Any-OPD: Heterogeneous On-Policy Distillation for Flow-Matching Models via Representation-Space Bridging

Hugging Face Daily Papers Papers
distillation flow-matching on-policy model-compression text-to-image

Summary

Any-OPD presents the first framework for on-policy distillation between arbitrary latent flow-matching generators, enabling distillation from a 12B FLUX model to a 2.5B SD3.5 model by bridging via a frozen vision representation. It improves the student's PickScore from 0.846 to 0.884, rivaling the teacher at a fifth of its size.

On-policy distillation, in which a teacher corrects samples that the student itself generates, presupposes that the two models speak the same language: identical VAE latents, matching architectures, and a common timestep grid. We ask what happens when none of this holds, as when the strongest teacher available and the student one wishes to deploy come from different model families, and find that the standard recipes have no answer: teacher latents cannot serve as targets in a foreign coordinate system, per-pixel losses against a teacher that stochastically re-draws local detail degenerate into blur or divergence, and timestep indices lose their meaning across mismatched schedules. We present Any-OPD, to our knowledge the first framework for on-policy distillation between arbitrary pairs of latent flow-matching generators. Any-OPD treats the teacher purely as a black-box sampler and connects the two models at exactly one point: a frozen, model-agnostic vision representation in which their independently decoded outputs are compared, sidestepping every assumption about latents, features, or architecture. Trajectory correspondence is recovered by matching continuous noise levels instead of step indices, and a brief anchoring phase, in which teacher samples are re-encoded through the student's own VAE, ensures the on-policy gradient measures sample quality rather than domain mismatch. Distilling the 12B FLUX.1-dev into the 2.5B SD3.5-Medium, Any-OPD lifts the student's PickScore from 0.846 to 0.884 and HPSv3 from 9.12 to 10.97, rivaling the teacher at a fifth of its size, where direct latent regression fails to train at all.
Original Article
View Cached Full Text

Cached at: 08/05/26, 05:44 AM

Paper page - Any-OPD: Heterogeneous On-Policy Distillation for Flow-Matching Models via Representation-Space Bridging

Source: https://huggingface.co/papers/2608.03316 Published on Aug 4

·

Submitted byhttps://huggingface.co/simingfu

fuon Aug 5

Abstract

On-policydistillation,inwhichateachercorrectssamplesthatthestudentitselfgenerates,presupposesthatthetwomodelsspeakthesamelanguage:identicalVAElatents,matchingarchitectures,andacommontimestepgrid.Weaskwhathappenswhennoneofthisholds,aswhenthestrongestteacheravailableandthestudentonewishestodeploycomefromdifferentmodelfamilies,andfindthatthestandardrecipeshavenoanswer:teacherlatentscannotserveastargetsinaforeigncoordinatesystem,per-pixellossesagainstateacherthatstochasticallyre-drawslocaldetaildegenerateintoblurordivergence,andtimestepindiceslosetheirmeaningacrossmismatchedschedules.WepresentAny-OPD,toourknowledgethefirstframeworkforon-policydistillationbetweenarbitrarypairsoflatentflow-matchinggenerators.Any-OPDtreatstheteacherpurelyasablack-boxsamplerandconnectsthetwomodelsatexactlyonepoint:afrozen,model-agnosticvisionrepresentationinwhichtheirindependentlydecodedoutputsarecompared,sidesteppingeveryassumptionaboutlatents,features,orarchitecture.Trajectorycorrespondenceisrecoveredbymatchingcontinuousnoiselevelsinsteadofstepindices,andabriefanchoringphase,inwhichteachersamplesarere-encodedthroughthestudent’sownVAE,ensurestheon-policygradientmeasuressamplequalityratherthandomainmismatch.Distillingthe12BFLUX.1-devintothe2.5BSD3.5-Medium,Any-OPDliftsthestudent’sPickScorefrom0.846to0.884andHPSv3from9.12to10.97,rivalingtheteacheratafifthofitssize,wheredirectlatentregressionfailstotrainatall.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2608\.03316

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2608.03316 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.03316 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2608.03316 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Flow-OPD: On-Policy Distillation for Flow Matching Models

Hugging Face Daily Papers

Flow-OPD is a research paper introducing a two-stage on-policy distillation framework for Flow Matching text-to-image models, significantly improving generation quality and alignment metrics using Stable Diffusion 3.5 Medium.

DanceOPD: On-Policy Generative Field Distillation

Hugging Face Daily Papers

DanceOPD proposes an on-policy generative field distillation framework for flow-matching models that unifies text-to-image generation, local editing, and global editing via capability-specific routing and velocity-based training, improving multi-capability composition while preserving anchor generation quality.

AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillation

Hugging Face Daily Papers

AnyFlow introduces a novel any-step video diffusion distillation framework that optimizes full ODE sampling trajectories through flow-map transition learning and backward simulation, achieving performance that matches or surpasses consistency-based counterparts while scaling with sampling step budgets.