MearcAll papers

arXiv 2010.02502 · cs.LG cs.CV

Denoising Diffusion Implicit Models

Jiaming Song, Chenlin Meng, Stefano Ermon

Claims were extracted automatically by reducer-1 on 2026-11-12. The paper's license does not allow republishing it, so long claims are shortened to a quotation. Read the full text on arXiv.

No reproduction recordedres_0b1422d6
We empirically demonstrate that DDIMs can produce high quality samples 10× to 50× faster in terms of wall-clock time compared to DDPMs, allow us to trade off computation for sample … (excerpt)

ABSTRACT · page 1 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_2f003e08
We show that the resulting variational training objectives have a shared surrogate objective, which is exactly the objective used to train DDPM.

I NTRODUCTION · page 2 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_f64c48bd
In Appendix A, we show that the non-Markovian perspective also applies beyond the Gaussian case.

V ARIATIONAL INFERENCE FOR NON -M ARKOVIAN FORWARD PROCESSES · page 3 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_78fa6a63
We show that only slight changes to the updates in Eq. (12) are needed to obtain the new, faster generative processes, which applies to DDPM, DDIM, as well as all … (excerpt)

A CCELERATED GENERATION PROCESSES · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_3bebaa17
In this section, we show that DDIMs outperform DDPMs in terms of image generation when fewer iterations are considered, giving speed ups of 10× to 100× over the original DDPM … (excerpt)

E XPERIMENTS · page 6 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_93197aca
We observe that DDIM ( η = 0) achieves the best sample quality when dim(τ) is small, and DDPM (η = 1 and ˆσ) typically has worse sample quality compared … (excerpt)

S AMPLE QUALITY AND EFFICIENCY · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_3812f4e6
In Figure 4, we show that the amount of time needed to produce a sample scales linearly with the length of the sample trajectory.

S AMPLE QUALITY AND EFFICIENCY · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_875a8409
Even though DDPM could also achieve reasonable sample quality with 100× steps, DDIM requires much fewer steps to achieve this; on CelebA, the FID score of the 100 step DDPM … (excerpt)

S AMPLE QUALITY AND EFFICIENCY · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_6250f6df
In Figure 6, we show that simple interpolations in xT can lead to semantically meaningful interpolations between two samples.

I NTERPOLATION IN DETERMINISTIC GENERATIVE PROCESSES · page 9 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_0f8412cf
Our results show that DDIMs have lower reconstruction error for largerS values and have properties similar to Neural ODEs and normalizing flows.

R ECONSTRUCTION FROM LATENT SPACE · page 9 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_209b87d6
This update is more stochastic than the update for η = 1, which explains why it achieves worse performance when dim(τ) is small.

C LOSED FORM EQUATIONS FOR EACH SAMPLING STEP · page 19 · Automated extraction, not reviewed · 0 reviewed reproductions

Reproduced one of these claims?

Record your result in the Mearc app on your own machine, then use Publish this result with this paper's id (2010.02502) and the claim id shown next to the claim. Mearc creates a file that holds only what you choose to publish. Add it to the public claim data by pull request; a maintainer other than you reviews it before it counts.