XStoryCloze

XStoryCloze translates the English StoryCloze validation set into 10 languages, testing commonsense selection of a story's correct ending.

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	reasoning
Subcategory	commonsense story-ending selection
Page status	active
Metric	accuracy
Direction	higher_is_better
Unit	percent
Dataset size	20581
Dataset licence	CC BY-SA 4.0
Publisher	Meta AI (FAIR)

What it measures

XStoryCloze tests commonsense narrative understanding across languages. Given a four-sentence story, the model must pick which of two possible one-sentence endings is the coherent continuation, using the same stories translated into 11 languages so results can be compared across them.

Task format

Binary choice; given a four-sentence story context, pick which of two candidate final sentences is the coherent ending.

Models reporting this benchmark

No model card in ModelSpec reports this benchmark yet.

Data

This page as JSON · Edit on GitHub