Large model data processing method, device, medium and computer program product

Abstract

The present disclosure relates to the technical field of artificial intelligence, and particularly provides a large model data processing method, device, medium and computer program product. The large model data processing method comprises: obtaining work order original data of a fault scene; extracting structured fine-tuning data and unstructured fine-tuning data based on the work order original data; the structured fine-tuning data comprises business parameters of the fault scene, and the unstructured fine-tuning data comprises multi-dimensional scene description data and a solution of the fault scene; integrating the structured fine-tuning data and the unstructured fine-tuning data to obtain a fine-tuning data set corresponding to the fault scene. Through the combination of structured and unstructured fine-tuning data, the trained large model of the Internet of Things card can automatically adapt to changing user scenarios and needs, more accurately diagnose and predict the faults of Internet of Things devices, thereby solving the technical problems that the user scenarios cannot be dynamically adapted for labeling and the multi-dimensional data labeling needs in complex scenarios cannot be met.

Images