作者位次 #2/7ACL 2026Findings· Findings of ACL
Infinite Babble: Inflating 3 D Vision-Language Model Inference Overhead via Adversarial Geometric Perturbation
Shuoyang Sun · Jiaxin Hong · Yv Zhang · Kuofeng Gao · Hao Fang · Bin Chen · Shu-Tao Xia
摘要
3D Vision-Language Models (3D-VLMs) have emerged as the critical cognitive backbone for spatial intelligence, enabling precise reasoning over unstructured 3D data. While these models serve as the foundation for downstream robotics and embodied systems, their reliance on autoregressive decoding in...