Denoising Task Difficulty-based Curriculum for Training Diffusion Models

arXiv:2403.10348v1 Announce Type: cross
Abstract: Diffusion-based generative models have emerged as powerful tools in the realm of generative modeling. Despite extensive research on denoising across various timesteps and noise levels, a conflict persists regarding the relative difficulties of the denoising tasks. While various studies argue that lower timesteps present more challenging tasks, others contend that higher timesteps are more difficult. To address this conflict, our study undertakes a comprehensive examination of task difficulties, focusing on convergence behavior and changes in …

abstract arxiv conflict cs.cv cs.lg curriculum denoising diffusion diffusion models generative generative modeling generative models modeling noise research studies tasks tools training type