Pingchuan Ma

Contrastive-Augmented Flow Matching for Style-Content Disentanglement

Yusong Li, Pingchuan Ma, Ming Gui, Vincent Tao Hu, Björn Ommer

Learning representations that separate content and style is crucial for controllable generation and compositional generalization. However, diffusion and flow-based models trained primarily with generative objectives often produce entangled or misaligned factors. To address this g…Read more