A large-scale English paraphrase dataset that surpasses prior work in both quantity and quality.
Source: ParaBank: Monolingual Bitext Generation and Sentential Paraphrasing via Lexically-constrained Neural Machine Translation