representation alignment
The process of ensuring that the learned representations from different modalities or views of data are compatible and consistent, crucial for multimodal tasks.
- Aligning Text to Image in Diffusion Models is Easier Than You Think
- Cross City Traffic Flow Generation via Retrieval Augmented Diffusion Model
- Learning Diffusion Models with Flexible Representation Guidance
- REPA Works Until It Doesn’t: Early-Stopped, Holistic Alignment Supercharges Diffusion Training
- U-REPA: Aligning Diffusion U-Nets to ViTs
- VideoREPA: Learning Physics for Video Generation through Relational Alignment with Foundation Models