Metraipolis › archive › dc_92e0576c9869

OmniOPD: Stable On-Policy Distillation for Joint Audio-Video Generation — arXiv:2611.02803

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.02803
License (as archived)
CC0-1.0
Archived
2026-11-03T10:37:58.363Z
SHA-256 of the archived text
ff28a5371b07dac27c6a85c4e761ef03be645768e6b0a10a500edc0e9a2280bf
Stable URL of this page
https://metraipolis.com/documento/dc_92e0576c9869

How to cite

OmniOPD: Stable On-Policy Distillation for Joint Audio-Video Generation — arXiv:2611.02803 — arXiv (en), https://arxiv.org/abs/2611.02803 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_92e0576c9869 (Metraipolis archive dc_92e0576c9869, sha256 ff28a5371b07dac27c6a85c4e761ef03be645768e6b0a10a500edc0e9a2280bf)
BibTeX
@misc{dc92e0576c9869,
  title = {OmniOPD: Stable On-Policy Distillation for Joint Audio-Video Generation — arXiv:2611.02803},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.02803},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_92e0576c9869; sha256 ff28a5371b07dac27c6a85c4e761ef03be645768e6b0a10a500edc0e9a2280bf},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.02803. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Reinforcement learning (RL) elevates joint audio-video generation, but optimizing heterogeneous, modality-specific audio and video rewards often introduces cross-modal interference. On-policy distillation (OPD) alleviates this issue by transferring the capabilities of specialized audio and video RL teachers into a single student model. However, a significant imbalance in cross-modal token counts leaves audio with sparse supervision, causing unstable distillation. We observe that naively increasing the audio loss weight fails to solve this problem since audio learning relies heavily on visual context, indicating richer visual cues are crucial to offset sparse audio signals. To address this, we propose OmniOPD, a modality-wise on-policy distillation framework designed for balanced and stable audio-video generation. First, we construct privileged visual information using asymmetric noise scheduling, allowing the audio teacher to generate more informative targets from cleaner video context. Second, we introduce grouped multi-noise sampling to expand audio supervision coverage and stabilize training. Third, we introduce timestep-wise importance weighting based on teacher-student transition divergence, preserving reliable supervision while down-weighting unreliable learning signals, to stabilize fluctuations in audio learning. Evaluated on JavisBench and VBench demonstrate OmniOPD preserves audio quality while significantly enhancing video generation, achieving effective joint distillation.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI); Sound (cs.SD).
Pagina del lavoro: https://arxiv.org/abs/2611.02803 — 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 2026-11-03T10:29:48.903Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).