Metraipolis › archive › dc_e901729d2ed7

Convergence rates for generative drifting flows: fixed-scale obstructions and multihead acceleration — arXiv:2609.15193

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.15193
License (as archived)
CC0-1.0
Archived
2026-10-16T00:11:27.563Z
SHA-256 of the archived text
17c82df687c95c568b986aa0555924a8bb03e5df59d4c8a1f7536da267bb5edf
Stable URL of this page
https://metraipolis.com/documento/dc_e901729d2ed7

How to cite

Convergence rates for generative drifting flows: fixed-scale obstructions and multihead acceleration — arXiv:2609.15193 — arXiv (en), https://arxiv.org/abs/2609.15193 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e901729d2ed7 (Metraipolis archive dc_e901729d2ed7, sha256 17c82df687c95c568b986aa0555924a8bb03e5df59d4c8a1f7536da267bb5edf)
BibTeX
@misc{dce901729d2ed7,
  title = {Convergence rates for generative drifting flows: fixed-scale obstructions and multihead acceleration — arXiv:2609.15193},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.15193},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e901729d2ed7; sha256 17c82df687c95c568b986aa0555924a8bb03e5df59d4c8a1f7536da267bb5edf},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.15193. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Drifting models offer a promising route to faster generative AI: they perform gradual transport during training, while generating new samples in a single step. This paper asks whether the underlying drifting process can converge rapidly to a target distribution under ideal conditions, before finite-data or optimization effects are introduced. We show that its convergence rate depends critically on how it handles spatial scale. With a single fixed resolution, fine-scale features of the target can become nearly invisible, leading to extremely slow convergence. We introduce a multihead approach that combines scale-normalized information across a continuum of resolutions. We prove that this multihead approach restores exponential convergence near standard reference distributions. These results identify fixed resolution as a key bottleneck and provide a simple route to faster one-step generative models.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-15 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.15193 — 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 2026-10-15T23:54:14.745Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).