UniWorld: Unified Multimodal Framework for Image Editing and Text-to-Image Generation
UniWorld: Unified Multimodal Framework for Image Editing and Text-to-Image Generation
UniWorld is an advanced multimodal framework that integrates high-resolution semantic encoders, replacing traditional VAE-based visual encoders. It excels in image editing and text-to-image generation tasks while utilizing significantly less training data, demonstrating effective unification of image understanding and manipulation capabilities.
来源: github.com
Last updated:
Table of contents