@inproceedings{zhao-etal-2026-question,
title = "Question Tells You Where the Answer Is: Intention-aware Long-Context {KV} Cache Compression",
author = "Zhao, Liang and
Feng, Xiaocheng and
Zhong, Weihong and
Huang, Lei and
Zhu, Kun and
Wang, Baoxin and
Wu, Dayong and
Hu, Guoping and
Liu, Ting and
Qin, Bing",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.acl-long.1250/",
pages = "27153--27169",
ISBN = "979-8-89176-390-6",
abstract = "The increasing context window greatly extends the capabilities of large language models, but on the other hand, it incurs an unaffordable memory overhead and computational latency due to the increasing Key-Value (KV) cache size. Recent KV cache compression methods manage to reduce the cache size by dropping irrelevant KVs. However, these methods often fail to identify crucial KVs for generation while excluding others accurately, resulting in severe information loss. To address this gap, we propose **IntentKV**, an intention-aware KV cache eviction method that identifies and retains crucial KVs according to the attention distribution of intention, which semantically reflects the user{'}s goal and determines which part of the context is relevant. The consistency between the semantics and attention distribution is further substantiated through meticulously designed experiments. On this basis, IntentKV first distinguishes intention tokens from the vanilla context tokens based on their attention distribution distances. Then, the block-wise cumulative attention is calculated via aggregating the intention token attention. Finally, blocks that acquire high cumulative attention are picked and stored in KV cache. We evaluate our method across diverse long-context tasks and models. Results demonstrate that IntentKV can effectively maintain the model performance while reducing the KV cache size from 128K to 2K, leading to a 6.3x increase in decoding speed and 7.8x enhancement in memory efficiency compared to the default setting."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhao-etal-2026-question">
<titleInfo>
<title>Question Tells You Where the Answer Is: Intention-aware Long-Context KV Cache Compression</title>
</titleInfo>
<name type="personal">
<namePart type="given">Liang</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaocheng</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weihong</namePart>
<namePart type="family">Zhong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lei</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kun</namePart>
<namePart type="family">Zhu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Baoxin</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dayong</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guoping</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ting</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bing</namePart>
<namePart type="family">Qin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Liakata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viviane</namePart>
<namePart type="given">P</namePart>
<namePart type="family">Moreira</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Jurgens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-390-6</identifier>
</relatedItem>
<abstract>The increasing context window greatly extends the capabilities of large language models, but on the other hand, it incurs an unaffordable memory overhead and computational latency due to the increasing Key-Value (KV) cache size. Recent KV cache compression methods manage to reduce the cache size by dropping irrelevant KVs. However, these methods often fail to identify crucial KVs for generation while excluding others accurately, resulting in severe information loss. To address this gap, we propose **IntentKV**, an intention-aware KV cache eviction method that identifies and retains crucial KVs according to the attention distribution of intention, which semantically reflects the user’s goal and determines which part of the context is relevant. The consistency between the semantics and attention distribution is further substantiated through meticulously designed experiments. On this basis, IntentKV first distinguishes intention tokens from the vanilla context tokens based on their attention distribution distances. Then, the block-wise cumulative attention is calculated via aggregating the intention token attention. Finally, blocks that acquire high cumulative attention are picked and stored in KV cache. We evaluate our method across diverse long-context tasks and models. Results demonstrate that IntentKV can effectively maintain the model performance while reducing the KV cache size from 128K to 2K, leading to a 6.3x increase in decoding speed and 7.8x enhancement in memory efficiency compared to the default setting.</abstract>
<identifier type="citekey">zhao-etal-2026-question</identifier>
<location>
<url>https://aclanthology.org/2026.acl-long.1250/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>27153</start>
<end>27169</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Question Tells You Where the Answer Is: Intention-aware Long-Context KV Cache Compression
%A Zhao, Liang
%A Feng, Xiaocheng
%A Zhong, Weihong
%A Huang, Lei
%A Zhu, Kun
%A Wang, Baoxin
%A Wu, Dayong
%A Hu, Guoping
%A Liu, Ting
%A Qin, Bing
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-390-6
%F zhao-etal-2026-question
%X The increasing context window greatly extends the capabilities of large language models, but on the other hand, it incurs an unaffordable memory overhead and computational latency due to the increasing Key-Value (KV) cache size. Recent KV cache compression methods manage to reduce the cache size by dropping irrelevant KVs. However, these methods often fail to identify crucial KVs for generation while excluding others accurately, resulting in severe information loss. To address this gap, we propose **IntentKV**, an intention-aware KV cache eviction method that identifies and retains crucial KVs according to the attention distribution of intention, which semantically reflects the user’s goal and determines which part of the context is relevant. The consistency between the semantics and attention distribution is further substantiated through meticulously designed experiments. On this basis, IntentKV first distinguishes intention tokens from the vanilla context tokens based on their attention distribution distances. Then, the block-wise cumulative attention is calculated via aggregating the intention token attention. Finally, blocks that acquire high cumulative attention are picked and stored in KV cache. We evaluate our method across diverse long-context tasks and models. Results demonstrate that IntentKV can effectively maintain the model performance while reducing the KV cache size from 128K to 2K, leading to a 6.3x increase in decoding speed and 7.8x enhancement in memory efficiency compared to the default setting.
%U https://aclanthology.org/2026.acl-long.1250/
%P 27153-27169
Markdown (Informal)
[Question Tells You Where the Answer Is: Intention-aware Long-Context KV Cache Compression](https://aclanthology.org/2026.acl-long.1250/) (Zhao et al., ACL 2026)
ACL
- Liang Zhao, Xiaocheng Feng, Weihong Zhong, Lei Huang, Kun Zhu, Baoxin Wang, Dayong Wu, Guoping Hu, Ting Liu, and Bing Qin. 2026. Question Tells You Where the Answer Is: Intention-aware Long-Context KV Cache Compression. In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 27153–27169, San Diego, California, United States. Association for Computational Linguistics.