Show simple item record

dc.contributor.authorBelinkov, Yonatan
dc.contributor.authorPoliak, Adam
dc.contributor.authorShieber, Stuart
dc.contributor.authorVan Durme, Benjamin
dc.contributor.authorRush, Alexander Sasha
dc.date.accessioned2019-07-03T14:31:33Z
dc.date.issued2019-07
dc.identifier.citationBelinkov, Yonatan, Adam Poliak, Stuart M. Shieber, Benjamin Van Durme, and Alexander Rush. Don’t take the premise for granted: Mitigating artifacts in natural language inference. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (ACL 2019), Florence, Italy, 28 July–2 August 2019. Assocation for Computational Linguistics.en_US
dc.identifier.urihttp://nrs.harvard.edu/urn-3:HUL.InstRepos:40827357*
dc.description.abstractNatural Language Inference (NLI) datasets often contain hypothesis-only biases—artifacts that allow models to achieve non-trivial performance without learning whether a premise entails a hypothesis. We propose two probabilistic methods to build models that are more robust to such biases and better transfer across datasets. In contrast to standard approaches to NLI, our methods predict the probability of a premise given a hypothesis and NLI label, discouraging models from ignoring the premise. We evaluate our methods on synthetic and existing NLI datasets by training on datasets containing biases and testing on datasets containing no (or different) hypothesis-only biases. Our results indicate that these methods can make NLI models more robust to dataset-specific artifacts, transferring better than a baseline architecture in 9 out of 12 NLI datasets. Additionally, we provide an extensive analysis of the interplay of our methods with known biases in NLI datasets, as well as the effects of encouraging models to ignore biases and fine-tuning on target datasets.en_US
dc.language.isoen_USen_US
dc.publisherAssociation of Computational Linguisticsen_US
dc.relationProceedings of the 57th Annual Meeting of the Association for Computational Linguistics (ACL 2019)en_US
dash.licenseOAP
dc.titleDon't Take the Premise for Granted: Mitigating Artifacts in Natural Language Inferenceen_US
dc.typeConference Paperen_US
dc.description.versionAccepted Manuscripten_US
dc.relation.journalProceedings of the Association of Computational Linguisticsen_US
dash.depositing.authorShieber, Stuart
dc.date.available2019-07-03T14:31:33Z
dash.affiliation.otherHarvard John A. Paulson School of Engineering and Applied Sciencesen_US
dash.contributor.affiliatedBelinkov, Yonatan
dash.contributor.affiliatedRush, Alexander Sasha
dash.contributor.affiliatedShieber, Stuart


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record