On the Efficacy of Adversarial Data Collection for Question Answering: Results from a Large-Scale Randomized StudyPublished in ACL 2021, 2021 dataset Read the paperShare on X (Twitter) Facebook Google+ LinkedIn Previous Next