Saved in:
Bibliographic Details
Main Authors: Wang, Yimeng, Zhao, Jiaxing, Xie, Hongbin, Ma, Hexing, Lei, Yuzhen, Liu, Shuangxue, Song, Xuan, Zhang, Zichen, Zhang, Haoran
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.19290
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918308036476928
author Wang, Yimeng
Zhao, Jiaxing
Xie, Hongbin
Ma, Hexing
Lei, Yuzhen
Liu, Shuangxue
Song, Xuan
Zhang, Zichen
Zhang, Haoran
author_facet Wang, Yimeng
Zhao, Jiaxing
Xie, Hongbin
Ma, Hexing
Lei, Yuzhen
Liu, Shuangxue
Song, Xuan
Zhang, Zichen
Zhang, Haoran
contents Large language models are increasingly deployed as multi-agent systems, where specialized roles communicate and collaborate through structured interactions to solve complex tasks that often exceed the capacity of a single agent. However, most existing systems still rely on a fixed role library and an execution-frozen interaction topology, a rigid design choice that frequently leads to task mismatch, prevents timely adaptation when new evidence emerges during reasoning, and further inflates inference cost. We introduce MetaGen, a training-free framework that adapts both the role space and the collaboration topology at inference time, without updating base model weights. MetaGen generates and rewrites query-conditioned role specifications to maintain a controllable dynamic role pool, then instantiates a constrained execution graph around a minimal backbone. During execution, it iteratively updates role prompts and adjusts structural decisions using lightweight feedback signals. Experiments on code generation and multi-step reasoning benchmarks show that MetaGen improves the accuracy and cost tradeoff over strong multi-agent baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2601_19290
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle MetaGen: Self-Evolving Roles and Topologies for Multi-Agent LLM Reasoning
Wang, Yimeng
Zhao, Jiaxing
Xie, Hongbin
Ma, Hexing
Lei, Yuzhen
Liu, Shuangxue
Song, Xuan
Zhang, Zichen
Zhang, Haoran
Computation and Language
Large language models are increasingly deployed as multi-agent systems, where specialized roles communicate and collaborate through structured interactions to solve complex tasks that often exceed the capacity of a single agent. However, most existing systems still rely on a fixed role library and an execution-frozen interaction topology, a rigid design choice that frequently leads to task mismatch, prevents timely adaptation when new evidence emerges during reasoning, and further inflates inference cost. We introduce MetaGen, a training-free framework that adapts both the role space and the collaboration topology at inference time, without updating base model weights. MetaGen generates and rewrites query-conditioned role specifications to maintain a controllable dynamic role pool, then instantiates a constrained execution graph around a minimal backbone. During execution, it iteratively updates role prompts and adjusts structural decisions using lightweight feedback signals. Experiments on code generation and multi-step reasoning benchmarks show that MetaGen improves the accuracy and cost tradeoff over strong multi-agent baselines.
title MetaGen: Self-Evolving Roles and Topologies for Multi-Agent LLM Reasoning
topic Computation and Language
url https://arxiv.org/abs/2601.19290