标题：用GPT-4增强CLIP:利用视觉描述作为提示源文链接：Maniparambil_Enhancing_CLIP_with_GPT-4_Harnessing_Visual_Descriptions_as_Prompts_ICCVW_2023_paper.pdf(thecvf.com)https://openaccess.thecvf.com/content/ICCV2023W/MMFM/papers/Maniparambil_Enhancing_CLIP_with