用关键词和嵌入双机制，让大模型更聪明地记住重要信息。
CAMeR: Keyword-Gated Hybrid Activation for Adaptive Memory Retention in LLM Agents
- 结合关键词匹配与嵌入相似度，动态判断记忆重要性。
- 在100轮对话中，关键记忆保留率比纯嵌入方法高1.6倍。
- 适合需要长期记忆的智能体，如客服机器人、个人助手。
大型语言模型（LLM）代理在长时间对话中积累大量信息，现有记忆系统或无差别保留所有内容，或使用统一遗忘规则，无法区分知识的相关性。本文提出CAMeR（上下文激活的记忆强化）框架，采用关键词门控的混合激活机制——结合词级杰卡德相似度（符号）与嵌入余弦相似度（非符号）——并引入自适应权重动态。对每对记忆-查询计算混合相似度得分，超过阈值的记忆获得强化，所有记忆均经历可控衰减。我们构建了CAMeR-Bench基准，包含76条记忆、100轮对话，覆盖8个主题簇，具有分层激活频率，可测试自适应保留能力，而现有基准（LoCoMO、LongMemEval）无法满足此需求。在该基准上，仅使用关键词门控的方案相较纯嵌入门控，高频率与从不引用记忆之间的保留差距提升1.6倍（剪刀差：0.039 vs. 0.024），时间驱动基线（Oblivion、SuperLocalMemory）在100轮后权重趋近于零。CAMeR的前5项检索节省83.2%的令牌（累计39k vs. 231k），同时提升检索精度。通过8种消融实验，证明关键词门控是该规模下的主要性能驱动因素。结果表明，符号-神经混合门控为LLM代理的自适应记忆保留提供了一种简单而有效的机制。
原文摘要 · Abstract (English)
Large language model (LLM) agents operating over extended dialogues accumulate vast amounts of information, yet existing memory systems either retain everything indiscriminately or apply uniform forgetting heuristics that fail to distinguish relevant from irrelevant knowledge. We present CAMeR (Context-Activated Memory Reinforcement), a memory retention framework combining keyword-gated hybrid activation -- a joint symbolic (word-level Jaccard) and sub-symbolic (embedding cosine) gating mechanism -- with adaptive weight dynamics. CAMeR computes a hybrid similarity score for each memory-query pair; memories exceeding a threshold receive reinforcement while all memories undergo controlled decay. We introduce CAMeR-Bench, a 76-memory, 100-round benchmark spanning 8 topic clusters with graded activation frequency, designed to test adaptive retention where existing benchmarks (LoCoMO, LongMemEval) cannot. On CAMeR-Bench, CAMeR's keyword gate achieves a 1.6$\times$ larger retention gap between high-frequency and never-referenced memories compared to embedding-only gating (scissors gap: 0.039 vs. 0.024), while time-driven baselines (Oblivion, SuperLocalMemory) collapse to near-zero weights over 100 rounds. CAMeR's top-5 retrieval saves 83.2\% tokens versus full-context approaches (39k vs. 231k cumulative) while producing weight signals that improve retrieval precision. Through 8 ablation conditions we establish that the keyword gate -- not learnable decay -- is the primary performance driver at this scale. Our findings demonstrate that hybrid symbolic-neural gating provides a simple yet effective mechanism for adaptive memory retention in LLM agents.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。