GRICE: A Grammar-based Dataset for Recovering Implicature and Conversational rEasoning

34000

Related Entities