Metraipolis › archive › dc_bdd06cc8acfb
Frame Differential On-Policy Self-Distillation for Video Reasoning — arXiv:2609.39021
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.39021
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T14:45:43.291Z
- SHA-256 of the archived text
83e21896f608fc05d4cfea8aea075ae080c9c571f8bf10a1be0c34d7739378bf- Stable URL of this page
https://metraipolis.com/documento/dc_bdd06cc8acfb
How to cite
Frame Differential On-Policy Self-Distillation for Video Reasoning — arXiv:2609.39021 — arXiv (en), https://arxiv.org/abs/2609.39021 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bdd06cc8acfb (Metraipolis archive dc_bdd06cc8acfb, sha256 83e21896f608fc05d4cfea8aea075ae080c9c571f8bf10a1be0c34d7739378bf)
BibTeX
@misc{dcbdd06cc8acfb,
title = {Frame Differential On-Policy Self-Distillation for Video Reasoning — arXiv:2609.39021},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.39021},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bdd06cc8acfb; sha256 83e21896f608fc05d4cfea8aea075ae080c9c571f8bf10a1be0c34d7739378bf},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39021. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Reinforcement learning (RL) has substantially improved the reasoning ability of multimodal language models through verifiable rewards and increasingly fine-grainedvisual or temporal credit assignment. In video reasoning, however, current RL methods typically train with a fixed sparse frame budget: increasing the number of frames makes autoregressive rollouts expensive, while too few frames may miss temporally localized events and fine-grained visual details. We present \textbf{Frame Differential On-Policy Self-Distillation (FD-OPSD)}, which transfers the useful evidence of dense frame observations to a sparse frame policy during RL training. FD-OPSD compares the policy's token level preferences for the same sampled response under sparse and dense views, and distills the resulting frame differential signal without an external teacher or dense autoregressive rollout. The method preserves sparse-frame rollouts and leaves inference unchanged. Across Qwen2.5-VL-7B and Qwen3-VL-4B on six video reasoning benchmarks, FD-OPSD yields higher overall average performance than the strongest corresponding GRPO, T-GRPO, or Video-KTR baselines across the 16, 32, and 64 frame evaluation settings. These results show that dense visual evidence can be transferred selectively during training through token level self-distillation while retaining sparse frame rollouts and unchanged inference.
Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-10-01 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.39021 — 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 2026-10-07T14:19:00.651Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).