Leakage-Adjusted Simulatability: Can Models Generate Non-Trivial Explanations of Their Behavior in Natural Language?
Authors:
- Peter HasePHPeter Hase
- Shiyue ZhangSZShiyue Zhang
- Harry XieHXHarry Xie
- Mohit BansalMBMohit Bansal
Open Access
- 1 January 2020
- conference paper
- Published by Association for Computational Linguistics (ACL)
- p. 4351-4367