← 研踪 PaperTrailarXiv:2603.22910 · v3
EchoKV: Efficient KV Cache Compression via Similarity-Based Reconstruction
首次提交 2026-03-24 15:58
版本更新 2026-11-03 20:27
雷达观测 2026-11-04 23:39
摘要
The increasing memory demand of the Key-Value (KV) cache poses a significant bottleneck for Large Language Models (LLMs) in long-context applications. Existing low-rank KV compression methods reduce this footprint by modifying model projections, limiting the flexibility to switch back to standard full-cache inference when sufficient memory is available. In this paper, we propose EchoKV, a flexible KV cache compression framework that supports on-demand transitions from full KV caching to compressed caching. Unlike traditional compression-decompression paradigms, EchoKV utilizes a lightweight network to reconstruct the discarded KV components from a partial subset, exploiting intrinsic inter-layer and intra-layer similarities among attention heads. We further introduce a lightweight two-stage fine-tuning strategy, requiring only a few minutes on a single A100 GPU for Llama-3.1-8B. Experimental results on LongBench and RULER demonstrate that EchoKV offers a competitive quality-cache trade-off across multiple compression ratios and backbone models while preserving the throughput of full-cache inference in short-context scenarios.
作者备注：Accepted at NeurIPS 2026
观测到的版本
相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。
v3 · 2026-11-03 20:27 · 最新观测版本
EchoKV: Efficient KV Cache Compression via Similarity-Based Reconstruction
The increasing memory demand of the Key-Value (KV) cache poses a significant bottleneck for Large Language Models (LLMs) in long-context applications. Existing low-rank KV compression methods reduce this footprint by modifying model projections, limiting the flexibility to switch back to standard full-cache inference when sufficient memory is available. In this paper, we propose EchoKV, a flexible KV cache compression framework that supports on-demand transitions from full KV caching to compressed caching. Unlike traditional compression-decompression paradigms, EchoKV utilizes a lightweight network to reconstruct the discarded KV components from a partial subset, exploiting intrinsic inter-layer and intra-layer similarities among attention heads. We further introduce a lightweight two-stage fine-tuning strategy, requiring only a few minutes on a single A100 GPU for Llama-3.1-8B. Experimental results on LongBench and RULER demonstrate that EchoKV offers a competitive quality-cache trade-off across multiple compression ratios and backbone models while preserving the throughput of full-cache inference in short-context scenarios.
查看此版本解读 →查看此版本原文 ↗