让大模型智能体持续积累经验，自动分类并优化策略。
Unified Context Evolution for LLM Agents
- 将经验拆分为记忆、策略、流程、技能四类，动态更新
- 在两个任务中成功率分别提升至96.3%和61.3%
- 可跨模型复用，无需重新训练
基于大模型的智能体通过结合推理与环境反馈解决多步交互任务，但每轮任务均从固定上下文开始，过程中发现的有效策略在任务结束后即丢失。现有方法或仅限于当前任务学习，或把所有经验存入单一无类型存储，无法区分知识类型、追踪使用质量，也难平衡知识缺口。我们提出统一上下文演化（UCE），一种无梯度框架，将智能体经验外化为可演化的类型化上下文单元（ECUs）库。UCE将经验分解为四类互补类型：记忆、策略、流程、技能，每类由特定条件下的轨迹生成，在决策时检索，通过重复使用结果评分，并在不再有价值时剔除。调度模块根据库中薄弱类型分配生成预算。在两个交互基准上，UCE将ALFWorld成功率从75.4%提升至96.3%，WebShop任务得分从45.1%提升至61.3%，且积累的库可直接迁移至其他智能体主干，无需重训。
原文摘要 · Abstract (English)
LLM-based agents can solve multi-step interactive tasks by combining reasoning with environment feedback, yet each episode starts from the same fixed context and any useful strategy discovered along the way is lost once the task ends. Existing approaches either limit learning to the current task or pool all experience into a single untyped store, without distinguishing knowledge types, tracking quality through use, or balancing what the library still lacks. We introduce Unified Context Evolution (UCE), a gradient-free framework that externalizes agent experience into an evolving library of typed Evolvable Context Units (ECUs). UCE decomposes experience into four complementary types (Memory, Strategy, Workflow, and Skill), each generated from trajectories under type-specific conditions, retrieved at decision time, scored through repeated usage outcomes, and pruned when no longer valuable. A scheduling module allocates each cycle's generation budget toward the types where the library is weakest. Across two interactive benchmarks, UCE raises ALFWorld success from 75.4% to 96.3% and WebShop task score from 45.1% to 61.3%, and the accumulated library transfers to alternative actor backbones without retraining.
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。