Learnable Sparsity for Vision Generative Models

Yang Zhang, Er Jin, Wenzhong Liang +5

Diffusion models have achieved impressive advancements in various vision tasks. However, these gains often rely on increasing model size, which escalates computational complexity a…