Matches in SemOpenAlex for { <https://semopenalex.org/work/W3138301265> ?p ?o ?g. }
- W3138301265 endingPage "242" @default.
- W3138301265 startingPage "226" @default.
- W3138301265 abstract "Abstract For natural language processing systems, two kinds of evidence support the use of text representations from neural language models “pretrained” on large unannotated corpora: performance on application-inspired benchmarks (Peters et al., 2018, inter alia), and the emergence of syntactic abstractions in those representations (Tenney et al., 2019, inter alia). On the other hand, the lack of grounded supervision calls into question how well these representations can ever capture meaning (Bender and Koller, 2020). We apply novel probes to recent language models— specifically focusing on predicate-argument structure as operationalized by semantic dependencies (Ivanova et al., 2012)—and find that, unlike syntax, semantics is not brought to the surface by today’s pretrained models. We then use convolutional graph encoders to explicitly incorporate semantic parses into task-specific finetuning, yielding benefits to natural language understanding (NLU) tasks in the GLUE benchmark. This approach demonstrates the potential for general-purpose (rather than task-specific) linguistic supervision, above and beyond conventional pretraining and finetuning. Several diagnostics help to localize the benefits of our approach.1" @default.
- W3138301265 created "2021-03-29" @default.
- W3138301265 creator A5036524439 @default.
- W3138301265 creator A5079854850 @default.
- W3138301265 creator A5088517824 @default.
- W3138301265 date "2021-01-01" @default.
- W3138301265 modified "2023-10-17" @default.
- W3138301265 title "Infusing Finetuning with Semantic Dependencies" @default.
- W3138301265 cites W1517853909 @default.
- W3138301265 cites W1970849810 @default.
- W3138301265 cites W2052449326 @default.
- W3138301265 cites W2059185913 @default.
- W3138301265 cites W2095589793 @default.
- W3138301265 cites W2122922578 @default.
- W3138301265 cites W2158847908 @default.
- W3138301265 cites W2413794162 @default.
- W3138301265 cites W2563574619 @default.
- W3138301265 cites W2594047108 @default.
- W3138301265 cites W2600702321 @default.
- W3138301265 cites W2604314403 @default.
- W3138301265 cites W2608787653 @default.
- W3138301265 cites W2799072540 @default.
- W3138301265 cites W2826721128 @default.
- W3138301265 cites W2888882903 @default.
- W3138301265 cites W2891602716 @default.
- W3138301265 cites W2906152891 @default.
- W3138301265 cites W2923014074 @default.
- W3138301265 cites W2951286828 @default.
- W3138301265 cites W2962731964 @default.
- W3138301265 cites W2962739339 @default.
- W3138301265 cites W2962784628 @default.
- W3138301265 cites W2962788148 @default.
- W3138301265 cites W2963653811 @default.
- W3138301265 cites W2963748441 @default.
- W3138301265 cites W2963846996 @default.
- W3138301265 cites W2963851958 @default.
- W3138301265 cites W2963876447 @default.
- W3138301265 cites W2964167098 @default.
- W3138301265 cites W2964263959 @default.
- W3138301265 cites W2964303116 @default.
- W3138301265 cites W2970120757 @default.
- W3138301265 cites W2972324944 @default.
- W3138301265 cites W2978670439 @default.
- W3138301265 cites W2982904446 @default.
- W3138301265 cites W2983845924 @default.
- W3138301265 cites W2984969137 @default.
- W3138301265 cites W2986009340 @default.
- W3138301265 cites W2986266667 @default.
- W3138301265 cites W2997244573 @default.
- W3138301265 cites W2998230451 @default.
- W3138301265 cites W2998665041 @default.
- W3138301265 cites W3034723486 @default.
- W3138301265 cites W3035204084 @default.
- W3138301265 cites W3104033643 @default.
- W3138301265 cites W4241738801 @default.
- W3138301265 cites W4246185962 @default.
- W3138301265 cites W4253067820 @default.
- W3138301265 doi "https://doi.org/10.1162/tacl_a_00363" @default.
- W3138301265 hasPublicationYear "2021" @default.
- W3138301265 type Work @default.
- W3138301265 sameAs 3138301265 @default.
- W3138301265 citedByCount "19" @default.
- W3138301265 countsByYear W31383012652021 @default.
- W3138301265 countsByYear W31383012652022 @default.
- W3138301265 countsByYear W31383012652023 @default.
- W3138301265 crossrefType "journal-article" @default.
- W3138301265 hasAuthorship W3138301265A5036524439 @default.
- W3138301265 hasAuthorship W3138301265A5079854850 @default.
- W3138301265 hasAuthorship W3138301265A5088517824 @default.
- W3138301265 hasBestOaLocation W31383012651 @default.
- W3138301265 hasConcept C111472728 @default.
- W3138301265 hasConcept C138885662 @default.
- W3138301265 hasConcept C140146324 @default.
- W3138301265 hasConcept C154945302 @default.
- W3138301265 hasConcept C162324750 @default.
- W3138301265 hasConcept C184337299 @default.
- W3138301265 hasConcept C187736073 @default.
- W3138301265 hasConcept C195324797 @default.
- W3138301265 hasConcept C199360897 @default.
- W3138301265 hasConcept C204321447 @default.
- W3138301265 hasConcept C2777530160 @default.
- W3138301265 hasConcept C2779439875 @default.
- W3138301265 hasConcept C2780451532 @default.
- W3138301265 hasConcept C2983448237 @default.
- W3138301265 hasConcept C41008148 @default.
- W3138301265 hasConcept C60048249 @default.
- W3138301265 hasConcept C67277372 @default.
- W3138301265 hasConcept C9354725 @default.
- W3138301265 hasConceptScore W3138301265C111472728 @default.
- W3138301265 hasConceptScore W3138301265C138885662 @default.
- W3138301265 hasConceptScore W3138301265C140146324 @default.
- W3138301265 hasConceptScore W3138301265C154945302 @default.
- W3138301265 hasConceptScore W3138301265C162324750 @default.
- W3138301265 hasConceptScore W3138301265C184337299 @default.
- W3138301265 hasConceptScore W3138301265C187736073 @default.
- W3138301265 hasConceptScore W3138301265C195324797 @default.
- W3138301265 hasConceptScore W3138301265C199360897 @default.
- W3138301265 hasConceptScore W3138301265C204321447 @default.