Metraipolis › archive › dc_944cc15cd9f2
LowRAM: Low-Rank Keys and Amortised Values for Composable KV Cache Compression at Test-Time — arXiv:2611.07025
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.07025
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-06T14:43:19.211Z
- SHA-256 of the archived text
c6c684fc0ac676fe74ea67336d45abd06878c0da39d4c0798aa5ce44740b2895- Stable URL of this page
https://metraipolis.com/documento/dc_944cc15cd9f2
How to cite
LowRAM: Low-Rank Keys and Amortised Values for Composable KV Cache Compression at Test-Time — arXiv:2611.07025 — arXiv (en), https://arxiv.org/abs/2611.07025 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_944cc15cd9f2 (Metraipolis archive dc_944cc15cd9f2, sha256 c6c684fc0ac676fe74ea67336d45abd06878c0da39d4c0798aa5ce44740b2895)
BibTeX
@misc{dc944cc15cd9f2,
title = {LowRAM: Low-Rank Keys and Amortised Values for Composable KV Cache Compression at Test-Time — arXiv:2611.07025},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.07025},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_944cc15cd9f2; sha256 c6c684fc0ac676fe74ea67336d45abd06878c0da39d4c0798aa5ce44740b2895},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.07025. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Autoregressive decoding relies on the KV cache, whose memory footprint grows linearly with context length and caps batch size and throughput. Existing post-training methods exploit redundancy across tokens, layers, and feature dimensions, or lower numerical precision. We identify a complementary axis: within each attention head, values are largely predictable from their corresponding keys. We exploit this with value cache amortisation, which fits a lightweight per-head predictor at test time and stores residuals only for the tokens it reconstructs worst, giving $4\times$ near-lossless value compression. The predictor can be initialised by meta-learning for substantially reduced prefill overhead, or fitted from scratch in settings that allow a one-off cost, such as multi-query inference. We pair value amortisation with cross-layer low-rank key compression to produce LowRAM, an asymmetric compression pipeline that stays close to lossless across three model families. By reconstructing only a selected subset of KV pairs at each decoding step, LowRAM can keep decoding overhead minimal and achieve up to $5\times$ higher peak decode throughput at $4\times$ compression. LowRAM further composes with quantisation and eviction, reaching $26.9\times$ compression at $94.5\%$ of baseline performance on long contexts. These results show that treating keys and values as structurally distinct enables efficient, aggressive, and composable KV cache compression. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2611.07025 — 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 2026-11-06T14:18:57.596Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).