New preprint out on Can Feedback Enhance Semantic Grounding in Large Vision-Language Models? on arXiv!