We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

cs

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: On Fast Sampling of Diffusion Probabilistic Models

Abstract: In this work, we propose FastDPM, a unified framework for fast sampling in diffusion probabilistic models. FastDPM generalizes previous methods and gives rise to new algorithms with improved sample quality. We systematically investigate the fast sampling methods under this framework across different domains, on different datasets, and with different amount of conditional information provided for generation. We find the performance of a particular method depends on data domains (e.g., image or audio), the trade-off between sampling speed and sample quality, and the amount of conditional information. We further provide insights and recipes on the choice of methods for practitioners.
Comments: Code is released
Subjects: Machine Learning (cs.LG)
Cite as: arXiv:2106.00132 [cs.LG]
  (or arXiv:2106.00132v2 [cs.LG] for this version)

Submission history

From: Zhifeng Kong [view email]
[v1] Mon, 31 May 2021 23:00:54 GMT (4565kb,D)
[v2] Wed, 23 Jun 2021 18:28:15 GMT (4565kb,D)

Link back to: arXiv, form interface, contact.