Metraipolis › archive › dc_3ec05aaa0891

SPHQuant: Efficient extreme low bit weight quantization for Vision-Language Models — arXiv:2609.24875

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.24875
License (as archived): CC0-1.0
Archived: 2026-10-09T22:06:26.061Z
SHA-256 of the archived text: b11229d3d2fdfd95b7e2723312467a2d28ca93eefa629c87751d5a3bb5352027
Stable URL of this page: https://metraipolis.com/documento/dc_3ec05aaa0891

How to cite

SPHQuant: Efficient extreme low bit weight quantization for Vision-Language Models — arXiv:2609.24875 — arXiv (en), https://arxiv.org/abs/2609.24875 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3ec05aaa0891 (Metraipolis archive dc_3ec05aaa0891, sha256 b11229d3d2fdfd95b7e2723312467a2d28ca93eefa629c87751d5a3bb5352027)

BibTeX

@misc{dc3ec05aaa0891,
  title = {SPHQuant: Efficient extreme low bit weight quantization for Vision-Language Models — arXiv:2609.24875},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.24875},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3ec05aaa0891; sha256 b11229d3d2fdfd95b7e2723312467a2d28ca93eefa629c87751d5a3bb5352027},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.24875. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Recent foundation models are moving toward native multimodal Vision-Language Models (VLMs), making VLMs a central form of next-generation foundation models. However, their large language backbones make edge deployment difficult due to high memory footprint and memory-bound autoregressive decoding. Weight-only post-training quantization is a practical solution, but pushing VLMs to extreme low bit-widths remains challenging: existing rotation-free methods suffer from outliers at 2-3 bits, while rotation-based methods improve accuracy at the cost of additional runtime overhead. We propose SPHQuant, a rotation-free spherical weight-only quantization framework for VLMs. Instead of quantizing weights directly in Cartesian coordinates, SPHQuant decomposes each 8D weight vector into coordinate signs, radius, and a positive unit direction. This representation isolates outlier magnitude into the radius while keeping directions bounded and statistically regular. Based on this insight, SPHQuant allocates extra precision to the radius to mitigate accuracy degradation induced by outliers. It further uses a compact positive-direction codebook and fine-tunes codebook entries through angular parameterization to preserve the unit-sphere constraint. We also design a hardware-friendly GEMV kernel that keeps the direction codebook small enough for shared-memory lookup and packs radial bits efficiently. Experiments show that SPHQuant matches the performance of state-of-the-art extreme low-bit quantization methods while improving decode throughput over QTIP by 30.3% on RTX A6000. Code will be released in this https URL .

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-22 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.24875 — 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 2026-10-09T21:45:21.101Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-22?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).