ValuePilot Dataset
收藏资源简介:
ValuePilot数据集是由清华大学知识工程实验室和BIGAI提出的一种新型数据集,旨在支持AI进行价值驱动的决策制定。该数据集通过DGT工具包生成,包含11938个场景和100,255个动作,每个动作根据六个价值维度(好奇心、活力、安全、幸福感、亲密感和公平性)进行评分。数据集的创建结合了自动化技术和人工审查,以确保场景与动作的合理性和价值维度的准确性。该数据集适用于训练AI在动态现实世界环境中做出具有个性化和解释性的决策。
The ValuePilot dataset is a novel dataset proposed by the Knowledge Engineering Laboratory of Tsinghua University and BIGAI, aiming to support AI in value-driven decision-making. Generated via the DGT toolkit, this dataset contains 11,938 scenarios and 100,255 actions, with each action scored against six value dimensions: curiosity, vitality, safety, well-being, intimacy, and fairness. The development of this dataset combines automated techniques and human review to ensure the rationality of scenarios and actions as well as the accuracy of the value dimension scores. This dataset is suitable for training AI to make personalized and interpretable decisions in dynamic real-world environments.




