LLaVA-Bench
收藏arXiv2025-09-30 收录
下载链接：
https://llava-vl.github.io/
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了24张图片和60个问题，内容丰富多样，不仅提供了详尽的描述，还覆盖了各种类型的问题。该数据集被用于比较MiniGPT-4和InstructionGPT-4模型的表现。规模上，数据集包含了24张图片和60个问题，任务是对多模态语言模型进行评估。
This dataset comprises 24 images and 60 questions, with diverse and comprehensive content that includes detailed descriptions and covers a wide range of question types. It is employed to compare the performance of MiniGPT-4 and InstructionGPT-4 models, and its core task is to evaluate multimodal language models.
