LIBRA:程序化构建可执行工具库
原标题:LIBRA: Programmatic Induction of Executable Tool Libraries
AI 摘要
论文提出 LIBRA 框架,将扁平的 LLM 工具目录转化为可执行工具库:由 LLM 为每个工具提出调用契约、可复用输出与状态变化、成功条件,并在沙箱中测试工具组合,仅保留运行时验证通过的关系。该库在用户查询前构建,无需依赖标注、使用轨迹或任务训练,可同时支持一次性检索和迭代式智能体检索。在 AppWorld、BFCL、SciToolEval 上,相比同一稠密检索器,一次性 Recall 和 NDCG 平均提升 7.9 和 3.7 个百分点;集成到 ToolReAGt 后,经十轮检索迭代,分别提升 4.5 和 2.9 个百分点。
正文节选
LIBRA: Programmatic Induction of Executable Tool Libraries Abstract Large language model agents increasingly retrieve tools from large catalogs, yet most retrievers rank each tool independently by request similarity. For multi tool tasks, this recovers the goal matching tool while omitting prerequisite tools required by the workflow. We introduce LIBRA, a framework for Library Induction by Runtime Assembly that turns a flat tool catalog into an executable tool library. An LLM proposes per tool c