None defined yet.
All modalities are equal, but video is more equal: Closing the Cross-Attention Gap in Joint Video Generation
MANCE: Manifold Aware Concept Erasure