智源研究院近日开源了一款拥有700亿参数的大模型Aquila2-70B-Expr（异构先驱版）。这款大模型是我国首个基于英伟达混合资源（A100集群+A800集群）和天数智芯混合资源（BI-V100集群+BI-V150集群）完成训练的成果。这一举措标志着我国在人工智能领域的研究取得了重要突破。
Aquila2-70B-Expr大模型的开源将有助于我国科研人员更深入地研究和开发人工智能技术，推动我国人工智能产业的快速发展。这款大模型具有极高的计算能力和广泛的应用潜力，可用于自然语言处理、计算机视觉、智能推荐等多个领域。
此次开源不仅展示了智源研究院在人工智能领域的技术实力，也体现了我国在高科技领域的持续投入和创新能力。英伟达和天数智芯的混合资源为Aquila2-70B-Expr大模型的训练提供了强大的支持，进一步提升了我国人工智能技术的竞争力。
英文翻译：
News Title: China’s Zhiyuan Research Institute Releases 700 Million Parameter Large Model
Keywords: Zhiyuan Research Institute, Open Source, Large Model
News Content:
The Zhiyuan Research Institute has recently opened up a large model with 700 million parameters, known as Aquila2-70B-Expr (Heterogeneous Pioneer Edition). This large model is the first in China to be trained using mixed resources from Nvidia (A100 cluster + A800 cluster) and Tianju ZhiXin mixed resources (BI-V100 cluster + BI-V150 cluster). This achievement marks a significant breakthrough in AI research in China.
The open-source of Aquila2-70B-Expr will contribute to deeper research and development of AI technology in China, and promote the rapid development of China’s AI industry. The large model has great computational capabilities and wide application potential, and can be used in various fields such as natural language processing, computer vision, and intelligent recommendation.
This move not only demonstrates the technical strength of the Zhiyuan Research Institute in the field of artificial intelligence but also reflects the continuous investment and innovation capabilities of China in the field of high-tech. The mixed resources of Nvidia and Tianju ZhiXin provide strong support for the training of Aquila2-70B-Expr, further enhancing China’s competitiveness in AI technology.
【来源】https://www.chinastarmarket.cn/detail/1531570
Views: 1