Matches in SemOpenAlex for { <https://semopenalex.org/work/W3103092622> ?p ?o ?g. }
- W3103092622 abstract "Natural language inference (NLI) data has proven useful in benchmarking and, especially, as pretraining data for tasks requiring language understanding. However, the crowdsourcing protocol that was used to collect this data has known issues and was not explicitly optimized for either of these purposes, so it is likely far from ideal. We propose four alternative protocols, each aimed at improving either the ease with which annotators can produce sound training examples or the quality and diversity of those examples. Using these alternatives and a fifth baseline protocol, we collect and compare five new 8.5k-example training sets. In evaluations focused on transfer learning applications, our results are solidly negative, with models trained on our baseline dataset yielding good transfer performance to downstream tasks, but none of our four new methods (nor the recent ANLI) showing any improvements over that baseline. In a small silver lining, we observe that all four new protocols, especially those where annotators edit *pre-filled* text boxes, reduce previously observed issues with annotation artifacts." @default.
- W3103092622 created "2020-11-23" @default.
- W3103092622 creator A5019552678 @default.
- W3103092622 creator A5076699933 @default.
- W3103092622 creator A5082761847 @default.
- W3103092622 creator A5089519482 @default.
- W3103092622 date "2020-01-01" @default.
- W3103092622 modified "2023-10-17" @default.
- W3103092622 title "New Protocols and Negative Results for Textual Entailment Data Collection" @default.
- W3103092622 cites W1599016936 @default.
- W3103092622 cites W1840435438 @default.
- W3103092622 cites W2130158090 @default.
- W3103092622 cites W2145755360 @default.
- W3103092622 cites W2219598741 @default.
- W3103092622 cites W2250790822 @default.
- W3103092622 cites W2396767181 @default.
- W3103092622 cites W2592170186 @default.
- W3103092622 cites W2739810148 @default.
- W3103092622 cites W2788496822 @default.
- W3103092622 cites W2804897457 @default.
- W3103092622 cites W2889468083 @default.
- W3103092622 cites W2891177506 @default.
- W3103092622 cites W2898662126 @default.
- W3103092622 cites W2898700502 @default.
- W3103092622 cites W2899771611 @default.
- W3103092622 cites W2945290257 @default.
- W3103092622 cites W2946659172 @default.
- W3103092622 cites W2956105246 @default.
- W3103092622 cites W2962736243 @default.
- W3103092622 cites W2962756502 @default.
- W3103092622 cites W2963090765 @default.
- W3103092622 cites W2963120843 @default.
- W3103092622 cites W2963159473 @default.
- W3103092622 cites W2963159690 @default.
- W3103092622 cites W2963310665 @default.
- W3103092622 cites W2963341956 @default.
- W3103092622 cites W2963351832 @default.
- W3103092622 cites W2963530300 @default.
- W3103092622 cites W2963691697 @default.
- W3103092622 cites W2963846996 @default.
- W3103092622 cites W2963854351 @default.
- W3103092622 cites W2963918774 @default.
- W3103092622 cites W2964352358 @default.
- W3103092622 cites W2965373594 @default.
- W3103092622 cites W2970062726 @default.
- W3103092622 cites W2970597249 @default.
- W3103092622 cites W2990704537 @default.
- W3103092622 cites W2994934025 @default.
- W3103092622 cites W2995643077 @default.
- W3103092622 cites W2996908057 @default.
- W3103092622 cites W3026404337 @default.
- W3103092622 cites W3034255912 @default.
- W3103092622 cites W3034850762 @default.
- W3103092622 cites W3037191812 @default.
- W3103092622 cites W2525127255 @default.
- W3103092622 doi "https://doi.org/10.18653/v1/2020.emnlp-main.658" @default.
- W3103092622 hasPublicationYear "2020" @default.
- W3103092622 type Work @default.
- W3103092622 sameAs 3103092622 @default.
- W3103092622 citedByCount "9" @default.
- W3103092622 countsByYear W31030926222021 @default.
- W3103092622 countsByYear W31030926222022 @default.
- W3103092622 countsByYear W31030926222023 @default.
- W3103092622 crossrefType "proceedings-article" @default.
- W3103092622 hasAuthorship W3103092622A5019552678 @default.
- W3103092622 hasAuthorship W3103092622A5076699933 @default.
- W3103092622 hasAuthorship W3103092622A5082761847 @default.
- W3103092622 hasAuthorship W3103092622A5089519482 @default.
- W3103092622 hasBestOaLocation W31030926221 @default.
- W3103092622 hasConcept C105795698 @default.
- W3103092622 hasConcept C111368507 @default.
- W3103092622 hasConcept C119857082 @default.
- W3103092622 hasConcept C12725497 @default.
- W3103092622 hasConcept C127313418 @default.
- W3103092622 hasConcept C133462117 @default.
- W3103092622 hasConcept C136764020 @default.
- W3103092622 hasConcept C142724271 @default.
- W3103092622 hasConcept C144133560 @default.
- W3103092622 hasConcept C150899416 @default.
- W3103092622 hasConcept C154945302 @default.
- W3103092622 hasConcept C162853370 @default.
- W3103092622 hasConcept C204321447 @default.
- W3103092622 hasConcept C204787440 @default.
- W3103092622 hasConcept C2776214188 @default.
- W3103092622 hasConcept C2776321320 @default.
- W3103092622 hasConcept C2780385302 @default.
- W3103092622 hasConcept C33923547 @default.
- W3103092622 hasConcept C41008148 @default.
- W3103092622 hasConcept C62230096 @default.
- W3103092622 hasConcept C71924100 @default.
- W3103092622 hasConcept C86251818 @default.
- W3103092622 hasConceptScore W3103092622C105795698 @default.
- W3103092622 hasConceptScore W3103092622C111368507 @default.
- W3103092622 hasConceptScore W3103092622C119857082 @default.
- W3103092622 hasConceptScore W3103092622C12725497 @default.
- W3103092622 hasConceptScore W3103092622C127313418 @default.
- W3103092622 hasConceptScore W3103092622C133462117 @default.
- W3103092622 hasConceptScore W3103092622C136764020 @default.
- W3103092622 hasConceptScore W3103092622C142724271 @default.
- W3103092622 hasConceptScore W3103092622C144133560 @default.