Metraipolis › archive › dc_80ce29eca87c

PReCache: Efficient KV Cache Sharing for Multi-LoRA Agents via Low-Rank Precomputation and Neutral Reconstruction — arXiv:2609.34054

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.34054
License (as archived)
CC0-1.0
Archived
2026-10-07T22:37:58.195Z
SHA-256 of the archived text
4d23e24127bd4948b6714ddcdd5314a1d910ee2dc15bb8294539db8ff37222a1
Stable URL of this page
https://metraipolis.com/documento/dc_80ce29eca87c

How to cite

PReCache: Efficient KV Cache Sharing for Multi-LoRA Agents via Low-Rank Precomputation and Neutral Reconstruction — arXiv:2609.34054 — arXiv (en), https://arxiv.org/abs/2609.34054 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_80ce29eca87c (Metraipolis archive dc_80ce29eca87c, sha256 4d23e24127bd4948b6714ddcdd5314a1d910ee2dc15bb8294539db8ff37222a1)
BibTeX
@misc{dc80ce29eca87c,
  title = {PReCache: Efficient KV Cache Sharing for Multi-LoRA Agents via Low-Rank Precomputation and Neutral Reconstruction — arXiv:2609.34054},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.34054},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_80ce29eca87c; sha256 4d23e24127bd4948b6714ddcdd5314a1d910ee2dc15bb8294539db8ff37222a1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34054. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Multi-LoRA agent systems enable efficient role specialization by sharing a common backbone model. However, each agent repeatedly processes the growing shared trajectory and constructs its own KV cache, introducing substantial memory and computation redundancy in long-horizon tasks. Existing KV cache sharing methods reduce this repeated prefill, but they either require additional training or architectural constraints or retain substantial model computation. Moreover, direct cache reuse causes the current agent to rely on cache states generated by the previous agent's adapter, weakening the role-specific behavior encoded by its own LoRA. We present PReCache, a training-free KV cache sharing framework with two designs, namely PreLRShared and ReBaseShared, that share the base cache computed using the pretrained weights and precompute a compact agent-specific low-rank (LR) cache. To remove repeated prefill, PreLRShared precomputes each agent's LR cache when the shared context is first processed, allowing the current agent to use its own LR cache without reprocessing context processed by previous agents. To improve sharing accuracy, ReBaseShared reconstructs the shared base cache from adapter-free hidden states, reducing the remaining error caused by the previous agent's adapted representation. To minimize its reconstruction cost, we propose two inference schemes tailored to single-stream inference and concurrent serving, performing the same reconstruction after each agent's turn or alongside its execution, respectively. Across multiple models and agent benchmarks, PreLRShared achieves up to a 3.1x TTFT speedup and a 2.3x improvement in per-request throughput over inference without KV cache sharing. ReBaseShared best preserves accuracy overall among the evaluated cache-sharing methods, with an average drop of only 1.1 points relative to inference without cache sharing.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.34054 — 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 2026-10-07T22:14:28.099Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).