Metraipolis › archive › dc_e8acdb358748

Are We Overthinking One-Step Generation? — arXiv:2610.20331

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.20331
License (as archived)
CC0-1.0
Archived
2026-10-15T14:20:05.329Z
SHA-256 of the archived text
94d4c2b305a6fba91e648b4ac56706dbd044d0c4b70f73a32de47c6f63939eca
Stable URL of this page
https://metraipolis.com/documento/dc_e8acdb358748

How to cite

Are We Overthinking One-Step Generation? — arXiv:2610.20331 — arXiv (en), https://arxiv.org/abs/2610.20331 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e8acdb358748 (Metraipolis archive dc_e8acdb358748, sha256 94d4c2b305a6fba91e648b4ac56706dbd044d0c4b70f73a32de47c6f63939eca)
BibTeX
@misc{dce8acdb358748,
  title = {Are We Overthinking One-Step Generation? — arXiv:2610.20331},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.20331},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e8acdb358748; sha256 94d4c2b305a6fba91e648b4ac56706dbd044d0c4b70f73a32de47c6f63939eca},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.20331. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Diffusion and flow matching are now dominant frameworks for generative modeling in continuous domains, but they typically require multiple network evaluations at inference. This has motivated extensive research on one-step generative modeling primarily built on diffusion and flow, though existing approaches often rely on complex training machinery. In this work, we introduce Priming Models, a simple two-stage framework for training one-step generators. Motivated by the difficulty of distributional training from scratch, Priming Models are first trained on a reconstruction objective, e.g., pixel-space denoising at a fixed noise level, before undergoing feature-space distributional post-training. Our central scientific finding is that this simple reconstruction stage is already enough to entirely alleviate the difficulty of distributional training. Empirically, across class-conditional image generation, text-to-image generation, and world modeling, Priming Models significantly advance the Pareto frontier of training compute vs. generation quality. On ImageNet at $256 \times 256$ resolution, we achieve a state-of-the-art FID of $1.19$ among one-step generators using approximately one-tenth the training compute of prior methods with comparable FID. We open-source our codebase at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.20331 — 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 2026-10-15T13:53:35.730Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).