arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2512.16144cs.LGcs.AI
收藏

INTELLECT-3:技术报告

INTELLECT-3: Technical Report

Prime Intellect, Inc.(Prime Intellect公司)
Prime Intellect Team, Mika Senghaas, Fares Obeid, Sami Jaghouar, William Brown, Jack Min Ong, Daniel Auras, Matej Sirovatka, Jannik Straube, Andrew Baker, Sebas… 展开作者

Prime Intellect Team, Mika Senghaas, Fares Obeid, Sami Jaghouar, William Brown, Jack Min Ong, Daniel Auras, Matej Sirovatka, Jannik Straube, Andrew Baker, Sebastian Müller, Justus Mattern, Manveer Basra, Aiman Ismail, Dominik Scherm, Cooper Miller, Ameen Patel, Simon Kirsten, Mario Sieg, Christian Reetz, Kemal Erdem, Vincent Weisser, Johannes Hagemann

AI总结:

INTELLECT-3是一款通过大规模强化学习训练的混合专家模型,开源了模型和基础设施栈,并引入了prime-rl框架以支持大规模异步强化学习。

摘要:

我们介绍了INTELLECT-3,这是一个具有10600亿参数的混合专家模型(1200亿个活跃参数),该模型在我们的端到端强化学习基础设施栈上通过大规模强化学习进行训练。INTELLECT-3在数学、代码、科学和推理基准测试中实现了其规模的最新性能,超越了许多更大的前沿模型。我们开源了该模型以及用于创建它的完整基础设施栈,包括强化学习框架、完整的配方以及广泛收集的环境,这些环境是使用验证器库构建的,用于从我们的环境枢纽社区平台进行训练和评估。为此次努力构建的,我们引入了prime-rl,一个专为大规模异步强化学习设计的开源框架,它可以无缝扩展从单个节点到数千个GPU,且专门针对代理强化学习,提供一流的多轮交互和工具使用的支持。使用此栈,我们在GLM-4.5-Air-Base模型之上运行了SFT和RL训练,将RL训练扩展到512个H200芯片,具有高训练效率。

英文摘要:

We present INTELLECT-3, a 106B-parameter Mixture-of-Experts model (12B active) trained with large-scale reinforcement learning on our end-to-end RL infrastructure stack. INTELLECT-3 achieves state of the art performance for its size across math, code, science and reasoning benchmarks, outperforming many larger frontier models. We open-source the model together with the full infrastructure stack used to create it, including RL frameworks, complete recipe, and a wide collection of environments, built with the verifiers library, for training and evaluation from our Environments Hub community platform. Built for this effort, we introduce prime-rl, an open framework for large-scale asynchronous reinforcement learning, which scales seamlessly from a single node to thousands of GPUs, and is tailored for agentic RL with first-class support for multi-turn interactions and tool use. Using this stack, we run both SFT and RL training on top of the GLM-4.5-Air-Base model, scaling RL training up to 512 H200s with high training efficiency.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑