Mitigating the Inconsistency Between Word Saliency and Model Confidence with Pathological Contrastive Training

32658

Related Entities