中国科学院孵化的中科闻歌公司于今日发布了雅意2.0大模型。这款大模型拥有300亿规模参数，支持128k上下文窗口输入，具备图文多模态能力，十余种语种对话以及智能工具调用。作为国产原生大模型，雅意2.0从头进行预训练，专注于安全可靠的政企大模型，同时开源支持商用。中科闻歌董事长王磊表示，雅意2.0旨在构建安全、金融、媒体、舆情等领域的大模型生态。
Title: Chinese native large model Yayi 2.0 released, expanding multimedia capabilities
Keywords: Yayi 2.0, large model, multimedia
News content:
Chinese Academy of Sciences-incubated company ZKWenGe released the Yayi 2.0 large model today. This large model has 300 billion scale parameters, supports 128k context window input, and possesses multimedia capabilities, more than 10 language dialogues, and intelligent tool calls. As a domestically produced native large model, Yayi 2.0 conducts pre-training from the beginning, focusing on secure and reliable political and enterprise-level large models, while also supporting open-source commercial use. Wang Lei, chairman of ZKWenGe, said that Yayi 2.0 aims to build a large model ecosystem in the fields of security, finance, media, and public opinion.
【来源】https://www.chinastarmarket.cn/detail/1545072
Views: 1