用动态检索的解题示例增强推理过程，提升大模型测试时计算的效率与准确性。
ThinkRetrieve: Retrieval-Augmented Reasoning Traces for Test-Time Scaling
- 在每步推理中检索相似解题例证并注入思考链，引导模型合理推理。
- 在AIME 2025上相对提升高达60%，多数任务中优于传统延长推理路径方法。
- 适合需要高精度推理的数学和科学问题求解场景，尤其适用于小参数模型。
大型推理模型（LRMs）通过分配额外的推理时计算来生成更长的思维链以提升性能。然而，近期研究表明，顺序式测试时扩展常导致收益递减甚至负面效果，因更长的推理链会引入更多不确定性、错误累积和偏离原问题。我们提出ThinkRetrieve，一种测试时扩展框架，该框架在每步推理中动态检索外部语料库中已解决的问题及其分步解答，并将相关例证直接注入推理链中，为模型提供如何推理的指导，而不仅仅是哪些事实相关。在GSM-8K、MATH-500、AIME 2025和SciQ上对五种推理模型（1.5B–8B参数）的实验表明，ThinkRetrieve在多数任务中持续提升准确率，其中在AIME 2025上相对提升最高达60%。
原文摘要 · Abstract (English)
Large Reasoning Models (LRMs) improve performance by allocating additional inference-time compute to generate extended chain-of-thought reasoning. However, recent studies reveal that sequential test-time scaling often yields diminishing or even negative returns, as longer traces exhibit increased uncertainty, error compounding, and drift from the original problem. We propose ThinkRetrieve, a test-time scaling framework that augments the reasoning traces of LRMs with dynamically retrieved solved examples at each reasoning step. Given an external corpus of problems paired with step-by-step solutions, ThinkRetrieve retrieves relevant exemplars at each intermediate step and injects them directly into the thinking trace, providing the model with guidance on how to reason rather than merely what facts are relevant. Experiments across five reasoning models (1.5B--8B parameters) on GSM-8K, MATH-500, AIME 2025, and SciQ demonstrate that ThinkRetrieve consistently improves accuracy over standard test-time scaling, with relative gains of up to $60\%$ on AIME 2025.
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。