自进化大模型会遗忘旧能力，新方法可有效防止能力退化。
Do Self-Evolving Agents Forget? Capability Degradation and Preservation in Lifelong LLM Agent Adaptation
- 提出能力保留演化框架，约束持续适应中的能力漂移。
- 在工作流演化中，简单任务性能从41.8%提升至52.8%。
- 适合构建长期稳定运行的自主智能体系统。
近期大模型智能体技术使系统能够自主优化流程、积累可复用技能、自我训练底层模型并维持持久记忆。然而，我们发现这种自进化往往非单调：适应新任务分布会逐步削弱已习得的各项能力。我们称此现象为‘自进化下的能力侵蚀’，并在流程、技能、模型和记忆四个演化维度上均观察到该现象。为此，我们提出‘能力保留演化’（CPE）机制，一种通用的稳定化原则，用于抑制持续适应中的破坏性能力漂移。在所有四个演化维度上，CPE均显著提升能力保留稳定性，同时保持适应性能。例如，在工作流演化中，使用GPT-5.1优化时，简单任务性能从41.8%提升至52.8%，同时实现更强的复杂任务适应能力。研究表明，稳定长周期自进化智能体不仅需要学习新能力，还需在持续适应中显式保护已有能力。
原文摘要 · Abstract (English)
Recent advances in LLM agents enable systems that autonomously refine workflows, accumulate reusable skills, self-train their underlying models, and maintain persistent memory. However, we show that such self-evolution is often non-monotonic: adapting to new task distributions can progressively degrade previously acquired capabilities across all major evolution channels. We identify this phenomenon as \emph{capability erosion under self-evolution} and show that it consistently emerges across workflow, skill, model, and memory evolution. To mitigate this issue, we propose \emph{Capability-Preserving Evolution} (CPE), a general stabilization principle that constrains destructive capability drift during continual adaptation. Across all four evolution dimensions, CPE consistently improves retained capability stability while preserving adaptation performance. For example, in workflow evolution, CPE improves retained simple-task performance from 41.8\% to 52.8\% under GPT-5.1 optimization while simultaneously achieving stronger complex-task adaptation. Our findings suggest that stable long-horizon self-evolving agents require not only acquiring new capabilities, but also explicitly preserving previously learned ones during continual adaptation.
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。