WizardLM: Empowering Large Pre-Trained Language Models to Follow Complex Instructions


GPT-4: WizardLM is a large language model that leverages Evol-Instruct, a novel method for generating open-domain instructions, to improve its performance in following complex instructions. The model outperforms ChatGPT in high-complexity tasks, demonstrating its ability to handle challenging problems. The project offers an online demo, training data, and fine-tuning instructions for users to experience and evaluate WizardLM’s capabilities.
Read more at GitHub…