Seonglae ChoConsistency loss
조건부 일관성을 학습하여, 모든 중간 상태에서 가 유지되도록 설계해서 posterior에 수렴 보장
- 적은 샘플링 단계로도 고품질 샘플을 생성
- Scaling 가능
- Diffusion Model use multi step noise reduction, while Consistency model generate once
Consistency Models
Consistency Models
Diffusion models have significantly advanced the fields of image, audio, and video generation, but they depend on an iterative sampling process that causes slow generation. To overcome this...
https://arxiv.org/abs/2303.01469
