Agree, Disagree, Explain: Decomposing Human Label Variation in NLI through the Lens of Explanations

Open in new window