让大模型在不超上下文限制下，通过并行推理实现百万级有效计算。
PaCoRe: Learning to Scale Test-Time Compute with Parallel Coordinated Reasoning
- 采用多轮消息传递机制，实现测试时并行推理
- 80亿模型在HMMT 2025达94.5%，超越GPT-5的93.2%
- 适合需要深度推理与高计算扩展性的任务
我们提出并行协调推理（PaCoRe），一种训练与推理框架，旨在解决当前语言模型无法在固定上下文窗口内大幅扩展测试时计算（TTC）的核心局限。PaCoRe突破传统串行推理范式，通过多轮消息传递架构驱动大规模并行探索。每轮启动多个并行推理轨迹，将结果压缩为上下文受限的消息，并合成这些消息以指导下一轮，最终生成答案。该模型通过大规模、基于结果的强化学习端到端训练，掌握所需合成能力，在不超出上下文限制的前提下，实现数百万有效TTC。该方法在多个领域表现优异，尤其在数学推理上显著领先：80亿参数模型在HMMT 2025上达到94.5%准确率，超越GPT-5的93.2%，其有效TTC扩展至约两百万令牌。我们开源模型检查点、训练数据和完整推理流程，以加速后续研究。
原文摘要 · Abstract (English)
We introduce Parallel Coordinated Reasoning (PaCoRe), a training-and-inference framework designed to overcome a central limitation of contemporary language models: their inability to scale test-time compute (TTC) far beyond sequential reasoning under a fixed context window. PaCoRe departs from the traditional sequential paradigm by driving TTC through massive parallel exploration coordinated via a message-passing architecture in multiple rounds. Each round launches many parallel reasoning trajectories, compacts their findings into context-bounded messages, and synthesizes these messages to guide the next round and ultimately produce the final answer. Trained end-to-end with large-scale, outcome-based reinforcement learning, the model masters the synthesis abilities required by PaCoRe and scales to multi-million-token effective TTC without exceeding context limits. The approach yields strong improvements across diverse domains, and notably pushes reasoning beyond frontier systems in mathematics: an 8B model reaches 94.5% on HMMT 2025, surpassing GPT-5's 93.2% by scaling effective TTC to roughly two million tokens. We open-source model checkpoints, training data, and the full inference pipeline to accelerate follow-up work.
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。