ParaBank
Semantic Textual SimilarityEnglish
ParaBank is a semantic textual similarity-focused dataset in English that provides 79.5M references labeled examples distributed in TSV format.
About ParaBank
Dataset contains paraphrases with 79.5 million references and on average 4 paraphrases per reference.
Details
- Task
- Semantic Textual Similarity
- Language
- English
- Format
- TSV
- Rows / instances
- 79.5M references
- Creator
- Hu et al.
- Year
- 2019