Investigating and addressing hallucinations in Vision-Language Models through visual prompt dependency.
