zhaode/cmmlu
收藏Hugging Face2025-10-29 更新2025-11-15 收录
下载链接：
https://hf-mirror.com/datasets/zhaode/cmmlu
下载链接
链接失效反馈官方服务：
资源简介：
CMMLU是一个综合性的中文评估基准，用于评估大语言模型在中文语言和文化背景下的高级知识和推理能力。本数据集是CMMLU的测试集部分，已经转换为统一的指令式JSONL格式，与C-Eval数据集的处理流程和数据结构完全相同，便于在统一框架下进行模型评测。
CMMLU is a comprehensive Chinese evaluation benchmark designed to assess the advanced knowledge and reasoning abilities of large language models in the context of Chinese language and culture. This dataset is the test split of CMMLU, converted into a unified instructional JSONL format, and follows the same processing pipeline and data structure as the C-Eval dataset, facilitating model evaluation within a unified framework.
提供机构：
zhaode
