MultiWikiQA

Full Official Name: MultiWikiQA
Submission date: Oct. 14, 2025, 1:58 p.m.

A reading comprehension dataset spanning 306 languages, based on Wikipedia articles. Questions are generated using the Gemini-1.5-pro language model, and answers appear verbatim in the articles. At most 5,000 samples are included for each language.

Creator(s)
Distributor(s)
Right Holder(s)