Adaptive network pruning in Denoising Diffusion Step-aware Models enhances efficiency in data generation.
Denoising Diffusion Step-aware Models (DDSM) optimize computational efficiency by adapting network sizes for each generative step.