---
id: 20261119-T0-05
title: "KV-LTM：把KV缓存变成个人助理的长期记忆"
title_en: "KV-LTM Turns Per-Session KV Caches into Long-Term Memory for Assistants"
url: https://ai.daily.yangsir.net/daily/20261119-T0-05
issue_date: 2026-11-19
publish_date: 2026-11-18T05:00:00.000Z
category: research
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2611.21998
---

# KV-LTM：把KV缓存变成个人助理的长期记忆

论文提出KV-LTM，将每个会话的KV缓存持久化保存，作为个人助理的长期记忆使用。现有长期记忆方案通常在问题和证据之间插入推导或检索步骤，带来信息损耗。在LongMemEval基准的系统级对比中，两款商业助手得分明显低于该方案。这意味着助手可以直接复用历史会话的缓存状态来回答问题，减少检索环节的误差。

## English Version

**KV-LTM Turns Per-Session KV Caches into Long-Term Memory for Assistants**

The paper introduces KV-LTM, which persists per-session KV caches as long-term memory for personal assistants. Existing memory stacks insert derivation or selection steps between a question and its evidence, causing information loss. In a system-level comparison on LongMemEval, two commercial assistants scored well below this approach. The implication: assistants can reuse cached states from past sessions directly to answer questions, cutting errors introduced by retrieval pipelines.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2611.21998)

**详情页**：https://ai.daily.yangsir.net/daily/20261119-T0-05

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*