Command A是专为企业场景设计的多语言大模型，支持23种商业语言。
Command A: An Enterprise-Ready Large Language Model
- 采用混合架构与去中心化训练，提升效率与性能平衡
- 具备顶尖的RAG能力，可支持工具调用与流程自动化
- 适合需要多语言支持的企业级AI应用开发者
本文介绍Command A的开发，这是一个专为真实企业应用场景打造的强大大语言模型。Command A是一个面向智能体优化、支持23种全球商业语言的多语言模型，采用新型混合架构，在效率与顶级性能间取得平衡。其具备业界领先的检索增强生成（RAG）能力，支持事实校准与工具使用，可自动化复杂业务流程。这些能力通过去中心化训练方法实现，包括自精炼算法和模型合并技术。我们还提供了具有相似能力与架构的Command R7B的评估结果。两个模型权重已开放供研究使用。本技术报告详述了原始训练流程，并在一系列企业相关任务与公开基准上进行了广泛评估，证明了模型在性能与效率上的优异表现。
原文摘要 · Abstract (English)
In this report we describe the development of Command A, a powerful large language model purpose-built to excel at real-world enterprise use cases. Command A is an agent-optimised and multilingual-capable model, with support for 23 languages of global business, and a novel hybrid architecture balancing efficiency with top of the range performance. It offers best-in-class Retrieval Augmented Generation (RAG) capabilities with grounding and tool use to automate sophisticated business processes. These abilities are achieved through a decentralised training approach, including self-refinement algorithms and model merging techniques. We also include results for Command R7B which shares capability and architectural similarities to Command A. Weights for both models have been released for research purposes. This technical report details our original training pipeline and presents an extensive evaluation of our models across a suite of enterprise-relevant tasks and public benchmarks, demonstrating excellent performance and efficiency.
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。