让大模型学会在不同任务中自动提取有用记忆，提升长期交互能力。
Self-Evolving LLM Memory Extraction Across Heterogeneous Tasks
- 按任务场景聚类训练数据，分组优化记忆提取提示。
- 在跨任务场景下相对提升9.04%，优于现有自进化方法。
- 适合需要持续学习和个性化的大模型应用开发。
随着基于大语言模型的助手变得持久且个性化，它们必须从过往对话中提取并保留有用信息作为记忆。然而，不同任务中值得记忆的信息类型差异显著。本文正式定义了‘异构记忆提取’任务，并提出BEHEMOTH基准，复用18个涵盖个性化、问题求解和代理型任务的现有数据集，采用下游效用驱动的评估指标进行系统性评测。实证分析表明，单一静态提取提示无法在所有任务类别中占优，且原有自进化提示优化框架在异构任务分布下性能下降。为此，本文提出CluE，一种基于聚类的自进化策略：根据提取场景将训练样本分组，独立分析每组，并融合跨组洞察更新提取提示。在BEHEMOTH上的实验显示，CluE在异构任务间具备良好泛化能力，相对提升9.04%，持续优于现有自进化框架。
原文摘要 · Abstract (English)
As LLM-based assistants become persistent and personalized, they must extract and retain useful information from past conversations as memory. However, the types of information worth remembering vary considerably across tasks. We formalize the \textit{heterogeneous memory extraction} task and introduce \textbf{BEHEMOTH}, a benchmark that repurposes 18 existing datasets spanning personalization, problem-solving, and agentic tasks, using a downstream utility-driven metric for systematic evaluation. Our empirical analysis confirms that no single static extraction prompt dominates across all task categories, and that existing self-evolving prompt optimization frameworks, originally designed for homogeneous distributions, degrade when training tasks are heterogeneous. To address this, we propose \textbf{CluE}, a cluster-based self-evolving strategy that groups training examples into clusters by extraction scenarios, analyzes each cluster independently, and synthesizes cross-cluster insights to update the extraction prompt. Experiments on BEHEMOTH show that CluE generalizes effectively across heterogeneous tasks ($+$9.04\% relative gain), consistently outperforming prior self-evolving frameworks.
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。