让大模型像人一样边推理边自检，提升准确性和可解释性。
Stepwise Think-Critique: A Unified Framework for Robust and Interpretable LLM Reasoning
- 在每一步推理中同步进行自我批判，形成闭环思考流程。
- 数学推理任务上准确率显著优于传统分步方法，且推理过程更清晰。
- 适合需要高可靠性与透明度的复杂决策场景，如科研与医疗辅助。
人类解决复杂问题依赖于批判性思维，即推理与评估相互交织以逼近正确答案。然而，现有大语言模型通常将推理与验证分离：要么生成推理但无即时自检，要么依赖外部验证器事后纠错。前者缺乏反馈，后者增加系统复杂度并阻碍协同学习。受人类批判性思维启发，我们提出一步式思辨-批判（Stepwise Think-Critique, STC）框架，一种统一的端到端可训练机制，在单个模型中于每个中间步骤交替执行推理与自我批判。STC采用混合强化学习目标，融合推理奖励与批判一致性奖励，联合优化解题正确性与自我评估可靠性。在数学推理基准上的实验表明，STC展现出强大的批判性思维能力，生成更具可解释性的推理轨迹，标志着大模型向内嵌批判性思维迈进一步。
原文摘要 · Abstract (English)
Human beings solve complex problems through critical thinking, where reasoning and evaluation are intertwined to converge toward correct solutions. However, most existing large language models (LLMs) treat the reasoning and verification as separate processes: they either generate reasoning without explicit self-checking or rely on external verifiers to detect errors post hoc. The former lacks immediate feedback, while the latter increases system complexity and hinders synchronized learning. Motivated by human critical thinking, we propose Stepwise Think-Critique (STC), a unified and end-to-end trainable framework that interleaves reasoning and self-critique at every intermediate step within a single model. STC is trained with a hybrid reinforcement learning objective that integrates reasoning rewards and critique-consistency rewards, thereby jointly optimizing solution correctness and reliability of self-evaluation. Experiments on mathematical reasoning benchmarks show that STC demonstrates strong critical-thinking capabilities and produces more interpretable reasoning traces, representing a step toward LLMs with built-in critical thinking.
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。