arXiv：cs.AI· Luoxi Tang, Yuqiao Meng, Nilesh Auradkar, Muchao Ye, Dazheng Zhang, Zhaohan Xi·· 7 小时前AI 评分43
如何理解并缓解 LLM 智能体的推理时记忆过度依赖
Understanding and Mitigating Inference-Time Overreliance Using Agentic Memory
AI 导读
研究提出"记忆过度依赖"失效模式：LLM 智能体检索到的记忆即使良性、存储正确、检索得当，也会扭曲推理，在查询与记忆仅部分重叠时失败最严重。为此提出即插即用框架 MEMTRIM，在写入时索引记忆证据、读取时控制复用，去除重复或冲突证据并保留有用的记忆特有信息，无需重新训练，适用于基于嵌入和结构化记忆，可在多模型与多种记忆设置下降低记忆过度依赖并保留有用记忆的收益。
正文
Abstract:Agentic memory allows LLM agents to reuse past experience, yet retrieved memories can also distort inference even when they are benign, correctly stored, and appropriately retrieved. We study this failure mode, which we call memory over-reliance. Across benchmarks and memory architectures, we find that memory is useful when past experience transfers to the current task, but can become misleading when only part of the evidence transfers. Failures are strongest under partial query-memory overlap, a pattern further confirmed by controlled experiments thatvary the amount of overlapping evidence. Motivated by this finding, we propose MEMTRIM, a plug-and-play framework that indexes memory evidence at write time and controls its reuse at read time. MEMTRIM removes repeated or conflicting evidence while preserving useful memory-specific information, requires no retraining, and applies to both embedding-based and structured memory this http URL show that MEMTRIM reduces memory overreliance while preserving the benefits of useful memory across models and memory settings.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.07311 [cs.AI] |
| (or arXiv:2610.07311v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.07311 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Luoxi Tang [view email]
[v1]
Mon, 5 Oct 2026 19:48:54 UTC (1,191 KB)
来源：arXiv：cs.AI · arxiv.org