OpenEvals/MuSR
收藏Hugging Face2025-12-12 更新2026-02-07 收录
下载链接：
https://hf-mirror.com/datasets/OpenEvals/MuSR
下载链接
链接失效反馈官方服务：
资源简介：
MuSR数据集旨在通过多步软推理测试思维链的极限。该数据集包含谋杀谜题、物品放置和团队分配等场景，每个场景都有叙事、问题、选项和答案。数据集使用ChatGPT创建，挑战了基于常识的多步推理能力。数据集在ICLR 2024上作为焦点展示，并提供了项目网站、论文和GitHub仓库的链接以供进一步参考。
MuSR: Testing the Limits of Chain-of-thought with Multistep Soft Reasoning. The dataset includes murder mysteries, object placements, and team allocation scenarios, each with narratives, questions, choices, and answers. Created using ChatGPT, it challenges multi-step reasoning with commonsense. Featured as a spotlight presentation at ICLR 2024, with links to the project website, paper, and GitHub repository for more details.
提供机构：
OpenEvals
