MuSR
收藏AI_Studio2025-09-01 更新2025-05-31 收录
下载链接：
https://aistudio.baidu.com/datasetdetail/333161
下载链接
链接失效反馈官方服务：
资源简介：
MuSR（Multistep Soft Reasoning）是一个专为评估大型语言模型（LLMs）在自然语言叙述中进行多步骤软推理能力而设计的基准测试集。该数据集由神经符号综合生成算法创建，包含复杂的推理实例，如谋杀之谜、物体放置问题或团队分配优化等，旨在测试模型在处理复杂约束满足问题（CSPs）时的表现。
MuSR (Multistep Soft Reasoning) is a benchmark dataset specifically designed to evaluate the multi-step soft reasoning capabilities of Large Language Models (LLMs) in natural language narratives. Developed using neuro-symbolic synthetic generation algorithms, this dataset includes complex reasoning instances such as murder mysteries, object placement problems, team assignment optimization scenarios and other similar cases, aiming to assess models' performance when handling complex Constraint Satisfaction Problems (CSPs).
提供机构：
keyjjjj
