Advancing End-to-End Pixel Space Generative Modeling via Self-supervised Pre-training

DiffMRe-assign community