用四种不同思维模式协作，让大模型更聪明地解难题。
PoTRE: Test-Time Reasoning inspired by Cognitive Heterogeneity
- 设计四种角色分工协作，分别负责纠错、规划、搜索和直觉推理。
- 在人类最难的考试题上达到49.92%准确率，超越此前最佳成绩。
- 比单一模式更高效，用更少计算量实现更强推理能力，适合复杂任务。
尽管大语言模型在众多任务中表现优异，但在需要长期规划与迭代纠错的复杂推理任务中仍常遇瓶颈。此外，标准单流提示在面对新抽象概念或严格领域约束时表现脆弱。本文提出PoTRE（多拓扑推理集成）框架，将推理过程拆分为四个异构代理：(1) 对抗精炼代理，(2) 分层战略规划代理，(3) 谱搜索代理，(4) 直接链式代理。最终通过任务自适应聚合层动态融合这些视角——包括候选选择、语义合成或神经符号验证——生成鲁棒全局解。我们在三个前沿基准测试上评估PoTRE：ARC-AGI-2、人类最后考试（HLE）和PRBench Finance。PoTRE在HLE上取得49.92%的最新最高准确率，超越此前官方最优成绩。实验表明，该异构架构在使用相似或更少推理令牌的情况下，相比大规模同质基线实现了更好的推理性能。
原文摘要 · Abstract (English)
While Large Language Models (LLMs) excel at many tasks, they frequently struggle with complex reasoning that requires long-horizon planning and iterative error correction. Furthermore, standard single-stream prompting proves brittle when models encounter novel abstractions or rigorous domain constraints. We introduce PoTRE (Poly-Topological Reasoning Ensembles), a heterogeneous framework that decouples inference into four agents: (1) Adversarial Refinement Agent, (2) Hierarchical strategic Planning Agent, (3) Spectrum Search Agent, and (4) Direct Chain Agent. A final Task-Adaptive Aggregation Layer dynamically reconciles these perspectives -- via final candidate selection, semantic synthesis, or neuro-symbolic verification -- to produce a robust global solution. We evaluate PoTRE on three frontier benchmarks: ARC-AGI-2, Humanity's Last Exam (HLE), and PRBench Finance. PoTRE achieves state-of-the-art accuracy of 49.92% on HLE, surpassing the previous best official score. We demonstrate that this architectural heterogeneity achieves improved reasoning performance using similar or fewer inference tokens compared to heavily scaled homogeneous 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。