Choice-75
收藏资源简介:
Choice-75是由匹兹堡大学和宾夕法尼亚大学合作创建的数据集,专注于研究脚本学习中的决策分支问题。该数据集包含75个脚本,每个脚本都有多个描述性场景,总计超过600个场景。创建过程中采用了人机交互的方法,确保场景的质量和挑战性。数据集的应用领域主要集中在提高智能系统在复杂决策场景中的表现,特别是在需要多步推理的难题上。
Choice-75 is a collaborative dataset developed by the University of Pittsburgh and the University of Pennsylvania, focused on investigating decision branching problems in script learning. It comprises 75 individual scripts, each containing multiple descriptive scenarios, with an overall count of over 600 scenarios. A human-computer interaction workflow was implemented during its creation to ensure the quality and challenging nature of all scenarios. The primary applications of this dataset center on enhancing the performance of intelligent systems in complex decision-making scenarios, particularly for challenging problems that demand multi-step reasoning capabilities.

- 1Choice-75: A Dataset on Decision Branching in Script Learning匹兹堡大学 · 2024年



