Kingsoft Office Unveils 'PPT Text-to-Image Model' at Global Machine Learning Conference, Elevating Creation Efficiency by More Than 80%
2025-04-21 / Read about 0 minute
Author:小编   

On April 19, the "2025 Global Machine Learning Technology Conference," co-hosted by CSDN and Boolan, commenced in Shanghai. During the event, Kingsoft Office presented its groundbreaking research achievement: the PPT Text-to-Image Model. Unlike traditional large-scale models that generate PPTs, Kingsoft Office's WPS intelligently crafts PPTs based on the content of user-uploaded documents. Technically, the WPS AI system first performs a structured analysis and reconstruction of the documents, then employs an LLM (Large Language Model) to create an outline, guaranteeing that the PPT content is closely aligned with user requirements. Zhang Jiarui, the head of AI application algorithms at Kingsoft Office, disclosed that this model can elevate creation efficiency by more than 80%, achieving an intelligent transformation from textual to visual expression.