AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

HitKV: Activation Frequency Knows Which Tokens Are Important

Sanle Zhao, Yujuan Tan, Yu Jing, Zhuoxin Bai, Yue Niu, Jiayi Guo, Zongjie Wang, Ao Ren

AAAI 2026 · Architectures and efficiency › Attention and transformers · approach: KV cache eviction and compression · 0 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
replace PyramidKV eviction indicator with hit rate (P-HitKV)PyramidKV Llama3-8B-Instruct · LongBench avg, B=128Laverage score: 36.35 -> 37.30Improved
replace PyramidKV eviction indicator with hit rate (P-HitKV)PyramidKV Llama3-8B-Instruct · LongBench avg, B=1024Laverage score: 40.93 -> 40.97Improved
replace PyramidKV eviction indicator with hit rate (P-HitKV)PyramidKV Mistral-7B-Instruct-v0.2 · LongBench avg, B=128Laverage score: 35.67 -> 36.41Improved
replace PyramidKV eviction indicator with hit rate (P-HitKV)PyramidKV Mistral-7B-Instruct-v0.2 · LongBench avg, B=1024Laverage score: 41.13 -> 41.32Improved
use hit-rate eviction instead of cumulative recent-window attention (HitKV vs SnapKV)SnapKV Llama3-8B-Instruct · LongBench avg, B=128Laverage score: 35.69 -> 37.16Improved
use hit-rate eviction instead of cumulative recent-window attention (HitKV vs SnapKV)SnapKV Llama3-8B-Instruct · LongBench avg, B=1024Laverage score: 41.16 -> 41.3No clear effect
use hit-rate eviction instead of cumulative recent-window attention (HitKV vs SnapKV)SnapKV Mistral-7B-Instruct-v0.2 · LongBench avg, B=128Laverage score: 35.17 -> 36.29Improved
use hit-rate eviction instead of cumulative recent-window attention (HitKV vs SnapKV)SnapKV Mistral-7B-Instruct-v0.2 · LongBench avg, B=1024Laverage score: 41.54 -> 41.63No clear effect
use hit-rate eviction instead of SnapKV scoringSnapKV Llama3-8B-Instruct · Needle-in-a-Haystack 8K, B=64Laccuracy: 69.3 -> 70.6Improved
use hit-rate eviction instead of SnapKV scoringSnapKV Mistral-7B-Instruct-v0.2 · Needle-in-a-Haystack 32K, B=64Laccuracy: 59.5 -> 77.0Improved
compress KV cache to fixed budget 128L with HitKV vs full cacheMistral-7B-Instruct-v0.2 · Decoding latency, 24k inputspeedup: 4× inference acceleration at 24k input lengthImproved
add hit-rate (activation-frequency) eviction in place of cumulative attention scores to P-HitKV for PG19 generationP-HitKV · PG19 perplexityperplexityHurt

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