Metraipolis › archive › dc_17c65bfbd1ed

Progressive-View On-Policy Distillation for Regional-to-Global Transfer in Multimodal LLMs — arXiv:2609.32333

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.32333
License (as archived)
CC0-1.0
Archived
2026-10-07T21:00:05.341Z
SHA-256 of the archived text
764fdb851f287589c36f5db4ce5ebec7511648b9deeca7ecdd6038b1c4629fcf
Stable URL of this page
https://metraipolis.com/documento/dc_17c65bfbd1ed

How to cite

Progressive-View On-Policy Distillation for Regional-to-Global Transfer in Multimodal LLMs — arXiv:2609.32333 — arXiv (en), https://arxiv.org/abs/2609.32333 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_17c65bfbd1ed (Metraipolis archive dc_17c65bfbd1ed, sha256 764fdb851f287589c36f5db4ce5ebec7511648b9deeca7ecdd6038b1c4629fcf)
BibTeX
@misc{dc17c65bfbd1ed,
  title = {Progressive-View On-Policy Distillation for Regional-to-Global Transfer in Multimodal LLMs — arXiv:2609.32333},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.32333},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_17c65bfbd1ed; sha256 764fdb851f287589c36f5db4ce5ebec7511648b9deeca7ecdd6038b1c4629fcf},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32333. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Regional-to-global distillation uses crop-conditioned guidance to improve full-image understanding. The challenge is to effectively transfer the teacher's crop-based advantage to the student's full-image inference. We propose progressive-view on-policy distillation (PVD), which shifts the student's view distribution from the crop toward the full image through an intermediate aspect-preserving padded crop. The padded crop preserves regional content while matching the full image's visual-token grid. Across stages, the view mixture assigns increasing probability to the full image. A lightweight regional-advantage weighting reallocates token-level supervision using the crop-conditioned teacher-student log-probability gap. Evaluated under each sampled input, it applies mild reweighting when the gap is small and emphasizes higher-gap tokens when the gap widens. A Jensen-Shannon metric decomposition interprets this schedule as a shift from matched-input imitation toward the deployment objective. Across benchmarks spanning perception, visual mathematics and general multimodal question answering, PVD-full reaches an average accuracy of 77.51 over three seeds, improving on the reward-free distillation baseline by 2.01 points and on its reward-matched variant by 1.00 point. In the reward-free setting, PVD-distill still gains 1.16 points.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.32333 — 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 2026-10-07T20:29:02.297Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).