FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3091784310> ?p ?o ?g. }

• W3091784310 abstract "Quite surprisingly, exact maximum a posteriori (MAP) decoding of neural language generators frequently leads to low-quality results. Rather, most state-of-the-art results on language generation tasks are attained using beam search despite its overwhelmingly high search error rate. This implies that the MAP objective alone does not express the properties we desire in text, which merits the question: if beam search is the answer, what was the question? We frame beam search as the exact solution to a different decoding objective in order to gain insights into why high probability under a model alone may not indicate adequacy. We find that beam search enforces uniform information density in text, a property motivated by cognitive science. We suggest a set of decoding objectives that explicitly enforce this property and find that exact decoding with these objectives alleviates the problems encountered when decoding poorly calibrated language generation models. Additionally, we analyze the text produced using various decoding strategies and see that, in our neural machine translation experiments, the extent to which this property is adhered to strongly correlates with BLEU." @default.
• W3091784310 created "2020-10-15" @default.
• W3091784310 creator A5029033876 @default.
• W3091784310 creator A5052346902 @default.
• W3091784310 creator A5061951606 @default.
• W3091784310 date "2020-10-06" @default.
• W3091784310 modified "2023-09-23" @default.
• W3091784310 title "If beam search is the answer, what was the question?" @default.
• W3091784310 cites W1498238796 @default.
• W3091784310 cites W1517947178 @default.
• W3091784310 cites W1535183358 @default.
• W3091784310 cites W1591706642 @default.
• W3091784310 cites W2009727972 @default.
• W3091784310 cites W2054125330 @default.
• W3091784310 cites W2101105183 @default.
• W3091784310 cites W2129955048 @default.
• W3091784310 cites W2130942839 @default.
• W3091784310 cites W2153653739 @default.
• W3091784310 cites W2157365695 @default.
• W3091784310 cites W2169528473 @default.
• W3091784310 cites W2184135559 @default.
• W3091784310 cites W2252272516 @default.
• W3091784310 cites W2340006107 @default.
• W3091784310 cites W2418993857 @default.
• W3091784310 cites W2463955103 @default.
• W3091784310 cites W2566564022 @default.
• W3091784310 cites W2757222607 @default.
• W3091784310 cites W2788277448 @default.
• W3091784310 cites W2889326796 @default.
• W3091784310 cites W2918914336 @default.
• W3091784310 cites W2933138175 @default.
• W3091784310 cites W2954460650 @default.
• W3091784310 cites W2962902089 @default.
• W3091784310 cites W2963096510 @default.
• W3091784310 cites W2963206148 @default.
• W3091784310 cites W2963260202 @default.
• W3091784310 cites W2963382396 @default.
• W3091784310 cites W2963403868 @default.
• W3091784310 cites W2963463964 @default.
• W3091784310 cites W2963506925 @default.
• W3091784310 cites W2963532001 @default.
• W3091784310 cites W2963546833 @default.
• W3091784310 cites W2963665552 @default.
• W3091784310 cites W2963706817 @default.
• W3091784310 cites W2963807318 @default.
• W3091784310 cites W2964103964 @default.
• W3091784310 cites W2964106094 @default.
• W3091784310 cites W2964199361 @default.
• W3091784310 cites W2964212410 @default.
• W3091784310 cites W2964265128 @default.
• W3091784310 cites W2964308564 @default.
• W3091784310 cites W2970295111 @default.
• W3091784310 cites W2970597249 @default.
• W3091784310 cites W2970692082 @default.
• W3091784310 cites W2996287690 @default.
• W3091784310 cites W3088227906 @default.
• W3091784310 cites W3115021520 @default.
• W3091784310 cites W648786980 @default.
• W3091784310 hasPublicationYear "2020" @default.
• W3091784310 type Work @default.
• W3091784310 sameAs 3091784310 @default.
• W3091784310 citedByCount "18" @default.
• W3091784310 countsByYear W30917843102020 @default.
• W3091784310 countsByYear W30917843102021 @default.
• W3091784310 countsByYear W30917843102022 @default.
• W3091784310 crossrefType "posted-content" @default.
• W3091784310 hasAuthorship W3091784310A5029033876 @default.
• W3091784310 hasAuthorship W3091784310A5052346902 @default.
• W3091784310 hasAuthorship W3091784310A5061951606 @default.
• W3091784310 hasConcept C111472728 @default.
• W3091784310 hasConcept C11413529 @default.
• W3091784310 hasConcept C125583679 @default.
• W3091784310 hasConcept C137293760 @default.
• W3091784310 hasConcept C138885662 @default.
• W3091784310 hasConcept C154945302 @default.
• W3091784310 hasConcept C177264268 @default.
• W3091784310 hasConcept C189950617 @default.
• W3091784310 hasConcept C19889080 @default.
• W3091784310 hasConcept C199360897 @default.
• W3091784310 hasConcept C203005215 @default.
• W3091784310 hasConcept C41008148 @default.
• W3091784310 hasConcept C48103436 @default.
• W3091784310 hasConcept C57273362 @default.
• W3091784310 hasConcept C75553542 @default.
• W3091784310 hasConcept C80444323 @default.
• W3091784310 hasConceptScore W3091784310C111472728 @default.
• W3091784310 hasConceptScore W3091784310C11413529 @default.
• W3091784310 hasConceptScore W3091784310C125583679 @default.
• W3091784310 hasConceptScore W3091784310C137293760 @default.
• W3091784310 hasConceptScore W3091784310C138885662 @default.
• W3091784310 hasConceptScore W3091784310C154945302 @default.
• W3091784310 hasConceptScore W3091784310C177264268 @default.
• W3091784310 hasConceptScore W3091784310C189950617 @default.
• W3091784310 hasConceptScore W3091784310C19889080 @default.
• W3091784310 hasConceptScore W3091784310C199360897 @default.
• W3091784310 hasConceptScore W3091784310C203005215 @default.
• W3091784310 hasConceptScore W3091784310C41008148 @default.
• W3091784310 hasConceptScore W3091784310C48103436 @default.
• W3091784310 hasConceptScore W3091784310C57273362 @default.
• W3091784310 hasConceptScore W3091784310C75553542 @default.

• next