Metraipolis › archive › dc_6c466ff136bc

RelaxKV: Recomputation Guided by the Query with Sparse Context Attention for Efficient KV Cache Reuse — arXiv:2609.33503

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.33503
License (as archived): CC0-1.0
Archived: 2026-10-07T20:56:51.610Z
SHA-256 of the archived text: b89779f3dc14f7d14428f80586ec9275a83bb9f138f218c4aba738b78a9ef892
Stable URL of this page: https://metraipolis.com/documento/dc_6c466ff136bc

How to cite

RelaxKV: Recomputation Guided by the Query with Sparse Context Attention for Efficient KV Cache Reuse — arXiv:2609.33503 — arXiv (en), https://arxiv.org/abs/2609.33503 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6c466ff136bc (Metraipolis archive dc_6c466ff136bc, sha256 b89779f3dc14f7d14428f80586ec9275a83bb9f138f218c4aba738b78a9ef892)

BibTeX

@misc{dc6c466ff136bc,
  title = {RelaxKV: Recomputation Guided by the Query with Sparse Context Attention for Efficient KV Cache Reuse — arXiv:2609.33503},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.33503},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6c466ff136bc; sha256 b89779f3dc14f7d14428f80586ec9275a83bb9f138f218c4aba738b78a9ef892},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33503. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Cross-request KV caching reduces the prefill cost of Retrieval-Augmented Generation (RAG), but conventional prefix caching severely limits cache reuse across requests. Position-Independent Caching (PIC) removes this constraint by reusing independent chunks, but their KV states miss cross-chunk interactions. Existing methods selectively recompute token states to recover these missing interactions, but primarily allocate the recomputation budget to selecting which states to recompute, while fixing the recomputation context to the full causal prefix. We introduce RelaxKV, which formulates selective cache repair as a joint allocation problem over repair targets and recomputation context. Guided by the user query, RelaxKV identifies layer-specific repair targets and restricts their recomputation to a query-relevant context, reducing attention computation. Across four decoder models, RelaxKV at a 15% anchor ratio improves aggregate LongBench performance over ProphetKV on all models. On Qwen3-14B, RelaxKV provides a stronger quality-TTFT trade-off than ProphetKV across a 5%-30% anchor-ratio sweep, and achieves the best selective results on RULER-MV and LV-Eval at 16K and 32K context lengths. Controlled ablations further demonstrate the importance of recomputation context selection.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.33503 — 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 2026-10-07T20:29:01.277Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).