SOURCE-LINKED INTELLIGENCE

MLLMs Hallucinate when Information Distribution Drifts in Synergy Heads

arXiv · AI, language, vision and robotics · article · Sep 5, 2026 · UTC

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 hallucinations. HEAL first employs causal noise intervention on multi-head outputs to filter out causally redundant heads. Subsequently, it disentangle

Read original source ↗ Open in workspace

recordType

paper

region

Global