Partial-input baselines show that NLI models can ignore context, but they don’t.

Neha Srikanth, Rachel Rudinger


Abstract
When strong partial-input baselines reveal artifacts in crowdsourced NLI datasets, the performance of full-input models trained on such datasets is often dismissed as reliance on spurious correlations. We investigate whether state-of-the-art NLI models are capable of overriding default inferences made by a partial-input baseline. We introduce an evaluation set of 600 examples consisting of perturbed premises to examine a RoBERTa model’s sensitivity to edited contexts. Our results indicate that NLI models are still capable of learning to condition on context—a necessary component of inferential reasoning—despite being trained on artifact-ridden datasets.
Anthology ID:
2022.naacl-main.350
Volume:
Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies
Month:
July
Year:
2022
Address:
Seattle, United States
Venue:
NAACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
4753–4763
Language:
URL:
https://aclanthology.org/2022.naacl-main.350
DOI:
10.18653/v1/2022.naacl-main.350
Bibkey:
Cite (ACL):
Neha Srikanth and Rachel Rudinger. 2022. Partial-input baselines show that NLI models can ignore context, but they don’t.. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4753–4763, Seattle, United States. Association for Computational Linguistics.
Cite (Informal):
Partial-input baselines show that NLI models can ignore context, but they don’t. (Srikanth & Rudinger, NAACL 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.naacl-main.350.pdf
Code
 nehasrikn/context-editing
Data
SNLI