大型语言模型 (LLM) 多代理系统通常依赖于严格的编排,，要么致力于平坦的每个查询路由，要么致力于手工设计的任务分解,，因此分解深度,工作人员选择,和推理预算不会在一个目标下联合优化。我们引入了 Uno-Orchestra, 一个统一的编排策略，它有选择地分解任务并将每个子任务分派给可接受的 (model, 原语) 对,，这两个决策都是从基于实际工作人员交互的精心策划的 RL 轨迹中共同学习的。与涵盖数学, 代码, 知识, 长上下文, 和代理工具使用, 的 13 个基准套件上的 22 个基线相比，Uno-Orchestra 达到了 77.0% 宏传递@1, 比最强工作流基线, 高出大约 16%，每个查询成本降低了大约一个数量级,，推进了选择性委派的准确性-效率前沿。
Large language model (LLM) multi-agent systems typically rely on rigid orchestration, committing either to flat per-query routing or to hand-engineered task decomposition, so decomposition depth, worker choice, and inference budget are not jointly optimized under one objective. We introduce Uno-Orchestra, a unified orchestration policy that selectively decomposes a task and dispatches each subtask to an admissible (model, primitive) pair, with both decisions learned together from curated RL trajectories grounded in real worker interactions. Against 22 baselines on a 13-benchmark suite spanning math, code, knowledge, long-context, and agentic tool-use, Uno-Orchestra reaches 77.0% macro pass@1, roughly 16% above the strongest workflow baseline, at roughly an order of magnitude lower per-query cost, advancing the accuracy-efficiency frontier of selective delegation.
科目:人工智能(cs.AI)
Subjects: Artificial Intelligence (cs.AI)