Metraipolis › archive › dc_d69b3c07ecfc

Learning via Self-Consistency for Diffusion-based Video Reasoning — arXiv:2609.36826

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.36826
License (as archived)
CC0-1.0
Archived
2026-10-07T18:38:31.353Z
SHA-256 of the archived text
16d685cd6975fcd982b9ddaf38dd2e83576a8a1f57a8d5f5714f96d64ed5ac98
Stable URL of this page
https://metraipolis.com/documento/dc_d69b3c07ecfc

How to cite

Learning via Self-Consistency for Diffusion-based Video Reasoning — arXiv:2609.36826 — arXiv (en), https://arxiv.org/abs/2609.36826 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d69b3c07ecfc (Metraipolis archive dc_d69b3c07ecfc, sha256 16d685cd6975fcd982b9ddaf38dd2e83576a8a1f57a8d5f5714f96d64ed5ac98)
BibTeX
@misc{dcd69b3c07ecfc,
  title = {Learning via Self-Consistency for Diffusion-based Video Reasoning — arXiv:2609.36826},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.36826},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d69b3c07ecfc; sha256 16d685cd6975fcd982b9ddaf38dd2e83576a8a1f57a8d5f5714f96d64ed5ac98},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.36826. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Video generation models have demonstrated emerging zero-shot capabilities for visual reasoning, perception, and other vision tasks. However, diffusion-based video generation is inherently stochastic, while many downstream vision tasks are deterministic. Motivated by the effectiveness of self-consistency in chain-of-thought reasoning for large language models, we investigate whether self-consistency can similarly improve diffusion-based video reasoning. We first introduce a training-free test-time scaling method that samples multiple video generations and aggregates their predictions through self-consistency. Specifically, we aggregate extracted paths, locations, or masks from multiple rollouts into a consensus prediction. To reduce the inference overhead of multi-rollout generation, we read out predictions early in the denoising trajectory, which preserves consensus quality while reducing denoising steps by more than half. We further propose Rejection Fine-Tuning (RFT) to distill consensus predictions into the video generation model. The resulting model internalizes the benefit of multi-sample consensus and requires only a single generation at inference time, while substantially outperforming the original model. Experiments on three tasks, including maze solving, visual search, and referring segmentation, show that both our self-consistency inference and consensus distillation dramatically improve video-based perception and reasoning, without requiring ground-truth videos or task-specific verification. For visual search, self-consistency raises task accuracy from 48.4% for a single generation to 99.0%. The distilled model retains much of the consensus benefit with a single rollout. For 4-by-4 maze solving, consensus-based training improves the single-generation strict success rate from 72.0% to 84.0% with the same inference latency.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-30 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.36826 — 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 2026-10-07T18:04:24.755Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).