AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

FreeKV: Boosting KV Cache Retrieval for Efficient LLM Inference

Guangda Liu, Chengwei Li, Zhenyu Ning, Jing Lin, Yiwu Yao, Danning Ke, Minyi Guo, Jieru Zhao

ICLR 2026 · Architectures and efficiency › Attention and transformers · approach: KV cache eviction and compression · 14 citations (OpenAlex)

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
add speculative retrieval using the last step query instead of the last layer query (ablation)KV retrieval (Qwen DeepSeek-R1-Qwen-7B) · AIME24avg@k accuracy: 33.00 -> 50.00Improved
use last step query instead of last layer query (ablation)DeepSeek-R1-Qwen-7B · GPQAavg@k accuracy: 30.50 -> 36.00Improved
use last step query instead of last layer query (ablation)Qwen-2.5-7B-Instruct · LongBench v2accuracy: 26.44 -> 26.63No clear effect
add fine-grained correction with tau=0.9 to speculative retrieval (ablation)DeepSeek-R1-Qwen-7B · AIME24avg@k accuracy: 50.00 -> 52.92Improved
add fine-grained correction with tau=0.9 (ablation)DeepSeek-R1-Qwen-7B · GPQAavg@k accuracy: 36.00 -> 39.50Improved
raise correction threshold from 0.8 to 0.9 (ablation)DeepSeek-R1-Qwen-7B · GPQAavg@k accuracy: 35.50 -> 39.50Improved
use mean pooling over softmax scores (MeanS) instead of MaxS (ablation)group-consistent selection · AIME24avg@k accuracy: 47.50 -> 52.92Improved
use mean pooling over softmax scores instead of mean query pooling (MeanQ) (ablation)group-consistent selection · AIME24avg@k accuracy: 46.67 -> 52.92Improved
max instead of mean pooling for group correction (ablation)group-consistent correction · AIME24avg@k accuracy: 52.92 -> 53.33No clear effect
use FreeKV instead of RazorAttention-style KV droppingFreeKV · LongGenBenchCR: 35.90 -> 78.03Improved
use FreeKV instead of ShadowKV on Qwen modelsFreeKV · LongGenBenchCR x Acc: 11.43 -> 32.81Improved
use FreeKV instead of ShadowKV on Llama-3.1-8BFreeKV · LongGenBenchCR x Acc: 30.66 -> 27.62Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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