Interactive Prompting

Interactive prompting explores how to effectively guide large language models (LLMs) to achieve desired outputs by carefully crafting input prompts, focusing on prompt engineering techniques, prompt structure optimization, and the integration of LLMs with other models (e.g., for vision-language tasks). Current research investigates methods to improve LLM performance through prompt design, including the use of chain-of-thought prompting, active learning for prompt selection, and the development of novel prompt formats like those incorporating symbols and attributes. This field is significant for advancing LLM capabilities across diverse applications, from education and code documentation to robot control and image generation, by enhancing the reliability, efficiency, and interpretability of LLM-based systems.