Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385570070> ?p ?o ?g. }
Showing items 1 to 47 of
47
with 100 items per page.
- W4385570070 abstract "In open-ended natural-language generation, existing text decoding methods typically struggle to produce text which is both diverse and high-quality. Greedy and beam search are known to suffer from text degeneration and linguistic diversity issues, while temperature, top-k, and nucleus sampling yield diverse but often lower-quality outputs. In this work, we build upon Minimum Bayes Risk Decoding (MBRD), a family of decoding methods based on Bayesian risk minimization, to address this diversity-quality trade-off. Inspired by the principle of the wisdom of the crowd, MBRD seeks to select a candidate from a pool of candidates that has the least expected risk under a generative model according to a given utility function. The crowd of candidates serves as an approximation for the distribution over human-generated references. We show that MBRD generalizes numerous decoding methods, including majority voting, and can be used as a drop-in replacement for existing sampling methods. Across a wide range of tasks—such as summarization, data-to-text, translation, and textual style transfer—MBRD yields 3-7 ROUGE and BLEU point improvements, including state-of-the-art results on WebNLG and WMT’16." @default.
- W4385570070 created "2023-08-05" @default.
- W4385570070 creator A5009980324 @default.
- W4385570070 creator A5079016134 @default.
- W4385570070 creator A5087088138 @default.
- W4385570070 date "2023-01-01" @default.
- W4385570070 modified "2023-09-24" @default.
- W4385570070 title "Follow the Wisdom of the Crowd: Effective Text Generation via Minimum Bayes Risk Decoding" @default.
- W4385570070 doi "https://doi.org/10.18653/v1/2023.findings-acl.262" @default.
- W4385570070 hasPublicationYear "2023" @default.
- W4385570070 type Work @default.
- W4385570070 citedByCount "0" @default.
- W4385570070 crossrefType "proceedings-article" @default.
- W4385570070 hasAuthorship W4385570070A5009980324 @default.
- W4385570070 hasAuthorship W4385570070A5079016134 @default.
- W4385570070 hasAuthorship W4385570070A5087088138 @default.
- W4385570070 hasBestOaLocation W43855700701 @default.
- W4385570070 hasConcept C11413529 @default.
- W4385570070 hasConcept C119857082 @default.
- W4385570070 hasConcept C154945302 @default.
- W4385570070 hasConcept C170858558 @default.
- W4385570070 hasConcept C204321447 @default.
- W4385570070 hasConcept C41008148 @default.
- W4385570070 hasConcept C57273362 @default.
- W4385570070 hasConceptScore W4385570070C11413529 @default.
- W4385570070 hasConceptScore W4385570070C119857082 @default.
- W4385570070 hasConceptScore W4385570070C154945302 @default.
- W4385570070 hasConceptScore W4385570070C170858558 @default.
- W4385570070 hasConceptScore W4385570070C204321447 @default.
- W4385570070 hasConceptScore W4385570070C41008148 @default.
- W4385570070 hasConceptScore W4385570070C57273362 @default.
- W4385570070 hasLocation W43855700701 @default.
- W4385570070 hasOpenAccess W4385570070 @default.
- W4385570070 hasPrimaryLocation W43855700701 @default.
- W4385570070 hasRelatedWork W2136308941 @default.
- W4385570070 hasRelatedWork W2285613413 @default.
- W4385570070 hasRelatedWork W2308250245 @default.
- W4385570070 hasRelatedWork W2351187795 @default.
- W4385570070 hasRelatedWork W2380641910 @default.
- W4385570070 hasRelatedWork W2389846579 @default.
- W4385570070 hasRelatedWork W2443684444 @default.
- W4385570070 hasRelatedWork W2561691764 @default.
- W4385570070 hasRelatedWork W2589098947 @default.
- W4385570070 hasRelatedWork W2887112617 @default.
- W4385570070 isParatext "false" @default.
- W4385570070 isRetracted "false" @default.
- W4385570070 workType "article" @default.