WikiText-103
Emerging2papers using it
2021first seen
Wikitext-103 is a benchmark dataset used to evaluate causal language modeling performance, containing a large collection of text extracted from Wikipedia.
Wikitext-103 is a benchmark dataset used to evaluate causal language modeling performance, containing a large collection of text extracted from Wikipedia.