arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2604.12948cs.AI
收藏

基于记忆的绘制:双轨迹编码提升LLM代理跨会话回忆

Drawing on Memory: Dual-Trace Encoding Improves Cross-Session Recall in LLM Agents

Tufts University(塔夫茨大学)

Benjamin Stern, Peter Nadel

AI总结:

本文提出双轨迹编码方法,通过将事实与具体场景描述结合,提升LLM代理在跨会话中的回忆能力,实验显示在时间推理、知识更新追踪和多会话聚合方面有显著提升。

摘要:

LLM代理具有持久记忆的存储信息作为扁平事实记录,提供很少的上下文用于时间推理、变化跟踪或跨会话聚合。受绘画效应[3]的启发,我们引入双轨迹记忆编码。在此方法中,每个存储的事实都配对一个具体的场景轨迹,即对信息学习时刻和上下文的叙述重建。代理在编码过程中被迫承诺特定的上下文细节,从而创建更丰富、更独特的记忆轨迹。使用LongMemEval-S基准(4575次会话,100个回忆问题),我们比较双轨迹编码与仅事实对照组,在匹配覆盖和格式下超过99个共享问题。双轨迹达到73.7%的总体准确率,与53.5%的对照组相比,提升20.2个百分点(95%置信区间:[+12.1, +29.3],Bootstrap p < 0.0001)。增益集中在时间推理(+40pp)、知识更新追踪(+25pp)和多会话聚合(+30pp),在单会话检索方面无益,与编码特异性理论[8]一致。令牌分析显示双轨迹编码在不增加额外成本的情况下实现此增益。我们还勾勒出将双轨迹编码适应于编码代理的建筑设计,并进行了初步试点验证。

英文摘要:

LLM agents with persistent memory store information as flat factual records, providing little context for temporal reasoning, change tracking, or cross-session aggregation. Inspired by the drawing effect [3], we introduce dual-trace memory encoding. In this method, each stored fact is paired with a concrete scene trace, a narrative reconstruction of the moment and context in which the information was learned. The agent is forced to commit to specific contextual details during encoding, creating richer, more distinctive memory traces. Using the LongMemEval-S benchmark (4,575 sessions, 100 recall questions), we compare dual-trace encoding against a fact-only control with matched coverage and format over 99 shared questions. Dual-trace achieves 73.7% overall accuracy versus 53.5%, a +20.2 percentage point (pp) gain (95% CI: [+12.1, +29.3], bootstrap p < 0.0001). Gains concentrate in temporal reasoning (+40pp), knowledge-update tracking (+25pp), and multi-session aggregation (+30pp), with no benefit for single-session retrieval, consistent with encoding specificity theory [8]. Token analysis shows dual-trace encoding achieves this gain at no additional cost. We additionally sketch an architectural design for adapting dual-trace encoding to coding agents, with preliminary pilot validation.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑