Paper Detail

No Channels Left Behind: Understanding Diffusion Acceleration through Residual Gates

Junha Park, Jaemyung Yu, Jaehui Hwang, Jong-Seok Lee, Dongyoon Han

huggingface Score 4.5

Published 2026-10-29 · First seen 2026-11-02

General AI

Abstract

Training acceleration methods for diffusion transformers, such as REPA and dispersive regularization, are commonly explained by additional supervision, spatial-structure priors, or improved feature geometry. We revisit this view and uncover a striking internal change: the AdaLN-Zero residual gate α, which directly controls how conditioned updates enter the residual stream, is substantially reorganized by these methods. This observation motivates our hypothesis that acceleration arises, at least in part, from reshaping the residual gating pathway rather than solely from the feature structures explicitly imposed by auxiliary objectives. Supporting this view, accelerated models are markedly more sensitive to perturbations of α, and their gate geometry reflects semantic relationships among classes. We further introduce the normalized participation ratio (nPR) to characterize the channel-wise organization of the gates and find that, across REPA teachers, projector architectures, and radial-objective targets, broader gate participation accompanies better generation quality. Guided by this insight, we propose a simple teacher-free radial objective that promotes broader gate participation through feature-norm regularization, without pairwise feature comparisons. On ImageNet 256times256 with SiT-B/2, our method improves FID over dispersive loss while increasing mean gate participation over timesteps and classes, further supporting the link between gate organization and training acceleration. Our results identify residual-gate reorganization as a measurable axis of diffusion training acceleration and suggest a simple direction for further acceleration.

Workflow Status

Review status
pending
Role
unreviewed
Read priority
later
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

BibTeX

@misc{park2026no,
  title = {No Channels Left Behind: Understanding Diffusion Acceleration through Residual Gates},
  author = {Junha Park and Jaemyung Yu and Jaehui Hwang and Jong-Seok Lee and Dongyoon Han},
  year = {2026},
  abstract = {Training acceleration methods for diffusion transformers, such as REPA and dispersive regularization, are commonly explained by additional supervision, spatial-structure priors, or improved feature geometry. We revisit this view and uncover a striking internal change: the AdaLN-Zero residual gate α, which directly controls how conditioned updates enter the residual stream, is substantially reorganized by these methods. This observation motivates our hypothesis that acceleration arises, at least },
  url = {https://huggingface.co/papers/2610.40914},
  keywords = {AdaLN-Zero, residual gate, diffusion transformers, REPA, dispersive regularization, normalized participation ratio, radial objective, feature-norm regularization, huggingface daily},
  eprint = {2610.40914},
  archiveprefix = {arXiv},
}

Metadata

{}