The dataset viewer is taking too long to fetch the data. Try to refresh this page.
Server-side error
Error code: ClientConnectionError
Dataset Card for "wiki-span"
This dataset is constructed by sampling 25%-50% of each wikipedia record twice, as positive pairs. It can be used to train unsupervised sentence representation models.