Fetching the paper…

LSceneLLM: Enhancing Large 3D Scene Understanding Using Adaptive Visual Preferences · Around