PVD partitions the diffusion trajectory into two temporal phases, coarse structure and fine detail, and trains a dedicated half-sized expert for each phase to approximate the localized average velocity.
Splitting diffusion distillation into two phase-specific half-sized experts beats a full-size student.
The authors show that partitioning the generation timeline into coarse and fine phases, each handled by a half-sized expert, improves quality while keeping compute at one full-backbone forward pass.
Academic
Zhen Guo · Rongyuan Wu · Qiaosi Yi · Chenxi Xie · Xinyu Wei · Lei Zhang
The Hong Kong Polytechnic University · OPPO Research Institute
Research Digest··2 min read
The authors propose Phase-wise Velocity Distillation (PVD), which splits the diffusion generation process into a structural phase and a refinement phase, each handled by a half-sized expert.
Why this paper
From The Hong Kong Polytechnic University and OPPO Research Institute
In one line
Phase-wise Velocity Distillation uses two half-sized phase-specific experts to generate high-quality images in one full-backbone forward pass.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors (2 noted)
- ✓Reports numbers on named benchmarks (2 benchmarks)
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§