"Deep Breathing" Enhances Large Model Performance! Google DeepMind Uses Large Language Models to Generate Prompts, as AI Understands AI Better

Translated Data: Google DeepMind has introduced the OPRO framework, which employs large language models (LLMs) for optimization, utilizing natural language descriptions to enhance solution generation. The term "Deep Breath" has emerged as the optimal prompt, significantly improving model performance. OPRO has also demonstrated potential in solving mathematical problems, validating the effectiveness of prompt optimization across various tasks and showcasing its broad application prospects. This research opens a new avenue for using large language models in optimization tasks, carrying significant importance.