FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4361766487> ?p ?o ?g. }

• W4361766487 endingPage "350" @default.
• W4361766487 startingPage "336" @default.
• W4361766487 abstract "Abstract This work presents a linguistic analysis into why larger Transformer-based pre-trained language models with more parameters and lower perplexity nonetheless yield surprisal estimates that are less predictive of human reading times. First, regression analyses show a strictly monotonic, positive log-linear relationship between perplexity and fit to reading times for the more recently released five GPT-Neo variants and eight OPT variants on two separate datasets, replicating earlier results limited to just GPT-2 (Oh et al., 2022). Subsequently, analysis of residual errors reveals a systematic deviation of the larger variants, such as underpredicting reading times of named entities and making compensatory overpredictions for reading times of function words such as modals and conjunctions. These results suggest that the propensity of larger Transformer-based models to ‘memorize’ sequences during training makes their surprisal estimates diverge from humanlike expectations, which warrants caution in using pre-trained language models to study human language processing." @default.
• W4361766487 created "2023-04-04" @default.
• W4361766487 creator A5066248885 @default.
• W4361766487 creator A5086362177 @default.
• W4361766487 date "2023-01-01" @default.
• W4361766487 modified "2023-09-30" @default.
• W4361766487 title "Why Does Surprisal From Larger Transformer-Based Language Models Provide a Poorer Fit to Human Reading Times?" @default.
• W4361766487 cites W1951724000 @default.
• W4361766487 cites W1977101708 @default.
• W4361766487 cites W1995875735 @default.
• W4361766487 cites W2019173092 @default.
• W4361766487 cites W2054125330 @default.
• W4361766487 cites W2064675550 @default.
• W4361766487 cites W2108010971 @default.
• W4361766487 cites W2130914914 @default.
• W4361766487 cites W2139450036 @default.
• W4361766487 cites W2157331557 @default.
• W4361766487 cites W2164418233 @default.
• W4361766487 cites W2795342569 @default.
• W4361766487 cites W2962784628 @default.
• W4361766487 cites W2962941914 @default.
• W4361766487 cites W2963073938 @default.
• W4361766487 cites W2963341956 @default.
• W4361766487 cites W2997938465 @default.
• W4361766487 cites W3083146265 @default.
• W4361766487 cites W3100748148 @default.
• W4361766487 cites W3160285835 @default.
• W4361766487 cites W3166464178 @default.
• W4361766487 cites W3168591151 @default.
• W4361766487 cites W3171953676 @default.
• W4361766487 cites W3173936365 @default.
• W4361766487 cites W3175306105 @default.
• W4361766487 cites W3210923133 @default.
• W4361766487 cites W3212496002 @default.
• W4361766487 cites W4214909510 @default.
• W4361766487 cites W4221159672 @default.
• W4361766487 cites W4229005866 @default.
• W4361766487 cites W4248358431 @default.
• W4361766487 cites W4285225959 @default.
• W4361766487 cites W4306842234 @default.
• W4361766487 doi "https://doi.org/10.1162/tacl_a_00548" @default.
• W4361766487 hasPublicationYear "2023" @default.
• W4361766487 type Work @default.
• W4361766487 citedByCount "4" @default.
• W4361766487 countsByYear W43617664872023 @default.
• W4361766487 crossrefType "journal-article" @default.
• W4361766487 hasAuthorship W4361766487A5066248885 @default.
• W4361766487 hasAuthorship W4361766487A5086362177 @default.
• W4361766487 hasBestOaLocation W43617664871 @default.
• W4361766487 hasConcept C100279451 @default.
• W4361766487 hasConcept C105795698 @default.
• W4361766487 hasConcept C121332964 @default.
• W4361766487 hasConcept C134306372 @default.
• W4361766487 hasConcept C137293760 @default.
• W4361766487 hasConcept C138885662 @default.
• W4361766487 hasConcept C154945302 @default.
• W4361766487 hasConcept C165801399 @default.
• W4361766487 hasConcept C204321447 @default.
• W4361766487 hasConcept C2781162219 @default.
• W4361766487 hasConcept C33923547 @default.
• W4361766487 hasConcept C41008148 @default.
• W4361766487 hasConcept C41895202 @default.
• W4361766487 hasConcept C554936623 @default.
• W4361766487 hasConcept C62520636 @default.
• W4361766487 hasConcept C66322947 @default.
• W4361766487 hasConcept C72169020 @default.
• W4361766487 hasConceptScore W4361766487C100279451 @default.
• W4361766487 hasConceptScore W4361766487C105795698 @default.
• W4361766487 hasConceptScore W4361766487C121332964 @default.
• W4361766487 hasConceptScore W4361766487C134306372 @default.
• W4361766487 hasConceptScore W4361766487C137293760 @default.
• W4361766487 hasConceptScore W4361766487C138885662 @default.
• W4361766487 hasConceptScore W4361766487C154945302 @default.
• W4361766487 hasConceptScore W4361766487C165801399 @default.
• W4361766487 hasConceptScore W4361766487C204321447 @default.
• W4361766487 hasConceptScore W4361766487C2781162219 @default.
• W4361766487 hasConceptScore W4361766487C33923547 @default.
• W4361766487 hasConceptScore W4361766487C41008148 @default.
• W4361766487 hasConceptScore W4361766487C41895202 @default.
• W4361766487 hasConceptScore W4361766487C554936623 @default.
• W4361766487 hasConceptScore W4361766487C62520636 @default.
• W4361766487 hasConceptScore W4361766487C66322947 @default.
• W4361766487 hasConceptScore W4361766487C72169020 @default.
• W4361766487 hasLocation W43617664871 @default.
• W4361766487 hasLocation W43617664872 @default.
• W4361766487 hasOpenAccess W4361766487 @default.
• W4361766487 hasPrimaryLocation W43617664871 @default.
• W4361766487 hasRelatedWork W1989705153 @default.
• W4361766487 hasRelatedWork W2496228846 @default.
• W4361766487 hasRelatedWork W2896411932 @default.
• W4361766487 hasRelatedWork W2936497627 @default.
• W4361766487 hasRelatedWork W3013624417 @default.
• W4361766487 hasRelatedWork W3049463507 @default.
• W4361766487 hasRelatedWork W3100913109 @default.
• W4361766487 hasRelatedWork W4287826556 @default.
• W4361766487 hasRelatedWork W4288365749 @default.
• W4361766487 hasRelatedWork W59929963 @default.
• W4361766487 hasVolume "11" @default.

• next