The paper "WizardLM: Empowering Large Language Models to Follow Complex Instructions" by Can Xu from Microsoft and Peking University introduces a novel method called *Evol-Instruct* to automatically generate diverse and complex instruction data for large language models (LLMs). The method uses LLMs to evolve initial instructions into more complex ones, addressing the limitations of manually created instruction datasets. The evolved instructions are then fine-tuned on the LLaMA model to create *WizardLM*. Human evaluations and automatic evaluations using GPT-4 show that *WizardLM* outperforms existing models like Alpaca and Vicuna, particularly in handling high-complexity tasks. The paper highlights the potential of AI-evolved instructions to enhance LLMs and discusses the ethical considerations of such methods. The code and data are publicly available at https://github.com/nlpxcuan/WizardLM.The paper "WizardLM: Empowering Large Language Models to Follow Complex Instructions" by Can Xu from Microsoft and Peking University introduces a novel method called *Evol-Instruct* to automatically generate diverse and complex instruction data for large language models (LLMs). The method uses LLMs to evolve initial instructions into more complex ones, addressing the limitations of manually created instruction datasets. The evolved instructions are then fine-tuned on the LLaMA model to create *WizardLM*. Human evaluations and automatic evaluations using GPT-4 show that *WizardLM* outperforms existing models like Alpaca and Vicuna, particularly in handling high-complexity tasks. The paper highlights the potential of AI-evolved instructions to enhance LLMs and discusses the ethical considerations of such methods. The code and data are publicly available at https://github.com/nlpxcuan/WizardLM.