##article.return## KR-VLM: Enhancing Factual Reasoning in Vision-Language Models via Knowledge Retrieval and Self-Verification Download Download PDF