Retrieved Images as Visual Thought: Training-Free Multimodal In-Context Learning for the Open-vs-Closed Gap figure
AlphaXiv 中文概览(可滚动查看)