2026-06-19原文2 min readUpdated: 2026-06-19

Learning When to Denoise: Optimizing Asynchronous Schedules for Latent Diffusion

The paper proposes learning an asynchronous schedule for denoising in multi-representation latent diffusion models. It introduces a schedule-corrected objective and a flexible parametric class that is convex and monotone. The schedule is learned with minimal additional compute (<1%). On ImageNet 256x256, the method achieves FID 1.05 in 200 epochs (matching a 800-epoch baseline) and FID 1.02 in 600 epochs (outperforming a 1B-parameter model). Unguided results also show significant improvements.

SourcearXiv Computer VisionAuthor: Bingshuo Qian, Xiang Cheng

[2606.19662] Learning When to Denoise: Optimizing Asynchronous Schedules for Latent Diffusion

[Submitted on 18 Jun 2026]

Title:Learning When to Denoise: Optimizing Asynchronous Schedules for Latent Diffusion

View a PDF of the paper titled Learning When to Denoise: Optimizing Asynchronous Schedules for Latent Diffusion, by Bingshuo Qian and 1 other authors

View PDF HTML (experimental)

Abstract:Multi-representation diffusion models can improve visual synthesis by denoising complementary views of an image, but their performance depends critically on the asynchronous schedule that determines when each representation is denoised. We propose to learn this schedule. Our method formulates asynchronous flow matching over multiple representation spaces and uses a schedule-corrected objective that keeps each representation's local noising-time weights fixed as the schedule changes. We instantiate the schedule with a flexible parametric class that is convex and monotone by construction, and learn it using a fast joint probe with less than 1% additional training compute. On ImageNet 256x256, the learned schedule substantially improves both convergence speed and final quality under a matched 675M-parameter XL backbone. With AutoGuidance, our 200-epoch model reaches FID 1.05, matching the 800-epoch SFD-XL baseline with 4x less training. Training to 600 epochs further improves to FID 1.02, outperforming the 1B-parameter SFD-XXL result of FID 1.04 while using a smaller model. In the unguided setting, our 200-epoch model reaches FID 2.37, already below the best 800-epoch SFD-XL result (2.54) at 4x less training, and improves to FID 2.14 at 600 epochs. Code is available at this https URL

Comments: 25 pages, 9 figures, 4 tables

Subjects:

Computer Vision and Pattern Recognition (cs.CV)

Cite as: arXiv:2606.19662 [cs.CV]

(or arXiv:2606.19662v1 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2606.19662

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Bingshuo Qian [view email] [v1] Thu, 18 Jun 2026 00:13:42 UTC (18,753 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Learning When to Denoise: Optimizing Asynchronous Schedules for Latent Diffusion, by Bingshuo Qian and 1 other authors

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.CV

new | recent | 2026-06

Change to browse by:

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)