ALIGNX
收藏资源简介:
ALIGNX是一个大规模数据集,包含超过130万个个性化偏好示例。该数据集由中国人民大学高灵人工智能学院和蚂蚁集团创建,通过抓取论坛互动中的多样化偏好模式,旨在有效建模个性化语言模型对齐。数据集涵盖了行为和描述性人物角色,以及基础偏好方向,通过整合用户生成的内容和成对比较反馈来构建。ALIGNX数据集的应用领域是推进语言模型向真正适应用户的AI系统发展。
ALIGNX is a large-scale dataset containing over 1.3 million personalized preference examples. Developed by the Gaoling School of Artificial Intelligence at Renmin University of China and Ant Group, this dataset extracts diverse preference patterns from forum interactions, aiming to effectively model personalized language model alignment. It covers behavioral and descriptive persona roles as well as basic preference directions, and is constructed by integrating user-generated content and pairwise comparison feedback. The ALIGNX dataset is designed to advance the development of language models toward truly user-adaptive AI systems.

- 1From 1,000,000 Users to Every User: Scaling Up Personalized Preference for User-level Alignment中国人民大学高灵人工智能学院,北京,中国;蚂蚁集团,北京,中国 · 2025年



