aiqbench analysis
This research introduces a geometry-aware time reparameterization technique for flow-map distillation, aiming to improve the efficiency of one- and few-step image generation in generative ODEs. By allocating more student time to trajectory segments with high normal acceleration, the method enhances sample quality without requiring teacher retraining or additional inference-time parameters. The approach demonstrates measurable gains over standard identity-time distillation on benchmarks like CIFAR-10 and CelebA-64, offering a practical optimization for accelerated diffusion models.
Geometry-Aware Time Reparameterization for Flow-Map Distillation
arXiv:2610.02427v1 Announce Type: cross Abstract: 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.
Abstract: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.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2610.02427 [cs.LG]
(or arXiv:2610.02427v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.02427 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 276 words
Related coverage
VEGA: Learning Navigation VLAs from In-the-Wild Egocentric Video with Geometric Trajectory Supervision
arXiv · Jul 30, 2026
ForeTime-VLA: Causal Future-Token Distillation from a World Action Model for Conveyor-Belt Manipulation
arXiv · Aug 24, 2026
OneFixer: High-Quality and Consistent One-Step Autoregressive 3DGS Refinement for Driving Scenes
arXiv · Sep 29, 2026
StraightDP: Geometry-Aware Differential Privacy for Rectified-Flow Transformers
arXiv · Aug 3, 2026