Metraipolis › archive › dc_4ced3e97572e

CACFG: Curvature-Aware Classifier-Free Guidance and Optimal Control — arXiv:2610.05845

Source
arXiv
Original
https://arxiv.org/abs/2610.05845
License (as archived)
CC0-1.0
Archived
2026-10-06T05:27:11.914Z
SHA-256 of the archived text
b1fcdf6d1563c087b53cd91a7f65bb52d240b9e73caa04c2ed0bf89be4648fd3
Stable URL of this page
https://metraipolis.com/documento/dc_4ced3e97572e

How to cite

CACFG: Curvature-Aware Classifier-Free Guidance and Optimal Control — arXiv:2610.05845 — arXiv, https://arxiv.org/abs/2610.05845 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_4ced3e97572e (Metraipolis archive dc_4ced3e97572e, sha256 b1fcdf6d1563c087b53cd91a7f65bb52d240b9e73caa04c2ed0bf89be4648fd3)
BibTeX
@misc{dc4ced3e97572e,
  title = {CACFG: Curvature-Aware Classifier-Free Guidance and Optimal Control — arXiv:2610.05845},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.05845},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_4ced3e97572e; sha256 b1fcdf6d1563c087b53cd91a7f65bb52d240b9e73caa04c2ed0bf89be4648fd3},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.05845. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Diffusion models generate samples by learning to reverse a fixed corruption process, and classifier-free guidance (CFG) is the standard mechanism for conditioning this process on a desired class or prompt. CFG can be applied at varying guidance strengths, and while higher strengths improve image quality and conditional alignment, too high a guidance strength can degrade image quality and diversity. Furthermore, CFG violates principled diffusion sampling dynamics, and existing explanations for why it works despite the violation disagree on the underlying theory or do not extend to deterministic samplers used in practice. We address both these issues. We first frame CFG sampling as a continuous-time optimal control problem, treating the sampling trajectory as a sequence of controls chosen to maximise the probability of the desired condition. Solving the resulting Hamilton--Jacobi--Bellman equation shows that CFG is recovered under specific path costs when using an unconstrained control set. We argue this lack of constraint is responsible for CFG's failure at high guidance strengths, since it permits the sampling path to move arbitrarily far from the current image estimate. To fix this, we propose curvature-aware CFG (CACFG), which constrains the control set to a hypersphere informed by the Gaussian regularisation used when training variational autoencoders. We show that the control inputs produced by CFG sampling routinely violate this bound, and that across diffusion models, datasets, and guidance schedules, CACFG achieves superior generative quality at mid-to-high guidance strengths with a less severe quality-diversity tradeoff than regular CFG.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.05845 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-06T05:24:06.075Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).