Multi-Agent Relay Thinking
Summary
The paper introduces MACR, a relay-based multi-agent framework that dynamically transfers reasoning ownership between agents using confidence-guided triggering and a bidding mechanism, improving accuracy on benchmarks like MathVista (+3.3%) and MMStar (+3.9%) while cutting token overhead by up to 6.0x versus other multi-agent strategies.
View Cached Full Text
Cached at: 11/03/26, 03:36 PM
Paper page - Multi-Agent Relay Thinking
Source: https://huggingface.co/papers/2610.35219 Published on Oct 27
·
Submitted byhttps://huggingface.co/67L11
Xuon Nov 3
Abstract
MACR introduces a relay-based multi-agent framework that dynamically transfers reasoning between agents using confidence-guided triggering and bidding to improve accuracy with lower token overhead.
Existingmulti-agent collaboration(MAC) methods have primarily explored a group thinking collaboration topology, where agents independently generate solutions and aggregate their outputs. In this work, we investigate an alternativerelay thinking topology, wherereasoning ownershipis dynamically transferred across agents, allowing subsequent agents to inherit and refine an evolving reasoning trajectory. We introduce Multi-Agent Confidence-guided Relay (MACR), a framework that enables dynamical relay collaboration among agents. The key challenge is determining when reasoning should be relayed and which agent should continue the process. method addresses this challenge throughconfidence-guided relay triggeringand abidding mechanismthat selects the next agent based on its confidence. Experiments on four benchmarks show that method outperforms the strongest single-model strategies in accuracy, e.g., +3.3% on MathVista and +3.9% on MMStar, and remains on par with multi-agent strategies while consuming 6.0x less token overhead on HallusionBench and MMStar. Further analysis demonstrates that relay collaboration enablesheterogeneous agentsto progressively integrate complementary insights, combining heterogeneous capabilities across reasoning stages. Our source code is available at https://github.com/67L1/MACR.
View arXiv pageView PDFGitHub0Add to collection
Get this paper in your agent:
hf papers read 2610\.35219
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2610.35219 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2610.35219 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2610.35219 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
Recursive Multi-Agent Systems
This paper introduces RecursiveMAS, a framework that extends recursive scaling principles to multi-agent systems for improved collaborative reasoning efficiency and accuracy. It demonstrates significant speedups and token reduction across various benchmarks compared to standard baselines.
TMAS: Scaling Test-Time Compute via Multi-Agent Synergy
TMAS introduces a multi-agent framework that enhances large language model reasoning by scaling test-time compute through structured collaboration and hierarchical memory systems. The approach uses specialized agents, cross-trajectory information flow, and hybrid reward reinforcement learning to improve iterative scaling and stability on challenging reasoning benchmarks.
Multi-Agent Computer Use
This paper proposes a multi-agent computer use (MACU) system that uses a manager model to decompose tasks into directed acyclic graphs for parallel execution by subagents. It demonstrates consistent improvements over single-agent baselines on multiple benchmarks and better test-time scaling.
Remember and Reweight: Enhancing Multi-Agent Debate with Experience Memory and Confidence Estimation
The paper proposes R2-MAD, a framework that enhances multi-agent debate in large language models by using experience memory and confidence estimation to address shared misconceptions, achieving consistent improvements over baselines.
Beyond Memory Majority: Latent-Source Reasoning for Multi-Agent Memory Arbitration
This paper proposes the Correlation-Aware Memory Arbitration (CAMA) framework to address Memory Correlation Bias in multi-agent memory systems by jointly decoupling memories and recovering independent evidence, demonstrating superior performance over baseline methods in experiments.