The authors apply pass@k, a metric adapted from code-generation evaluation, to image and causal video diffusion models.
Few-step diffusion can sacrifice output coverage for single-sample quality
Across image and video generation, the authors find that coverage retention depends strongly on the objective used to distill a diffusion model.
Big Tech
Yifei Wang · Xiaoyu Wu · Tsu-Jui Fu · Chen Chen · Liang-Chieh Chen · Zhe Gan · +1 more
Rice University · Apple
Research Digest··2 min read
Wang et al.
Why this paper
From Apple and Rice University
In one line
Few-step diffusion distillation can improve single-sample quality while narrowing valid-output coverage, especially when trained with distribution-matching objectives.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§