arXiv：cs.AI· Xiangyi Zeng, Baihang Liu, Xutong Wang, Ze Jin, Yunpeng Li, Qixu Liu·· 4 小时前
Hippocam：面向 LLM 智能体记忆与学习的意图结构化经验整合架构
Use and Disuse: Intent-Structured Experience Consolidation for Memory and Learning in LLM Agents
AI 导读
研究者提出分层记忆与持续学习架构 Hippocam，将 LLM 智能体的工作组织为嵌套意图，已完成意图被整合为与任务相关的成果和状态而非完整工作细节。其递归前缀整合机制让长期未使用的经验愈发抽象，原始交互仍被保留，智能体可按层级逐步恢复细节并在信息足够时停止。经验被召回后会与新经验一同再整合，从而在无需参数更新的情况下通过自身经验学习并演化能力。
正文
Abstract:The evolution of Large Language Model agents from single-task execution to long-term autonomous operation highlights the critical challenge of transforming continuous experiences into reusable knowledge. To address this, we propose Hippocam, a hierarchical memory and continual learning architecture. Hippocam draws inspiration from two characteristics of human memory: cognitive processes selectively maintain information relevant to current goals, while long-term memories form gradually through repeated consolidation. Accordingly, Hippocam structures an agent's ongoing work as nested intents. The active context remains centered on the current intent, while completed intents are consolidated into the task-relevant outcomes and state needed for subsequent work, rather than carrying forward their full working details. Concurrently, a recursive prefix consolidation mechanism repeatedly consolidates earlier history, causing long-unused experiences to become increasingly abstract. Original interactions are preserved, allowing the agent to progressively recover finer-grained details through the hierarchy and stop once sufficient information is available. Crucially, when past experiences are recalled and reintegrated into active work, they undergo subsequent consolidation alongside new experiences, thereby being reinforced, supplemented, and updated. Through this memory dynamic of use and disuse, Hippocam connects working context, long-term memory, knowledge accumulation, and skill learning within a single continuously evolving experiential process. This enables agents to learn and evolve capabilities through their own experiences without parameter updates.
| Comments: | 33 pages, including references and appendices |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.12124 [cs.AI] |
| (or arXiv:2610.12124v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.12124 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Xiangyi Zeng [view email]
[v1]
Thu, 8 Oct 2026 15:16:56 UTC (774 KB)
来源：arXiv：cs.AI · arxiv.org