MultiQ

该数据集名为MultiQ，包含了来自137种不同语言类型的27,400个简单开放式问题，旨在评估大型语言模型的基本多语言能力。这些问题涵盖了从代数到地理，再到天文学的多个主题，并使用了谷歌翻译API的第三版进行翻译。该数据集被视为评估大型语言模型的银标准基准。规模上，该数据集包含了27,400个问题，任务集中在基本的开放式问题回答。

This dataset, named MultiQ, consists of 27,400 simple open-ended questions across 137 distinct language varieties, and is designed to evaluate the basic multilingual capabilities of large language models. These questions cover a broad range of topics spanning from algebra, geography to astronomy, and were translated using Google Translate API v3. This dataset is recognized as a silver standard benchmark for evaluating large language models. In terms of scale, it contains 27,400 questions, with the tasks focused on basic open-ended question answering.