Official Implementation for LLaMA-VID: An Image is Worth 2 Tokens in Large Language Models - View it on GitHub