The paper discusses the development of WildfireGPT, an LLM agent tailored for wildfire analysis. The model is designed to provide accurate, specific information on wildfire risks, enhanced with additional context such as climate projections and scientific literature. The aim is to make this crucial information accessible to a diverse range of users, including researchers, urban planners, and emergency managers. The paper also discusses the limitations of generalized LLMs and the need for more domain-specific information to improve their performance.
Publication date: 13 Feb 2024
Project Page: https://arxiv.org/abs/2402.07877
Paper: https://arxiv.org/pdf/2402.07877
