发表机构
Beihang University（北京航空航天大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
JustMem通过沿发现广度和读取保真度两个维度自适应调整内存访问，实现高效长期对话记忆，在LoCoMo和LongMemEval-S上取得最优性能并降低令牌开销。
AI 中文摘要
高效的长期对话记忆需要在不过度扩展提供给语言模型的上下文的情况下检索足够的证据。这具有挑战性，因为相关证据可能分布在多个会话中，而压缩可能会丢弃回答问题所需的细节。因此，不同的查询需要不同形式的内存访问。为了捕捉这些需求，我们沿两个维度来表述内存访问：发现广度，它控制证据搜索的广泛程度；以及读取保真度，它控制证据是以紧凑形式读取还是从原始对话中恢复。基于这一表述，我们引入了JustMem，它将对话历史存储为紧凑的原子记忆，并沿这两个维度调整每个查询的内存访问。具体来说，LOOKUP处理局部证据，COMPOSE拓宽发现范围以处理分布的证据，而REPLAY提高对保真度敏感的证据的读取保真度。在LoCoMo和LongMemEval-S上，JustMem在比较的记忆系统中实现了最高的平均准确率和检索召回率，同时使用显著更少的生成模型令牌进行记忆构建和推理。
英文摘要
Efficient long-term conversational memory requires retrieving sufficient evidence without indiscriminately expanding the context presented to the language model. This is challenging because relevant evidence may be distributed across multiple sessions, while compression may discard details needed for answering. Different queries therefore require different forms of memory access. To capture these demands, we formulate memory access along two dimensions: discovery breadth, which controls how broadly evidence is searched, and reading fidelity, which controls whether evidence is read in compact form or recovered from the original conversation. Based on this formulation, we introduce JustMem, which stores conversation history as compact atomic memories and adapts memory access along these two dimensions to each query. Specifically, LOOKUP handles local evidence, COMPOSE broadens discovery for distributed evidence, and REPLAY increases reading fidelity for fidelity-sensitive evidence. On LoCoMo and LongMemEval-S, JustMem achieves the highest mean accuracy and retrieval recall among the compared memory systems while using substantially fewer generative-model tokens for memory construction and inference.
Comments12 pages, 8 tables, 3 figures. Includes appendix
