RSChat:Intelligent Question Answering Model for Railway Safety Knowledge

The release of ChatGPT heralds a new era in artificial intelligence, as large language models (LLMs) empower myriad industries. This paper introduces RSChat, a model based on supervised fine-tuning of multiple large language models (LLMs), exploring a novel paradigm for applying knowledge in railway safety. Initially, we perform prompt engineering using collected domain data, including key information extraction and knowledge expansion. Subsequently, leveraging the constructed fine-tuning datasets, we conduct supervised fine-tuning using multiple large language models (LLMs) as base models. Finally, experimental evaluations demonstrate that RSChat can generate more professional responses in the railway safety domain, exhibiting robust potential for application.