
- W3172181153 hasRelatedWork W2896317853 @default.
- W3172181153 hasRelatedWork W2987019345 @default.
- W3172181153 hasRelatedWork W3008756425 @default.
- W3172181153 hasRelatedWork W3107474891 @default.
- W3172181153 hasRelatedWork W3172181153 @default.
- W3172181153 hasRelatedWork W3176899693 @default.
- W3172181153 hasRelatedWork W4287864142 @default.
- W3172181153 isParatext "false" @default.
- W3172181153 isRetracted "false" @default.
- W3172181153 magId "3172181153" @default.
- W3172181153 workType "article" @default.