Fetching the paper…

LEO-VL: Efficient Scene Representation for Scalable 3D Vision-Language Learning · Around