Metraipolis › archive › dc_d8a90a7e1e77

CONDUIT: A Unified Residual-Stream Restoration Framework for KV Cache Reuse in Vision-Language Models — arXiv:2609.05821

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.05821
License (as archived)
CC0-1.0
Archived
2026-10-16T12:34:45.904Z
SHA-256 of the archived text
fda20dc70547875a139e6eb7bfc8f4979bf03d1cae88e7a9945ad6f98dda8bb5
Stable URL of this page
https://metraipolis.com/documento/dc_d8a90a7e1e77

How to cite

CONDUIT: A Unified Residual-Stream Restoration Framework for KV Cache Reuse in Vision-Language Models — arXiv:2609.05821 — arXiv (en), https://arxiv.org/abs/2609.05821 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d8a90a7e1e77 (Metraipolis archive dc_d8a90a7e1e77, sha256 fda20dc70547875a139e6eb7bfc8f4979bf03d1cae88e7a9945ad6f98dda8bb5)
BibTeX
@misc{dcd8a90a7e1e77,
  title = {CONDUIT: A Unified Residual-Stream Restoration Framework for KV Cache Reuse in Vision-Language Models — arXiv:2609.05821},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.05821},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d8a90a7e1e77; sha256 fda20dc70547875a139e6eb7bfc8f4979bf03d1cae88e7a9945ad6f98dda8bb5},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.05821. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision-language models (VLMs) often answer new questions about recurring visual content, where reusing the key-value (KV) cache can avoid re-encoding expensive visual prefixes. Exact-prefix reuse, however, fails when the same visual content appears under a changed prefix. Selective recomputation can recover quality under a small visual-token budget, but only when the right stale tokens are refreshed. Raw-attention selection can waste budget on high-attention tokens with small value-norm proxy scores and on query-irrelevant images. To address these failure modes, we propose CONDUIT, a training-free refresh policy that unifies single- and multi-image reuse as residual-stream restoration. Building on norm-weighted attention, CONDUIT ranks cached visual tokens using cached-key query attention and an accessible pre-output cached-value-norm proxy, then applies empirical image-level relevance amplification before one global selection. With one image, the coefficient is one and the rule reduces to intra-image token selection. The method preserves model architecture and weights, adding only a single query-conditioned scoring pass at inference. At a 10% refresh budget, CONDUIT achieves 97.0-99.5% of the corresponding full-prefill five-dataset average across three VLM backbones and leads budgeted methods on average; on the MMLongBench-Doc latency subset, it uses 13.5% of full-prefill FLOPs and achieves a 2.99x time-to-first-token speedup.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-09 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.05821 — 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 2026-10-16T12:30:24.988Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).