VITAL
收藏资源简介:
VITAL数据集是一个专门针对医疗领域构建的全面基准数据集,包含13.1万个具有价值观负载的场景和5.4万个多项选择题。该数据集由墨尔本大学计算与信息系统学院和Macquarie大学计算学院共同构建,旨在评估和基准化多元对齐方法。数据集覆盖了医疗场景中多个角度和观点的多样性,为大型语言模型在医疗领域的多元对齐提供了专门的评估工具。
The VITAL dataset is a comprehensive benchmark dataset specifically constructed for the medical domain. It contains 131,000 value-laden scenarios and 54,000 multiple-choice questions. Jointly developed by the School of Computing and Information Systems at the University of Melbourne and the School of Computing at Macquarie University, this dataset aims to evaluate and benchmark multi-dimensional alignment methods. It captures the diversity of multiple perspectives and angles within medical scenarios, serving as a specialized evaluation tool for the multi-dimensional alignment of large language models (LLMs) in the medical field.




