MLLMs Hallucinate when Information Distribution Drifts in Synergy Heads

Excerpt

arXiv:2609.09206v1 Announce Type: cross Abstract: Multimodal Large Language Models (MLLMs) often struggle with hallucinations, thus hindering their reliable practical applications. Existing attention-based mitigation methods mainly rely on indirect signals (e.g., attention weights) that fail to accurately reflect the actual information shift underlying hallucination generation. In this paper, we propose HEAL, Head-lEvel information disentAnglement and caLibration for identifying and mitigating h