🏢 School of Data Science, University of Science and Technology of China
Stabilize the Latent Space for Image Autoregressive Modeling: A Unified Perspective
·2011 words·10 mins·
loading
·
loading
Computer Vision
Image Generation
🏢 School of Data Science, University of Science and Technology of China
DiGIT stabilizes image autoregressive models’ latent space using a novel discrete tokenizer from self-supervised learning, achieving state-of-the-art image generation.