LiveMem: Maintaining Memory State Continuity in Long-Running LLM Inference

ID: 154f3940
原文链接: https://arxiv.org/abs/2608.02515
PDF: https://arxiv.org/pdf/2608.02515v1
作者: Zhichen Liu, Ruihan Sun, Hengjie Yang, Zipeng Wu, Zhaohan Chen, Xiaofan Zhang, Yang Xu
日期: 2026-08-03
更新: 2026-08-03
分类: models
来源类型: paper
标签: llm-memory, long-running-agents, inference, state-continuity
质量评分: 4/5
抓取时间: 2026-08-05T04:19:03Z

中文导读

LiveMem 将长运行助手和 Agent 的问题表述为 context turnover 下的 state continuity：当工作上下文被替换时，固定容量的记忆状态仍要承载历史计算方法在全注意力 LLM 中加入可持续的 memory state，结合记忆导向 post-training 和 state-aware serving，在 LongMemEval 上显示证据移出当前窗口后依然可被利用

为什么值得关注

LiveMem frames long-running LLM inference as state continuity under context turnover.

Grounded relevance: authors, date, arXiv categories, and abstract claims below; no extra experimental claims beyond the abstract/metadata.

关键信息

论文标题：LiveMem: Maintaining Memory State Continuity in Long-Running LLM Inference
作者：Zhichen Liu, Ruihan Sun, Hengjie Yang, Zipeng Wu, Zhaohan Chen, Xiaofan Zhang, Yang Xu
arXiv：https://arxiv.org/abs/2608.02515
发布时间：2026-08-03
arXiv 分类：cs.CL, cs.LG
关联标签：llm-memory, long-running-agents, inference, state-continuity

English Abstract

Long-running assistants and agents consume interaction streams that eventually outgrow the context. Existing context retention, summarization, and retrieval preserve access to selected history, but do not provide a persistent state over the full lifecycle when working context changes. We formulate this missing inference capability as \emph{state continuity under context turnover}: carrying computation forward through a fixed-capacity memory state whose lifetime is independent of the active context. We introduce an intrinsic memory method, \textbf{LiveMem}, which augments a pretrained full-attention LLM with a memory state that preserves the historical information over the whole lifecycle while the main attention path retains a bounded KV window. Context turnover and memory state maintaining, memory-oriented post-training, and state-aware serving jointly make this memory state load bearing after its originating tokens are released. Our experiments show that LiveMem achieves leading overall performance among evaluated systems and other intrinsic memory methods. Experiments on LongMemEval show that LiveMem is able to answer the question based on the memory state, even when the supporting evidence has been removed from the current context, and evidence-distance analysis shows that useful information persists beyond the active window. LiveMem thus establishes state continuity as a distinct and complementary abstraction for continual LLM inference.

English Summary

Obsidian Notes

内容由 opencli arxiv paper 拉取 arXiv 元数据与摘要生成。
中文导读与价值判断均锚定在条目已有摘要、论文摘要、作者、日期与分类信息上；未补充论文摘要之外的实验细节。
content-fetcher run; entry id 154f3940; arXiv 2608.02515.