视觉语言模型中思维链路提示调优（英文版）

英文标题：Chain of Thought Prompt Tuning in Vision Language Models中文摘要：本文提出了一种基于连锁式思维提示调整的视觉语言建模方法，经过广泛的实验验证，我们的方法在图像分类任务中的泛化能力更强，在单个数据集之外具有更强的可转移性和更强的领域泛化性能，而且在需要更多推理能力的图像文本检索和视觉问答方面表现更好。英文摘要：Language-Image Pre-training has demonstrated promising results on zero-shotand few-shot downstream tasks by prompting visual models with natural languageprompts. However, most recent studies only use a single prompt for tuning,neglecting the inherent step-to-step cognitive reasoning process that humansconduct i