Reproduce “Dimensionality-Varying Diffusion Process” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Han Zhang, Ruili Feng, Zhantao Yang, Lianghua Huang, Yu Liu, Yifei Zhang, Yujun Shen, Deli Zhao and 2 more
DVDP varies image representation during diffusion, attenuating selected components before downsampling and restoring dimension during generation. It reduces measured training and sampling time without reducing diffusion steps. On LSUN Bedroom 256 × 256, it reports FID 4.88 versus DDPM’s 5.74 and speedups of 2.2× and 2.4×; high-resolution FFHQ results use a different network structure than Score-SDE.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.