Update on GitHub

				

				from nlp import load_dataset

				dataset = load_dataset("wiki40b")
				
			

Description

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.

Citation

Models trained or fine-tuned on wiki40b

None yet. Start fine-tuning now =)