HeGraphAdapter: Tuning Multi-Modal Vision-Language Models with Heterogeneous Graph Adapter | Read Paper on Bytez