Turning a LLaMA model into a LLaVA

Hello. I have a custom fine-tuned LLaMA 2 model. Is it possible to make it LLaVA and fine-tune it using an image Q&A dataset?