XSum

XSum is an English single-document summarization benchmark of 226,711 BBC articles paired with a one-sentence summary each.

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.

Category	generation
Subcategory	single-document abstractive summarization
Page status	active
Metric	ROUGE-1/2/L
Direction	higher_is_better
Unit	F1
Dataset size	226711
Publisher	University of Edinburgh (EdinburghNLP)

What it measures

XSum tests abstractive single-document summarization. Given a full BBC news article, the model must produce a single, fluent sentence answering "what is this article about", a format designed to discourage extractive copy-paste strategies and require genuine synthesis.

Task format

Abstractive summarization; generate a single-sentence summary of a full news article.

Models reporting this benchmark

No model card in ModelSpec reports this benchmark yet.

Data

This page as JSON · Edit on GitHub