Improving reasoning capabilities in Vision-Language Models (VLMs) by transferring capabilities from Large-Language Models (LLMs).
