AI Researchers from China Introduce ImageBind-LLM: A Multi-Modality Instruction Tuning Method of Large Language Models (LLMs) via ImageBind