[论文] Laguna M.1/XS.2 Technical Report

论文概要

研究领域: AI
作者: Julien Abadji, Marah Abdin, Connor Adams, et al.
发布时间: 2026-05-28
arXiv: 2605.27605

中文摘要

本文介绍了Laguna M.1和Laguna XS.2--两个专为长程智能体编码构建的混合专家基础模型:M.1拥有2258亿总参数量(每token激活234亿),XS.2拥有334亿总参数量(每token激活30亿)。两个模型均在称为"模型工厂"的内部系统中从零开始端到端训练,这是一个将模型开发转化为工业流程的紧密集成的版本化数据、训练、评估和推理组件栈。论文详细描述了模型工厂的设计原则和选择,以及模型端到端训练过程的细节,涵盖预训练数据与架构、后训练阶段、评估和量化。在智能体软件工程和终端基准(SWE-bench Verified、SWE-bench Multilingual、SWE-Bench Pro和Terminal-Bench 2.0)上,M.1和XS.2在各自权重级别中与最先进的开源模型具有竞争力。Laguna XS.2权重已在Apache 2.0许可证下于Hugging Face开源。

原文摘要

We present Laguna M.1 and Laguna XS.2, two Mixture-of-Experts foundation models built for long-horizon, agentic coding: M.1 has 225.8B total parameters (23.4B activated per token) and XS.2 has 33.4B total (3B activated). Both models were trained from scratch end-to-end inside the same internal system that we refer to as our Model Factory: a tightly-integrated stack of versioned data, training, evaluation, and inference components that turn model development into an industrial process. We describe the principles and design choices of the Model Factory and also detail the end-to-end training process of our models, throughout pre-training data and architecture, post-training stages, evaluation, and quantization. On agentic software engineering and terminal benchmarks (SWE-bench Verified, SWE-bench Multilingual, SWE-Bench Pro, and Terminal-Bench 2.0) M.1 and XS.2 are competitive with state-o...

自动采集于 2026-05-29

#论文 #arXiv #AI #小凯