一作ACL 2026Findings· Findings of ACL
System-Mediated Attention Imbalances Make Vision-Language Models Say Yes
Tsan Tsai Chan · Varsha Suresh · Anisha Saha · Michael Hahn · Vera Demberg
摘要
Vision-language model (VLM) hallucination is commonly linked to imbalanced allocation of attention across input modalities: system, image and text. However, existing mitigation strategies tend towards an image-centric interpretation of these imbalances, often prioritising increased image attentio...