Test-time scaling has emerged as an effective way to improve language models on challenging reasoning tasks. However, most existing methods treat each problem in isolation and do not systematically reuse knowledge from prior reasoning trajectories. In particular, they underutilize procedural knowledge: how to reframe a problem, choose an approach, and verify or backtrack when needed. We introduce \textbf{Reasoning Memory}, a retrieval-augmented generation (RAG) framework for reasoning models that explicitly retrieves and reuses procedural knowledge at scale. Starting from existing corpora of step-by-step reasoning trajectories, we decompose each trajectory into self-contained subquestion-subroutine pairs, yielding a datastore of 32 million compact procedural knowledge entries. At inference time, a lightweight in-thought prompt lets the model verbalize the core subquestion, retrieve relevant subroutines within its reasoning trace, and reason under diverse retrieved subroutines as implicit procedural priors. Across six math, science, and coding benchmarks, Reasoning Memory consistently outperforms RAG with document, trajectory, and template knowledge, as well as a compute-matched test-time scaling baseline. With a higher inference budget, averaged across models and tasks, it improves over no retrieval by 19.5% and over the compute-matched baseline by 8.9%. Ablation studies show that these gains come from two key factors: the broad procedural coverage of the source trajectories and our decomposition and retrieval design, which together enable effective extraction and reuse of procedural knowledge. Our experiment code is available at https://github.com/facebookresearch/reasoning-memory.
AutoPRM: Automating Procedural Supervision for Multi-Step Reasoning via Controllable Question Decomposition
University of Chicago(芝加哥大学);Peking University(北京大学);University of California, Berkeley(加州大学伯克利分校);Carnegie Mellon University(卡内基梅隆大学);UNC-Chapel Hill(北卡罗来纳大学教堂山分校)
Chain-of-Procedure: Hierarchical Visual-Language Reasoning for Procedural QA
NLP;CT Lab, Department of Computer and Information Science, University of Macau(计算机与信息科学系,澳门大学CT实验室);Department of Electrical and Computer Engineering, University of Macau(电气与计算机工程系,澳门大学);Centre for Cognitive and Brain Sciences, University of Macau(认知与脑科学中心,澳门大学)
Beyond Factual Knowledge: Benchmarking and Learning Step-Level Procedural Rule Reasoning in Large Language Models
University of Chinese Academy of Sciences(中国科学院大学);Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所);Academy of Mathematics and Systems Science, Chinese Academy of Sciences(中国科学院数学与系统科学研究院)
Transferring Procedural Knowledge across Commonsense Tasks
Information Sciences Institute(信息科学研究所);Viterbi School of Engineering(维特比工程学院);University of Southern California(南加州大学);Department of Computer Science(计算机科学系);Faculty of Science(理学院);Vrije Universiteit Amsterdam(阿姆斯特丹自由大学);Tencent AI Lab(腾讯AI实验室)
Meta-Cognitive Analysis: Evaluating Declarative and Procedural Knowledge in Datasets and Large Language Models
University of Chinese Academy of Sciences(中国科学院大学);Chinese Information Processing Laboratory(中文信息处理实验室);State Key Laboratory of Computer Science(计算机科学国家重点实验室);Institute of Software, Chinese Academy of Sciences(中国科学院软件研究所)
A Benchmark for Structured Procedural Knowledge Extraction from Cooking Videos
Carnegie Mellon University(卡内基梅隆大学);Microsoft Research Asia(微软亚洲研究院);Beijing Institute of Technology(北京理工大学);University of Southern California(南加利福尼亚大学)
BioProBench: A Corpus and Benchmark for Biological Protocol Reasoning in Autonomous Science
School of Electronic and Computer Engineering, Peking University(北京大学电子与计算机工程学院);School of Chemical Biology&Biotechnology, Peking University(北京大学化学生物学与生物技术学院)
PlaSma: Making Small Language Models Better Procedural Knowledge Models for (Counterfactual) Planning
Allen Institute for Artificial Intelligence(艾伦人工智能研究所);University of Washington(华盛顿大学);University of Southern California(南加州大学);Tohoku University(东北大学)
MoReBench: Evaluating Procedural and Pluralistic Moral Reasoning in Language Models, More than Outcomes
University of Washington(华盛顿大学);New York University(纽约大学);Scale AI;Harvard University(哈佛大学);University of Michigan(密歇根大学);UNC Chapel Hill(北卡罗来纳大学教堂山分校);Center for AI Safety(人工智能安全中心);Stanford University(斯坦福大学);MIT(麻省理工学院);University of Oxford(牛津大学)