通过协作代理机制提升大模型推理性能，且有理论保证。
ALIGN: Aligned Delegation with Performance Guarantees for Multi-Agent LLM Reasoning
- 设计代理协作机制，让多个模型分工生成答案并由主模型选择最优解。
- 在多个基准测试中，性能优于单模型和传统集成方法。
- 首次提供理论证明：该方法能提升推理效果，不依赖独立假设。
大语言模型在复杂推理任务中常因单一生成与筛选流程而表现不佳。现有推理时集成方法虽能通过采样多样推理路径或聚合多个候选答案提升性能，但通常将候选答案视为独立，且无法提供集成必然改善推理质量的正式保证。本文提出一种新方法——多代理大模型推理中的对齐委派（ALIGN），将推理过程建模为一个对齐的委托博弈。在此框架中，主模型向多个代理分配任务，代理在激励机制下生成候选解，主模型从中选出最终答案。该机制促进代理间结构化协作，同时保持代理目标与主目标一致。我们建立了理论保证：在同等访问候选解条件下，ALIGN在期望性能上严格优于单代理生成。分析支持相关候选答案，并放宽了以往工作普遍采用的独立性假设。在广泛的大模型推理基准上的实证结果一致表明，ALIGN显著超越强基线的单代理与集成方法。
原文摘要 · Abstract (English)
LLMs often underperform on complex reasoning tasks when relying on a single generation-and-selection pipeline. Inference-time ensemble methods can improve performance by sampling diverse reasoning paths or aggregating multiple candidate answers, but they typically treat candidates independently and provide no formal guarantees that ensembling improves reasoning quality. We propose a novel method, Aligned Delegation for Multi-Agent LLM Reasoning (ALIGN), which formulates LLM reasoning as an aligned delegation game. In ALIGN, a principal delegates a task to multiple agents that generate candidate solutions under designed incentives, and then selects among their outputs to produce a final answer. This formulation induces structured interaction among agents while preserving alignment between agent and principal objectives. We establish theoretical guarantees showing that, under a fair comparison with equal access to candidate solutions, ALIGN provably improves expected performance over single-agent generation. Our analysis accommodates correlated candidate answers and relaxes independence assumptions that are commonly used in prior work. Empirical results across a broad range of LLM reasoning benchmarks consistently demonstrate that ALIGN outperforms strong single-agent and ensemble baselines.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。