Poly-OPD: Heterogeneous Multi-Teacher On-Policy Distillation for Capability-Selectable Flow Models

Hugging Face Daily Papers Papers
distillation flow-matching text-to-image multi-teacher on-policy

Summary

Poly-OPD is a framework for distilling complementary strengths from heterogeneous text-to-image flow models into a single compact flow-matching student, using pixel bridges and gradient-compatible adapters. It improves GenEval and DrawBench scores while consolidating multiple teacher capabilities.

Leading open text-to-image models often carry complementary strengths: one may lead on preference-aligned aesthetics while another follows compositional instructions more faithfully. However, differences in their autoencoders and noise schedules make it difficult to transfer these strengths across models. In this paper, we present Poly-OPD, a framework that can consolidate complementary strengths of heterogeneous teachers into a single compact flow-matching student. To bridge the incompatible latent spaces of different teachers, Poly-OPD performs on-policy distillation through a pixel bridge. Each student-generated image is re-encoded by a selected teacher's encoder and refined from a noise level matched by magnitude under the teacher's noise schedule. The resulting target is further matched to the student in frozen DINOv2 space, enabling supervision across incompatible latent spaces. To retain complementary capabilities without cross-teacher interference, Poly-OPD uses a gradient compatibility diagnostic to organize its adapters: attention LoRA modules are shared across teachers, whereas feed-forward adapters remain teacher-specific. During distillation, a gap-aware curriculum devotes more training to compositional categories where the student still falls short of the teacher. As each gap narrows, training shifts toward categories with larger remaining gaps. By distilling FLUX.1-dev and Z-Image into a 2.5B SD3.5-Medium student, Poly-OPD improves GenEval from 67.3 to 73.3, surpassing both larger teachers, and raises DrawBench HPSv3 from 9.34 to 11.35, consolidating both strengths within a switchable model.
Original Article
View Cached Full Text

Cached at: 08/06/26, 05:50 AM

Paper page - Poly-OPD: Heterogeneous Multi-Teacher On-Policy Distillation for Capability-Selectable Flow Models

Source: https://huggingface.co/papers/2608.04349 Published on Aug 5

·

Submitted byhttps://huggingface.co/simingfu

fuon Aug 6

Authors:

,

,

,

,

,

,

,

,

,

,

Abstract

Leadingopentext-to-imagemodelsoftencarrycomplementarystrengths:onemayleadonpreference-alignedaestheticswhileanotherfollowscompositionalinstructionsmorefaithfully.However,differencesintheirautoencodersandnoiseschedulesmakeitdifficulttotransferthesestrengthsacrossmodels.Inthispaper,wepresentPoly-OPD,aframeworkthatcanconsolidatecomplementarystrengthsofheterogeneousteachersintoasinglecompactflow-matchingstudent.Tobridgetheincompatiblelatentspacesofdifferentteachers,Poly-OPDperformson-policydistillationthroughapixelbridge.Eachstudent-generatedimageisre-encodedbyaselectedteacher’sencoderandrefinedfromanoiselevelmatchedbymagnitudeundertheteacher’snoiseschedule.TheresultingtargetisfurthermatchedtothestudentinfrozenDINOv2space,enablingsupervisionacrossincompatiblelatentspaces.Toretaincomplementarycapabilitieswithoutcross-teacherinterference,Poly-OPDusesagradientcompatibilitydiagnostictoorganizeitsadapters:attentionLoRAmodulesaresharedacrossteachers,whereasfeed-forwardadaptersremainteacher-specific.Duringdistillation,agap-awarecurriculumdevotesmoretrainingtocompositionalcategorieswherethestudentstillfallsshortoftheteacher.Aseachgapnarrows,trainingshiftstowardcategorieswithlargerremaininggaps.BydistillingFLUX.1-devandZ-Imageintoa2.5BSD3.5-Mediumstudent,Poly-OPDimprovesGenEvalfrom67.3to73.3,surpassingbothlargerteachers,andraisesDrawBenchHPSv3from9.34to11.35,consolidatingbothstrengthswithinaswitchablemodel.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2608\.04349

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/2608.04349 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.04349 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2608.04349 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.

OPD-V: Visual On-Policy Self-Distillation with Modality Balance

Hugging Face Daily Papers

Introduces OPD-V, a visual on-policy self-distillation paradigm for multimodal large language models that leverages positive and negative teachers to exploit modality balance as privileged information, improving reasoning performance across benchmarks while reducing training cost.