ScaleAI/VisualToolBench
收藏Hugging Face2025-12-16 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/ScaleAI/VisualToolBench
下载链接
链接失效反馈官方服务：
资源简介：
VisuAlToolBench是一个评估工具增强型视觉感知、转换和推理在多模态大型语言模型中的挑战性基准数据集。该数据集包含单轮和多轮任务，跨越不同领域，每个任务都配备了详细的评分标准，用于评估模型是否能够通过操作图像并结合通用工具来解决复杂问题。
VisuAlToolBench is a challenging benchmark for evaluating tool-enhanced visual perception, transformation, and reasoning in multimodal large language models. The dataset includes single-turn and multi-turn tasks across diverse domains, each equipped with detailed rubrics for assessing the models ability to manipulate images and integrate general-purpose tools to solve complex tasks.
提供机构：
ScaleAI
