GraSP:用于LLM代理的图结构技能组合
GraSP: Graph-Structured Skill Compositions for LLM Agents
AI总结:
本文提出GraSP,一种可执行技能图架构,通过引入编译层提升LLM代理技能编排效率,通过五种类型操作符将扁平技能集转化为带前提-效应边的有向无环图,提升任务完成效率和奖励。
摘要:
LLM代理的技能生态系统发展迅速,但最新基准表明,提供更多技能并不总是提升性能——聚焦的2-3个技能优于全面文档,过多技能反而有害。瓶颈从技能可用性转向技能编排:代理不需要更多技能,而是需要结构化机制来选择、组合和执行技能,明确因果依赖。我们提出GraSP,首个可执行技能图架构,在技能检索与执行之间引入编译层。GraSP将扁平技能集转化为带前提-效应边的有向无环图(DAG),通过节点级验证执行,并通过五种类型操作符进行局部受限修复,将重计划复杂度从O(N)降至O(d^h)。在ALFWorld、ScienceWorld、WebShop和InterCode八个LLM基础架构上,GraSP在所有配置中均优于ReAct、Reflexion、ExpeL和扁平技能基线,在最强基线上提升奖励达+19点,同时减少环境步骤达41%。GraSP的优势随任务复杂性增加而增强,对技能过检索和质量退化具有鲁棒性,证实结构化编排——而非更大技能库——是可靠代理执行的关键。
英文摘要:
Skill ecosystems for LLM agents have matured rapidly, yet recent benchmarks show that providing agents with more skills does not monotonically improve performance -- focused sets of 2-3 skills outperform comprehensive documentation, and excessive skills actually hurt. The bottleneck has shifted from skill availability to skill orchestration: agents need not more skills, but a structural mechanism to select, compose, and execute them with explicit causal dependencies. We propose GraSP, the first executable skill graph architecture that introduces a compilation layer between skill retrieval and execution. GraSP transforms flat skill sets into typed directed acyclic graphs (DAGs) with precondition-effect edges, executes them with node-level verification, and performs locality-bounded repair through five typed operators -- reducing replanning from O(N) to O(d^h). Across ALFWorld, ScienceWorld, WebShop, and InterCode with eight LLM backbones, GraSP outperforms ReAct, Reflexion, ExpeL, and flat skill baselines in every configuration, improving reward by up to +19 points over the strongest baseline while cutting environment steps by up to 41%. GraSP's advantage grows with task complexity and is robust to both skill over-retrieval and quality degradation, confirming that structured orchestration -- not larger skill libraries -- is the key to reliable agent execution.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。