Metraipolis › archive › dc_ab1df9b38ea9

CapField-OPD: Learning Continuous Capability Fields via Joint-Anchored Multi-Teacher On-Policy Distillation for Flow Models — arXiv:2609.34658

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.34658
License (as archived)
CC0-1.0
Archived
2026-10-07T23:52:21.081Z
SHA-256 of the archived text
06b61e2ed6f9a9cf2eb482a31443735da835540d09e4df2739252690b030a751
Stable URL of this page
https://metraipolis.com/documento/dc_ab1df9b38ea9

How to cite

CapField-OPD: Learning Continuous Capability Fields via Joint-Anchored Multi-Teacher On-Policy Distillation for Flow Models — arXiv:2609.34658 — arXiv (en), https://arxiv.org/abs/2609.34658 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ab1df9b38ea9 (Metraipolis archive dc_ab1df9b38ea9, sha256 06b61e2ed6f9a9cf2eb482a31443735da835540d09e4df2739252690b030a751)
BibTeX
@misc{dcab1df9b38ea9,
  title = {CapField-OPD: Learning Continuous Capability Fields via Joint-Anchored Multi-Teacher On-Policy Distillation for Flow Models — arXiv:2609.34658},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.34658},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ab1df9b38ea9; sha256 06b61e2ed6f9a9cf2eb482a31443735da835540d09e4df2739252690b030a751},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34658. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Reward-specialized post-training produces strong experts for flow-based generative models, while multi-teacher on-policy distillation (OPD) consolidates their capabilities into a single student. Existing methods, however, route each prompt to a single teacher according to its semantic category, implicitly binding the desired capability to prompt content. This coupling makes capability invocation vulnerable to prompt perturbations and prevents users from explicitly adjusting the strength of the desired capability at inference time. In this work, we introduce CapField-OPD, an OPD framework that integrates multiple teachers into a continuous capability field through explicit capability coordinates. We use teacher models as anchors to construct this field, with the coordinates determining how their outputs are combined. Each capability configuration thus receives a unique supervision target, and capability control no longer depends on prompt semantics. Since the training anchors may not be optimal at inference time, we further profile the learned field on a small calibration set. The coordinate with the highest mean reward serves as the recommended default, while coordinates that are frequently optimal offer a promising candidate set for test-time scaling. Extensive experiments on compositional generation, text rendering, and visual aesthetics demonstrate that CapField-OPD consolidates multiple specialized teachers into a single student while preserving or surpassing their performance, reliably invokes the desired capabilities under semantics-preserving prompt variations, and supports continuous capability control and coordinate-based test-time scaling.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.34658 — 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 2026-10-07T23:29:27.425Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).