让大模型像人一样持久推理，解决多轮对话中记忆丢失和偏见问题。
CogMem: A Cognitive Memory Architecture for Sustained Multi-Turn Reasoning in Large Language Models
- 设计三层记忆架构：长期记忆、直接访问记忆和注意力焦点机制。
- 在TurnBench测试中显著减少推理偏差与记忆衰减，保持上下文稳定。
- 适合需要长时推理的复杂任务，如科学论证、法律分析等场景。
大语言模型在单轮推理中表现优异，但在多轮交互中常出现准确率下降、连贯性丧失等问题。近期的TurnBench评估揭示了常见的失败模式：推理偏见、任务漂移、幻觉、过度自信和记忆衰减。现有方法通常拼接完整对话历史，导致上下文无限增长、计算开销上升且推理效率下降。本文提出CogMem，一种受认知启发的记忆增强型大模型架构，支持持续迭代推理。该架构包含三层：长期记忆（LTM）用于固化跨会话推理策略；直接访问（DA）记忆维护会话级笔记并检索相关长期记忆；注意力焦点（FoA）机制动态重构每轮任务相关的简洁上下文。在TurnBench上的实验表明，该设计有效缓解推理失败，控制上下文增长，并提升长链推理的一致性，朝着更可靠、类人的推理能力迈进。
原文摘要 · Abstract (English)
Large language models (LLMs) excel at single-turn reasoning but often lose accuracy and coherence over extended, multi-turn interactions. Recent evaluations such as TurnBench highlight recurring failure modes-reasoning bias, task drift, hallucination, overconfidence, and memory decay. Current approaches typically append full conversational histories, causing unbounded context growth, higher computational costs, and degraded reasoning efficiency. We introduce CogMem, a cognitively inspired, memory-augmented LLM architecture that supports sustained iterative reasoning through structured, persistent memory. CogMem incorporates three layers: a Long-Term Memory (LTM) that consolidates cross-session reasoning strategies; a Direct Access (DA) memory that maintains session-level notes and retrieves relevant long-term memories; and a Focus of Attention (FoA) mechanism that dynamically reconstructs concise, task-relevant context at each turn. Experiments on TurnBench show that this layered design mitigates reasoning failures, controls context growth, and improves consistency across extended reasoning chains, moving toward more reliable, human-like reasoning in LLMs.
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。