1 paper
Tsan Tsai Chan, Varsha Suresh, Anisha Saha +2
Vision-language model (VLM) hallucination is commonly linked to imbalanced allocation of attention across input modalities: system, image and text. However, existing mitigation str…