发表机构
Monash University（莫纳什大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
EASy是一种可训练的智能体框架，通过强化学习优化任务性能与计算效率，采用里程碑-规划-执行工作流，在多类基准上实现了更优的性能-效率权衡。
AI 中文摘要
智能体系统作为一种有前景的范式，通过协调专门的基于大语言模型（LLM）的智能体来解决复杂任务。然而，现有多数系统主要优化任务成功率，却较少考虑执行器能力、计算成本等实际约束下的执行效率。现有基于路由器的方法在对丰富、动态变化的任务上下文、多步骤依赖关系及中间执行反馈进行推理时能力有限，且对未见过的执行器泛化能力较差。我们提出EASy，这是一种可训练的智能体框架，通过强化学习同时优化任务性能与计算效率。EASy为基于LLM的协调器配备异构执行器的能力与成本概况的显式知识，使其能实现超越仅基于性能的路由的上下文感知协调。它还引入里程碑-规划-执行工作流，将复杂任务分解为可管理的里程碑，构建感知依赖关系的执行图，分配合适的执行器，并行化独立步骤，并根据中间结果调整后续决策。为训练协调器，我们开发树状展开流程，探索替代的里程碑分解与执行计划，同时采用多组件奖励，涵盖任务正确性、执行效率与轨迹完整性。在数学推理、具身决策及深度研究基准上开展的大量实验表明，EASy相较于强大的智能体基线，始终能实现更优的性能-效率权衡。
英文摘要
Agentic orchestration enables multiple autonomous agents to solve complex tasks through adaptive decomposition, delegation, and execution. However, existing orchestrators often rely on hand-crafted logic and prompting strategies, limiting adaptation and generalization across tasks and executor configurations. We propose E$^3$-Orch, a reinforcement learning framework for effective, efficient, and extensible agentic orchestration based on a milestone-plan-act workflow. Instead of planning all subtasks upfront, E$^3$-Orch organizes execution around milestones, a mid-level abstraction that scopes planning around meaningful intermediate objectives and allows orchestration decisions to adapt as execution progresses. For each milestone, the orchestrator builds a dependency-aware plan, assigns subtasks to suitable executors, and executes independent subtasks in parallel. We train the orchestration policy from execution feedback, using milestone and plan decisions as units for fine-grained credit assignment. Tree-structured rollouts compare alternative decisions under shared execution histories, while complementary rewards optimize task performance, execution cost, and planning completeness, including an uncertainty-aware performance reward for stochastic downstream outcomes. Across seven benchmarks, E$^3$-Orch achieves the best task performance under multiple executor configurations, improving over the strongest baselines by $0.7$--$3.8$ points and delivering $1.16$--$1.59\times$ higher intelligence efficiency. The learned policy also transfers to unseen executor configurations introduced only at evaluation time, supporting extensible agentic orchestration.
CommentsWork in progress
