Metraipolis › archive › dc_eb400dad88fb
DSAQuant: Denoising-Stage-Aligned Quantization-Aware Training for Video Generation — arXiv:2609.04031
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.04031
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T02:56:32.957Z
- SHA-256 of the archived text
eb3358c06b26e5260eeba827c3ad1816a1088fe15e0bedc460da002f4c8c4a2c- Stable URL of this page
https://metraipolis.com/documento/dc_eb400dad88fb
How to cite
DSAQuant: Denoising-Stage-Aligned Quantization-Aware Training for Video Generation — arXiv:2609.04031 — arXiv (en), https://arxiv.org/abs/2609.04031 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_eb400dad88fb (Metraipolis archive dc_eb400dad88fb, sha256 eb3358c06b26e5260eeba827c3ad1816a1088fe15e0bedc460da002f4c8c4a2c)
BibTeX
@misc{dceb400dad88fb,
title = {DSAQuant: Denoising-Stage-Aligned Quantization-Aware Training for Video Generation — arXiv:2609.04031},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.04031},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_eb400dad88fb; sha256 eb3358c06b26e5260eeba827c3ad1816a1088fe15e0bedc460da002f4c8c4a2c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.04031. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Video diffusion models (VDMs) have achieved impressive progress in text-to-video generation, but their high memory and computational costs hinder practical deployment. Quantization-aware training (QAT) is an effective solution for compressing and accelerating advanced generative models without runtime overhead at inference. However, existing QAT methods suffer from a distinctive challenge in VDMs: while they often preserve prompt semantics, global layout, and coarse motion, the quantized model severely degrades visual details, texture fidelity, and sharpness. In this paper, we trace this degradation to the timestep-agnostic design of conventional quantization pipelines, which overlooks the stage-wise functionality of video denoising. In VDMs, early denoising steps mainly establish global structure and motion, whereas middle and late steps refine local appearance and high-frequency details. Based on this insight, we propose DSAQuant, a Denoising-Stage-Aligned Quantization-aware training framework for VDMs. During training, Denoising-Stage Oriented Supervision preserves teacher distillation in early steps for stable structure planning, while shifting later steps toward target-driven optimization to enhance detail reconstruction. During inference, Denoising-Stage Gated Guidance disables CFG in the final denoising steps to prevent it from amplifying quantization-induced errors into high-frequency artifacts. Extensive experiments on the Wan and CogVideoX families under W4A4 and W3A3 settings show that DSAQuant consistently outperforms the SOTA QAT baseline, improving the VBench average score by up to 6.60 under aggressive W3A3 quantization while preserving strong text-video alignment. These results demonstrate that effective VDM quantization requires not only reducing quantization error, but also aligning quantization training and inference with the stage-wise nature of video diffusion. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-04 in cs.CV). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2609.04031 — 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 2026-10-17T02:27:19.006Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-04?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).