Metraipolis › archive › dc_d31e70c05ea7

Geometry-Aware Time Reparameterization for Flow-Map Distillation — arXiv:2610.02427

Source
arXiv
Original
https://arxiv.org/abs/2610.02427
License (as archived)
CC0-1.0
Archived
2026-10-05T22:50:00.264Z
SHA-256 of the archived text
28f12abe42683f647b26af5d2d470f11dafa7b780b28604be41fbd2d445f9642
Stable URL of this page
https://metraipolis.com/documento/dc_d31e70c05ea7

How to cite

Geometry-Aware Time Reparameterization for Flow-Map Distillation — arXiv:2610.02427 — arXiv, https://arxiv.org/abs/2610.02427 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d31e70c05ea7 (Metraipolis archive dc_d31e70c05ea7, sha256 28f12abe42683f647b26af5d2d470f11dafa7b780b28604be41fbd2d445f9642)
BibTeX
@misc{dcd31e70c05ea7,
  title = {Geometry-Aware Time Reparameterization for Flow-Map Distillation — arXiv:2610.02427},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.02427},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d31e70c05ea7; sha256 28f12abe42683f647b26af5d2d470f11dafa7b780b28604be41fbd2d445f9642},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.02427. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Flow-map distillation enables one- and few-step generation by learning finite-time transitions of a pretrained generative ODE. We investigate whether changing the teacher's time parameterization can make these transitions easier to learn. Motivated by the hypothesis that trajectory segments with large normal acceleration are harder to distill, we propose a geometry-aware time reparameterization that allocates more student time to these regions while preserving the teacher's geometric paths and terminal distribution. We derive a shared clock that equalizes a population normal-acceleration statistic under suitable assumptions, and construct a practical approximation from robust, regularized estimates across teacher trajectories. We incorporate this clock into Lagrangian flow-map distillation, using the transformed time coordinate to condition the student. The clock is estimated once before distillation and requires neither teacher retraining nor additional student parameters or inference-time network evaluations. Experiments on synthetic data, CIFAR-10, and CelebA-64 show improved sample quality over identity-time distillation at matched inference budgets, including improvements in one- and two-step image generation. The gains in one-step generation, where no intermediate sampling times can be adjusted, highlight the benefits of time reparameterization during distillation.

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/2610.02427 — 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 2026-10-05T22:45:55.752Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).