The rapid progress in the field of large multimodal foundation models, especially vision-language models, has dramatically transformed the landscape of machine learning, computer vision, and natural language processing.