测试不同记忆方法在长对话问答中的表现，发现记忆能大幅减少token使用。
Evaluating Long-Term Memory for Long-Context Question Answering
- 对比五种记忆机制：全上下文、检索增强、代理记忆、事件记忆和提示优化
- 记忆增强使token消耗降低90%以上，准确率仍保持竞争力
- 基础模型适合RAG，强指令微调模型更适合通过反思实现事件记忆
为实现大语言模型真正的对话连续性与经验学习，需引入记忆能力。尽管研究聚焦于复杂记忆系统开发，但何种记忆对长上下文对话任务最有效仍不明确。本文系统评估了五类记忆增强方法在标注有问答任务的长对话数据上的表现，涵盖全上下文提示、通过检索增强生成的语义记忆、代理型记忆、通过上下文学习的事件记忆，以及通过提示优化的程序记忆。结果表明，记忆增强方法可将token使用量减少超过90%，同时保持竞争性准确率。记忆架构复杂度应与模型能力匹配：基础模型从RAG中获益最大，而更强的指令微调模型则更受益于通过反思实现的事件记忆及更复杂的代理语义记忆。特别地，事件记忆有助于模型识别自身知识边界。
原文摘要 · Abstract (English)
In order for large language models to achieve true conversational continuity and benefit from experiential learning, they need memory. While research has focused on the development of complex memory systems, it remains unclear which types of memory are most effective for long-context conversational tasks. We present a systematic evaluation of memory-augmented methods on long-context dialogues annotated for question-answering tasks that require diverse reasoning strategies. We analyse full-context prompting, semantic memory through retrieval-augmented generation and agentic memory, episodic memory through in-context learning, and procedural memory through prompt optimization. Our findings show that memory-augmented approaches reduce token usage by over 90\% while maintaining competitive accuracy. Memory architecture complexity should scale with model capability, with foundation models benefitting most from RAG, and stronger instruction-tuned models gaining from episodic learning through reflections and more complex agentic semantic memory. In particular, episodic memory can help LLMs recognise the limits of their own knowledge.
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。