comin/ViVerBench
收藏Hugging Face2025-10-17 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/comin/ViVerBench
下载链接
链接失效反馈官方服务：
资源简介：
ViVerBench是一个全面的多模态推理视觉验证基准，涵盖16个关键任务类别，用于评估视觉-语言模型和统一多模态模型在推理和生成过程中的反思和细化能力。
ViVerBench is a comprehensive benchmark for visual verification in multimodal reasoning, spanning 16 categories of critical tasks designed to assess the reflection and refinement capabilities of vision-language models and unified multimodal models during the reasoning and generation process.
提供机构：
comin
