[ECCV 2026] 3DZip: Spatial-Aware Feature Diversity-Guided Token Compression for 3D Question Answering
efficiency token-pruning vision-language-model multimodal-llm 3d-question-answering token-compression 3d-vlm eccv-2026 3dzip llava-3d scanqa sqa3d
-
Updated
Jul 2, 2026 - Python