Leakage-Adjusted Simulatability: Can Models Generate Non-Trivial Explanations of Their Behavior in Natural Language?

Leakage-Adjusted Simulatability: Can Models Generate Non-Trivial Explanations of Their Behavior in Natural Language? Peter Hase author Shiyue Zhang author Harry Xie author Mohit Bansal author 2020-11 text Findings of the Association for Computational Linguistics: EMNLP 2020 Trevor Cohn editor Yulan He editor Yang Liu editor Association for Computational Linguistics Online conference publication hase-etal-2020-leakage 10.18653/v1/2020.findings-emnlp.390 https://aclanthology.org/2020.findings-emnlp.390/ 2020-11 4351 4367