Paralex learns from a collection of 18 million question-paraphrase pairs scraped from WikiAnswers.
Variants: Paralex
This dataset is used in 1 benchmark:
| Task | Model | Paper | Date |
|---|---|---|---|
| Paraphrase Generation | HRQ-VAE | Hierarchical Sketch Induction for Paraphrase … | 2022-03-07 |
| Paraphrase Generation | Separator | Factorising Meaning and Form for … | 2021-05-31 |
Recent papers with results on this dataset: