Adversarially Constructed Evaluation Sets Are More Challenging, but May Not Be Fair

Adversarially Constructed Evaluation Sets Are More Challenging, but May Not Be Fair Jason Phang author Angelica Chen author William Huang author Samuel R Bowman author 2022-07 text Proceedings of the First Workshop on Dynamic Adversarial Data Collection Max Bartolo editor Hannah Kirk editor Pedro Rodriguez editor Katerina Margatina editor Tristan Thrush editor Robin Jia editor Pontus Stenetorp editor Adina Williams editor Douwe Kiela editor Association for Computational Linguistics Seattle, WA conference publication phang-etal-2022-adversarially 10.18653/v1/2022.dadc-1.8 https://aclanthology.org/2022.dadc-1.8/ 2022-07 62 62