Structural Pruning for Diffusion Models Gongfan Fang Xinyin Ma Xinchao Wang National University of Singapore

The impressive capability of these models, however, often entails significant computational overhead during both training and inference.