Matches in SemOpenAlex for { <https://semopenalex.org/work/W4306311904> ?p ?o ?g. }
Showing items 1 to 75 of
75
with 100 items per page.
- W4306311904 abstract "Most current popular subword tokenizers are trained based on word frequency statistics over a corpus, without considering information about co-occurrence or context. Nevertheless, the resulting vocabularies are used in language models' highly contextualized settings. We present SaGe, a tokenizer that tailors subwords for their downstream use by baking in the contextualized signal at the vocabulary creation phase. We show that SaGe does a better job than current widespread tokenizers in keeping token contexts cohesive, while not incurring a large price in terms of encoding efficiency or domain robustness. SaGe improves performance on English GLUE classification tasks as well as on NER, and on Inference and NER in Turkish, demonstrating its robustness to language properties such as morphological exponence and agglutination." @default.
- W4306311904 created "2022-10-15" @default.
- W4306311904 creator A5008477284 @default.
- W4306311904 creator A5087386943 @default.
- W4306311904 date "2022-10-13" @default.
- W4306311904 modified "2023-09-27" @default.
- W4306311904 title "Incorporating Context into Subword Vocabularies" @default.
- W4306311904 doi "https://doi.org/10.48550/arxiv.2210.07095" @default.
- W4306311904 hasPublicationYear "2022" @default.
- W4306311904 type Work @default.
- W4306311904 citedByCount "0" @default.
- W4306311904 crossrefType "posted-content" @default.
- W4306311904 hasAuthorship W4306311904A5008477284 @default.
- W4306311904 hasAuthorship W4306311904A5087386943 @default.
- W4306311904 hasBestOaLocation W43063119041 @default.
- W4306311904 hasConcept C104317684 @default.
- W4306311904 hasConcept C127413603 @default.
- W4306311904 hasConcept C138885662 @default.
- W4306311904 hasConcept C151730666 @default.
- W4306311904 hasConcept C154945302 @default.
- W4306311904 hasConcept C185592680 @default.
- W4306311904 hasConcept C201995342 @default.
- W4306311904 hasConcept C204321447 @default.
- W4306311904 hasConcept C2776214188 @default.
- W4306311904 hasConcept C2777601683 @default.
- W4306311904 hasConcept C2779135771 @default.
- W4306311904 hasConcept C2779343474 @default.
- W4306311904 hasConcept C2780451532 @default.
- W4306311904 hasConcept C2781121862 @default.
- W4306311904 hasConcept C28490314 @default.
- W4306311904 hasConcept C38652104 @default.
- W4306311904 hasConcept C41008148 @default.
- W4306311904 hasConcept C41895202 @default.
- W4306311904 hasConcept C48145219 @default.
- W4306311904 hasConcept C55493867 @default.
- W4306311904 hasConcept C63479239 @default.
- W4306311904 hasConcept C86803240 @default.
- W4306311904 hasConceptScore W4306311904C104317684 @default.
- W4306311904 hasConceptScore W4306311904C127413603 @default.
- W4306311904 hasConceptScore W4306311904C138885662 @default.
- W4306311904 hasConceptScore W4306311904C151730666 @default.
- W4306311904 hasConceptScore W4306311904C154945302 @default.
- W4306311904 hasConceptScore W4306311904C185592680 @default.
- W4306311904 hasConceptScore W4306311904C201995342 @default.
- W4306311904 hasConceptScore W4306311904C204321447 @default.
- W4306311904 hasConceptScore W4306311904C2776214188 @default.
- W4306311904 hasConceptScore W4306311904C2777601683 @default.
- W4306311904 hasConceptScore W4306311904C2779135771 @default.
- W4306311904 hasConceptScore W4306311904C2779343474 @default.
- W4306311904 hasConceptScore W4306311904C2780451532 @default.
- W4306311904 hasConceptScore W4306311904C2781121862 @default.
- W4306311904 hasConceptScore W4306311904C28490314 @default.
- W4306311904 hasConceptScore W4306311904C38652104 @default.
- W4306311904 hasConceptScore W4306311904C41008148 @default.
- W4306311904 hasConceptScore W4306311904C41895202 @default.
- W4306311904 hasConceptScore W4306311904C48145219 @default.
- W4306311904 hasConceptScore W4306311904C55493867 @default.
- W4306311904 hasConceptScore W4306311904C63479239 @default.
- W4306311904 hasConceptScore W4306311904C86803240 @default.
- W4306311904 hasLocation W43063119041 @default.
- W4306311904 hasOpenAccess W4306311904 @default.
- W4306311904 hasPrimaryLocation W43063119041 @default.
- W4306311904 hasRelatedWork W147166030 @default.
- W4306311904 hasRelatedWork W1728883788 @default.
- W4306311904 hasRelatedWork W1977671488 @default.
- W4306311904 hasRelatedWork W2100172690 @default.
- W4306311904 hasRelatedWork W2112039799 @default.
- W4306311904 hasRelatedWork W2428638475 @default.
- W4306311904 hasRelatedWork W2806128650 @default.
- W4306311904 hasRelatedWork W2810280135 @default.
- W4306311904 hasRelatedWork W3128216712 @default.
- W4306311904 hasRelatedWork W61303690 @default.
- W4306311904 isParatext "false" @default.
- W4306311904 isRetracted "false" @default.
- W4306311904 workType "article" @default.