Clean-up text for 40+ Wikipedia languages editions of pages
correspond to entities. The datasets have train/dev/test splits per language.
The dataset is cleaned up by page filtering to remove disambiguation pages,
redirect pages, deleted pages, and non-entity pages. Each example contains the
wikidata id of the entity, and the full Wikipedia article after page processing
that removes non-content sections and structured objects.… See the full description on the dataset page: https://huggingface.co/datasets/google/wiki40b.
