Response-Conditioned Parallel-to-Sequential Orchestration for Multi-Agent Systems
Quick Answer
The Nexa framework introduces a hybrid response-conditioned policy for multi-agent systems, combining parallel and sequential orchestration to enhance communication efficiency and response accuracy.
Quick Take
It utilizes a lightweight transformer model to create a directed acyclic communication graph, allowing for flexible agent interactions without external judges. The approach demonstrates generalizability across varying tasks and agent configurations.
Key Points
- Nexa combines parallel execution with sequential message propagation for improved agent collaboration.
- The framework avoids external judges and hand-crafted topology searches.
- Results show that learned communication policies are reusable across different tasks and agent numbers.
- The hybrid model is based on a lightweight transformer architecture.
- The directed acyclic graph ensures efficient communication without unnecessary latency.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Multi-agent systems can solve complex tasks through collaboration between multiple Large Language Model agents. Existing collaboration frameworks typically operate in either a parallel or a sequential mode. In the parallel mode, agents respond independently to queries followed by aggregation of responses. In contrast, sequential systems allow agents to communicate via a directed topology and refine one another step by step. However, both modes are inadequate for achieving the desired objectives of minimizing communication and latency while simultaneously maximizing the accuracy of the final response. In this work, we introduce a hybrid paradigm called Nexa, a trainable response-conditioned policy that bridges the gap between the two modes. Nexa begins with a parallel execution stage, embeds the resulting responses into a shared semantic space, and then predicts a sparse directed acyclic communication graph. If the graph is empty, the system remains purely parallel; if it is non-empty, the system performs one sequential message propagation. The policy is a lightweight transformer model, and the method avoids the need for external LLM judges or reward models, as well as hand-crafted test-time topology search. We formalize this hybrid execution problem, show that the resulting graph is acyclic by construction, and that the framework strictly subsumes pure parallel execution, and present a training procedure based on policy-gradient optimization. Results demonstrate that the response-conditioned policy learned by Nexa under one setting can be reused when the number of agents, the task, or the underlying agent changes, thus emphasizing the generalizability of the learned communication policy.
| Subjects: | Computation and Language (cs.CL); Machine Learning (cs.LG); Multiagent Systems (cs.MA) |
| Cite as: | arXiv:2605.15573 [cs.CL] |
| (or arXiv:2605.15573v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2605.15573 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Nurbek Tastan [view email]
[v1]
Fri, 15 May 2026 03:33:20 UTC (116 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CL
See more →TriAgent: Divergence-Aware Committees for Cost-Efficient Financial Sentiment Analysis
TriAgent introduces a cost-efficient multi-agent system for financial sentiment analysis, combining VADER, FinBERT, and Qwen2.5. It achieves an F1 score of ~0.87 with significant savings of $9.3M/year at a 10M-user scale compared to GPT-4o-mini, while also detecting hallucinations with an AUC of 0.90.