WikiText-103

E1790159 UNEXPLORED

WikiText-103 is a large, high-quality word-level language modeling dataset derived from Wikipedia articles, widely used to evaluate and benchmark neural language models.

Try in SPARQL Jump to: Surface forms Referenced by

All labels observed (1)

Label	Occurrences
WikiText-103 canonical	1

Referenced by (1)

Full triples — surface form annotated when it differs from this entity's canonical label.

Transformer-XL → benchmarkedOn → WikiText-103 ⓘ