一作ACL 2026Findings· Findings of ACL
Fixing Semantic Blind Spots in Anchor Tokens of d MLLM s
Ruixuan Xu · Jiexi Xu · Qiyan Zhao · Xiaofeng Zhang
摘要
Recent advances in diffusion-based Multimodal Large Language Models (dMLLMs) offer a compelling alternative to autoregressive counterparts; however, they remain prone to hallucinations. Through information flow analysis on LLaDA-V, we identify two intertwined factors contributing to this issue. F...