LEMON: Learning Executable Multi-Agent Orchestration via Counterfactual Reinforcement Learning

• URL: http://arxiv.org/abs/2605.14483v1
• Date: Thu, 14 May 2026 07:24:09 GMT
• Title: LEMON: Learning Executable Multi-Agent Orchestration via Counterfactual Reinforcement Learning
• Abstract summary: Large language models (LLMs) have become a strong foundation for multi-agent systems, but their effectiveness depends heavily on orchestration design.<n>We propose LEMON, an orchestrator that generates an executable orchestration specification.<n> Experiments on six reasoning and coding benchmarks, including MMLU, GSM8K, AQuA, MultiArith, SVAMP, and HumanEval, show that LEMON achieves state-of-the-art performance.
• Score: 31.870185345616733
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models (LLMs) have become a strong foundation for multi-agent systems, but their effectiveness depends heavily on orchestration design. Across different tasks, role design, capacity assignment, and dependency construction jointly affect both solution quality and execution efficiency. Existing approaches automate parts of this design process, yet they often optimize these decisions partially or sequentially, and rely on execution-level feedback that provides limited credit assignment for local orchestration decisions. We propose LEMON (\textbf{L}earning \textbf{E}xecutable \textbf{M}ulti-agent \textbf{O}rchestratio\textbf{N} via Counterfactual Reinforcement Learning), an LLM-based orchestrator that generates an executable orchestration specification. The specification integrates task-specific roles, customized duties, capacity levels, and dependency structure into a single deployable system. To train the orchestrator, we augment the orchestration-level GRPO objective with a localized counterfactual signal that edits role, capacity, or dependency fields and applies the resulting reward contrast only to the edited spans. Experiments on six reasoning and coding benchmarks, including MMLU, GSM8K, AQuA, MultiArith, SVAMP, and HumanEval, show that LEMON achieves state-of-the-art performance among the evaluated multi-agent orchestration methods. Our code is available at https://anonymous.4open.science/r/LEMON-B23C.

Related papers

• EASy: Towards Efficient LLM-Based Agentic System [54.10819421625103]
  We propose EASy, a trainable agentic framework that jointly optimize task performance and computational efficiency through reinforcement learning.<n> EASy equips an LLM-based orchestrator with explicit knowledge of the capability and cost profiles of heterogeneous executors, enabling context-sensitive coordination beyond performance-only routing.<n>It further introduces a milestone-plan-act workflow that decomposes complex tasks into manageable milestones, constructs dependency-aware execution graphs, assigns suitable executors, and parallelizes independent steps while adapting subsequent decisions to intermediate outcomes.
  arXiv  Detail & Related papers  (2026-08-05T08:50:36Z)
• Reward Modeling for Multi-Agent Orchestration [43.79233309038303]
  Multi-Agent Systems (MAS) built on Large Language Models (LLMs) require effective orchestration to coordinate specialized agents.<n>We propose Orchestration Reward Modeling (OrchRM), a self-supervised framework for evaluating orchestration quality without human annotations.
  arXiv  Detail & Related papers  (2026-06-11T17:16:24Z)
• Orchestra-o1: Omnimodal Agent Orchestration [72.34907006336108]
  Orchestra-o1 is an omnimodal agent orchestration framework designed to support efficient agent collaboration across multiple modalities.<n>This work introduces a unified orchestration mechanism that enables modality-aware task decomposition, online sub-agent specialization, and parallel sub-task execution.<n>This scalable design allows agent systems to effectively tackle complex real-world tasks involving heterogeneous information sources, surpassing the second-best approach by 10.3% accuracy on the OmniGAIA benchmark.
  arXiv  Detail & Related papers  (2026-06-10T04:50:35Z)
• Training LLMs for Multi-Step Tool Orchestration with Constrained Data Synthesis and Graduated Rewards [76.49428173793386]
  LLMs must invoke multiple dependent APIs in the correct order while propagating intermediate outputs.<n>Training models to handle such faces two obstacles: existing environments focus on simple per-turn function calls with simulated data, and binary rewards.<n>We present a framework addressing both challenges. First, we construct a reinforcement learning environment backed by a large-scale cache of real API responses, enabling a data synthesis pipeline that samples valid multi-step orchestration traces.<n>Second, we propose a graduated reward design that decomposes correctness into atomic validity and orchestration.
  arXiv  Detail & Related papers  (2026-03-25T18:31:39Z)
• MAS-Orchestra: Understanding and Improving Multi-Agent Reasoning Through Holistic Orchestration and Controlled Benchmarks [86.05918381895555]
  We propose MASOrchestra as a training-time framework that formulates MAS orchestration as a function-calling reinforcement learning problem.<n>In MAS-Orchestra, complex, goal-oriented subagents are abstracted as callable functions, enabling global reasoning over system structure.<n>Our analysis reveals that MAS gains depend critically on task structure, verification protocols, and the capabilities of both orchestrator and subagents.
  arXiv  Detail & Related papers  (2026-01-21T04:57:02Z)
• Difficulty-Aware Agentic Orchestration for Query-Specific Multi-Agent Workflows [5.376980683850103]
  We propose Difficulty-Aware Agentic Orchestration (DAAO), which generates query-specific multi-agent guided by predicted difficulty.<n>DAAO surpasses prior multi-agent systems in both accuracy and inference efficiency.
  arXiv  Detail & Related papers  (2025-09-14T03:57:43Z)
• How to Train a Leader: Hierarchical Reasoning in Multi-Agent LLMs [16.853362180877593]
  We introduce a hierarchical multi-agent framework that trains only a single leader LLM to coordinate a team of untrained peer agents.<n>Our results highlight the effectiveness and efficiency of training a single, flexible leader for collaborative reasoning in multi-agent LLM systems.
  arXiv  Detail & Related papers  (2025-07-11T18:34:07Z)
• Multi-Agent Collaboration via Evolving Orchestration [55.574417128944226]
  Large language models (LLMs) have achieved remarkable results across diverse downstream tasks, but their monolithic nature restricts scalability and efficiency in complex problem-solving.<n>We propose a puppeteer-style paradigm for LLM-based multi-agent collaboration, where a centralized orchestrator ("puppeteer") dynamically directs agents ("puppets") in response to evolving task states.<n> Experiments on closed- and open-domain scenarios show that this method achieves superior performance with reduced computational costs.
  arXiv  Detail & Related papers  (2025-05-26T07:02:17Z)
• Multiple Weaks Win Single Strong: Large Language Models Ensemble Weak Reinforcement Learning Agents into a Supreme One [28.264011412168347]
  Model ensemble is a useful approach in reinforcement learning (RL) for training effective agents.<n>We propose LLM-Ens, a novel approach that enhances RL model ensemble with task-specific semantic understandings.
  arXiv  Detail & Related papers  (2025-05-21T09:35:43Z)
• Transformer-Squared: Self-adaptive LLMs [29.1326358746118]
  We introduce Transformer-Squared, a novel self-adaptation framework that adapts large language models for unseen tasks in real-time.<n>Our method consistently outperforms ubiquitous approaches such as LoRA, with fewer parameters and greater efficiency.<n> Transformer-Squared represents a significant leap forward, offering a scalable, efficient solution for enhancing the adaptability and task-specific performance of LLMs.
  arXiv  Detail & Related papers  (2025-01-09T01:19:21Z)
• Small LLMs Are Weak Tool Learners: A Multi-LLM Agent [73.54562551341454]
  Large Language Model (LLM) agents significantly extend the capabilities of standalone LLMs. We propose a novel approach that decomposes the aforementioned capabilities into a planner, caller, and summarizer. This modular framework facilitates individual updates and the potential use of smaller LLMs for building each capability.
  arXiv  Detail & Related papers  (2024-01-14T16:17:07Z)
• A Unified and Efficient Coordinating Framework for Autonomous DBMS Tuning [34.85351481228439]
  We propose a unified coordinating framework to efficiently utilize existing ML-based agents. We show that it can effectively utilize different ML-based agents and find better configurations with 1.414.1X speedups on the workload execution time.
  arXiv  Detail & Related papers  (2023-03-10T05:27:23Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.