Skywork/CSVQA
收藏官方服务：
资源简介：
CSVQA是一个用于评估视觉语言模型在STEM领域推理能力的中英文多模态数据集。该数据集包含1378个精心构建的问题-答案对，涵盖了STEM学科的多个领域，需要领域知识、视觉证据整合和高级推理。数据集还包含了多种视觉模态和现实世界的STEM场景。
CSVQA is a Chinese-English multimodal dataset designed for evaluating the reasoning capabilities of Vision-Language Models in the STEM field. The dataset contains 1,378 carefully constructed question-answer pairs spanning multiple disciplines within STEM, requiring domain knowledge, integration of visual evidence, and higher-order reasoning. It also includes diverse visual modalities and real-world STEM scenarios.
提供机构：
Skywork
