This project, trained on a large-scale Chinese dataset (~200 million text-image pairs), offers several pre-trained models in various sizes and technical reports, enabling users to perform feature extraction of Chinese text and images with just a few lines of code.