3DGraphLLM is a model that uses a 3D scene graph and an LLM to perform 3D vision-language tasks. - View it on GitHub