Metraipolis › archive › dc_5f6f4e292f3f

Centering before Pruning: Lightweight Geometry Correction for Diversity-Based Visual Token Pruning in LVLMs — arXiv:2608.30263

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.30263
License (as archived)
CC0-1.0
Archived
2026-10-17T07:52:32.076Z
SHA-256 of the archived text
127f6032c0bca57f013019d4000389572b72bf7910c154e60e81e5baea6c36db
Stable URL of this page
https://metraipolis.com/documento/dc_5f6f4e292f3f

How to cite

Centering before Pruning: Lightweight Geometry Correction for Diversity-Based Visual Token Pruning in LVLMs — arXiv:2608.30263 — arXiv (en), https://arxiv.org/abs/2608.30263 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5f6f4e292f3f (Metraipolis archive dc_5f6f4e292f3f, sha256 127f6032c0bca57f013019d4000389572b72bf7910c154e60e81e5baea6c36db)
BibTeX
@misc{dc5f6f4e292f3f,
  title = {Centering before Pruning: Lightweight Geometry Correction for Diversity-Based Visual Token Pruning in LVLMs — arXiv:2608.30263},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.30263},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5f6f4e292f3f; sha256 127f6032c0bca57f013019d4000389572b72bf7910c154e60e81e5baea6c36db},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30263. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large vision-language models (LVLMs) incur substantial inference costs due to their long and highly redundant visual-token sequences. Diversity-based pruning mitigates this cost by selecting token subsets based on pairwise cosine similarity. We find, however, that similarities between raw visual tokens are strongly concentrated in the positive range, limiting their ability to distinguish non-redundant tokens. A natural way to improve this resolution is to center token features before computing cosine similarity. Centering indeed reveals a substantially richer pairwise structure, yet unexpectedly degrades pruning performance when used alone. We show that this apparent contradiction arises because the raw geometry does more than represent pairwise diversity: it also implicitly favors globally distinctive tokens, which tend to contain semantically informative content. Centering better resolves subset diversity but loses this useful token-wise preference, revealing that diversity and distinctiveness are entangled in the raw geometry. Based on this analysis, we propose the \textbf{Cen}tered Geometry \textbf{Prune}r (Cen-Prune), which measures subset diversity using centered cosine similarity while retaining raw-space distinctiveness as a complementary token-wise preference. This lightweight, plug-and-play correction leaves the underlying selection mechanism unchanged and incurs negligible computational overhead. Extensive experiments across multiple image- and video-understanding benchmarks and LVLM architectures demonstrate that Cen-Prune provides robust improvements in overall performance across existing diversity-based pruners.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.AI).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.30263 — 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 2026-10-17T07:28:26.846Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).