##article.return## RMAdapter: Reconstruction-based Multi-Modal Adapter for Vision-Language Models Download Download PDF