Metraipolis › archive › dc_751dcb458550

Deflating the Hessian: Rank-4 W4A4 Quantization for Multimodal Diffusion Transformers — arXiv:2610.11315

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.11315
License (as archived)
CC0-1.0
Archived
2026-10-09T06:23:16.673Z
SHA-256 of the archived text
f6ed409f2ac5cce137d1894efc5074fa6268396e3ae5d76aed6124817b5fc86d
Stable URL of this page
https://metraipolis.com/documento/dc_751dcb458550

How to cite

Deflating the Hessian: Rank-4 W4A4 Quantization for Multimodal Diffusion Transformers — arXiv:2610.11315 — arXiv (en), https://arxiv.org/abs/2610.11315 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_751dcb458550 (Metraipolis archive dc_751dcb458550, sha256 f6ed409f2ac5cce137d1894efc5074fa6268396e3ae5d76aed6124817b5fc86d)
BibTeX
@misc{dc751dcb458550,
  title = {Deflating the Hessian: Rank-4 W4A4 Quantization for Multimodal Diffusion Transformers — arXiv:2610.11315},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.11315},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_751dcb458550; sha256 f6ed409f2ac5cce137d1894efc5074fa6268396e3ae5d76aed6124817b5fc86d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.11315. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

In diffusion transformers, low-rank branches can mitigate 4-bit weight--activation (W4A4) post-training quantization (PTQ) loss by decomposing each weight into a low-bit residual and a high-precision low-rank component. Existing low-rank PTQ approaches, however, either optimize low-rank compensation and residual quantization separately, often requiring higher ranks, or rely on second-order weight updates without explicitly modeling activation quantization error, which becomes particularly pronounced under 4-bit quantization. To address these limitations, we present \method{}, a unified framework modeling low-rank-assisted W4A4 PTQ as a coupled calibration problem and deriving optimization-based solvers from the joint objective. Eliminating the output-side low-rank factor yields a \emph{deflated Hessian} that discounts residual errors already captured by the low-rank component, while an activation-noise surrogate is incorporated to suppress activation quantization error. Across five diffusion backbones, rank-4 \method{} consistently outperforms rank-4 SVDQuant in PSNR and LPIPS. It further surpasses rank-32 SVDQuant on SANA-1.6B, FLUX.1-schnell, and FLUX.1-dev with an $8\times$ smaller rank and up to $6.25\times$ faster quantization. Furthermore, on the Qwen3-8B LLM, rank-4 \method{} improves MMLU accuracy from 61.50\% to 68.17\% over rank-32 SVDQuant. Overall, \method{} achieves better W4A4 performance with substantially lower rank and quantization cost.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Computer Vision and Pattern Recognition (cs.CV); Optimization and Control (math.OC).
Pagina del lavoro: https://arxiv.org/abs/2610.11315 — 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 2026-10-09T05:50:31.208Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).