Fig. 2
From: LLM-based intelligent Q&A system for railway locomotive maintenance standardization
Process of constructing a customized LLM for locomotive data standardization. First, locomotive maintenance data are subjected to comprehensive data preprocessing to build corpora specifically for this purpose. The UIE model performs P-tuning via corpora to develop a customized LLM optimized for locomotive maintenance-related object information extraction. Finally, this customized LLM is used to obtain accurate information extraction results from the maintenance data, which include the system name, component name, accessory name, fault label, and number name.
