HiTZ/informative_judge

这是一个用于训练评估文本信息性的语言模型的多语言数据集。包含英语、巴斯克语、加泰罗尼亚语、加利西亚语和西班牙语等多种语言。每个实例由一个问题、一个正确的答案和一个要评估的回答组成，目的是训练模型判断回答的信息性。

This is a multilingual dataset for training language models to evaluate the informativeness of text. It includes multiple languages such as English, Basque, Catalan, Galician, and Spanish. Each instance consists of a question, a correct answer, and an answer to be evaluated, aiming to train models to judge the informativeness of the answers.