Metraipolis › archive › dc_a53b435477ba
MoEZip: Routing-Aware KV Cache Compression for Sparse Mixture-of-Experts LLMs — arXiv:2610.30206
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.30206
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-25T08:33:45.810Z
- SHA-256 of the archived text
936825f17da1df690f2f21bb7a82e09182ff1cb0e08e96b3869a26b23ff55300- Stable URL of this page
https://metraipolis.com/documento/dc_a53b435477ba
How to cite
MoEZip: Routing-Aware KV Cache Compression for Sparse Mixture-of-Experts LLMs — arXiv:2610.30206 — arXiv (en), https://arxiv.org/abs/2610.30206 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a53b435477ba (Metraipolis archive dc_a53b435477ba, sha256 936825f17da1df690f2f21bb7a82e09182ff1cb0e08e96b3869a26b23ff55300)
BibTeX
@misc{dca53b435477ba,
title = {MoEZip: Routing-Aware KV Cache Compression for Sparse Mixture-of-Experts LLMs — arXiv:2610.30206},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.30206},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a53b435477ba; sha256 936825f17da1df690f2f21bb7a82e09182ff1cb0e08e96b3869a26b23ff55300},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.30206. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Mixture-of-Experts (MoE) language models scale model capacity efficiently by activating a small subset of parameters per token. However, in long-context inference, memory remains a major bottleneck because the KV cache grows linearly with sequence length. Existing KV cache compression methods reduce this cost for dense transformers, but overlook a problem specific to MoE: KV eviction can change which experts are selected during answer generation, degrading the generation quality. We propose MoEZip, a routing-aware KV cache compression for sparse MoE language models. To estimate the sensitivity of expert routing to perturbation, we propose a novel metric based on the Fisher information matrix of routing distributions. MoEZip combines this metric with attention weights for routing-aware scoring of contexts. Finally, we compose a complementary scoring system that combines eviction methods for dense and sparse-MoE architectures. Experiments show that MoEZip achieves robust performance under aggressive cache compression on various benchmarks. At a KV retention ratio of 10\%, MoEZip improves over the SoTA baseline by +24 and +20 points on average on LongBench and RULER, corresponding to 2.3$\times$ and 4.9$\times$ higher scores, respectively. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.30206 — 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 2026-10-25T08:31:13.124Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).