Return to Preprint Details What Can a Model Learn Without the Premise? Explaining Hypothesis-Only Shortcuts in Natural Language Inference Download Download PDF