AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

xKV: Cross-Layer KV-Cache Compression via Aligned Singular Vector Extraction

Chi-Chih Chang, Wei-Cheng Lin, Chien-Yu Lin, Hung-Yueh Chiang, Yash Akhauri, Xilai Dai, Huiqiang Jiang, Yucheng Li, Luis Ceze, Kai-Chiang Wu, Mohamed S. Abdelfattah

ICML 2026 · Architectures and efficiency › Attention and transformers · approach: Long-context KV and memory attention

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
share a cross-layer SVD basis instead of per-layer SVDSingle SVD · RULER 64K (Llama-3.1-8B-Instruct)avg accuracy: 45.71 -> 88.50Improved
use cross-layer SVD instead of cosine-similarity SLERP mergingMiniCache · RULER 64K (Llama-3.1-8B-Instruct)avg accuracy: 45.04 -> 88.50Improved
use cross-layer SVD instead of cosine-similarity mergingMiniCache · LongBench (Llama-3.1-8B-Instruct)avg score: 30.63 -> 42.27Improved
use cross-layer SVD instead of KIVI-2 quantizationKIVI-2 · RULER 64K (Llama-3.1-8B-Instruct)avg accuracy: 86.87 -> 88.50Improved
use cross-layer SVD instead of token eviction (SnapKV)SnapKV · RULER 64K (Llama-3.1-8B-Instruct)avg accuracy: 89.68 -> 88.50No clear effect
use cross-layer SVD instead of token eviction on a compact-KV modelSnapKV/PyramidKV · RULER 64K (Qwen2.5-7B-Instruct-1M)avg accuracy: 89.22Improved
replace single-layer SVD key compression with cross-layer (xK-SR) in ShadowKVShadowKV · RULER 64K (selective reconstruction)avg accuracy: 87.17 -> 89.70Improved
use cross-layer compression at extreme rate (1.71x, 11.59x GPU memory)ShadowKV · RULER 64K (selective reconstruction)avg accuracy: 43.59 -> 79.18Improved
compress both keys and values with cross-layer SVD (xKV-SR) vs ShadowKV value compressionShadowKV-V · RULER 64K (selective reconstruction)avg accuracy: 70.94 -> 89.69Improved
use xK-SR/xKV-SR instead of QuestQuest · RULER 64K (selective reconstruction)avg accuracy: 84.87 -> 89.70Improved
use selective reconstruction instead of dense reconstructionxKV · Efficiency (A100)kernel speedup: below 1.0x -> 3.5xImproved
keep compressed cache on GPU (xKV-SR) rather than offloading values (ShadowKV)ShadowKV · Efficiency (A100)throughput: approx. 30% higherImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy