CaraJ/MME-CoT
收藏Hugging Face2025-02-13 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/CaraJ/MME-CoT
下载链接
链接失效反馈官方服务：
资源简介：
MME-CoT是一个针对大型多模态模型（LMMs）链式思维（CoT）推理性能的专业基准，涵盖数学、科学、OCR、逻辑、时空和一般场景六个领域。该数据集旨在通过一个全面的评估套件，包括三个新颖的评估指标，对推理质量、鲁棒性和效率进行细粒度评估。
MME-CoT is a specialized benchmark for evaluating the Chain-of-Thought (CoT) reasoning performance of Large Multimodal Models (LMMs), spanning six domains: math, science, OCR, logic, space-time, and general scenes. It aims to conduct a thorough evaluation of reasoning quality, robustness, and efficiency at a fine-grained level through a comprehensive evaluation suite that includes three novel metrics.
提供机构：
CaraJ
