AI芯片初创公司SambaNova最近推出了一个名为Samba-1的万亿参数企业级AI模型,这很可能是迄今为止发布的最大的大模型之一。Samba-1不是单一的模型,而是50多个高质量AI模型的组合,通过SambaNova称为“专家组合架构”的方法集成在一起。尽管Samba-1的总体规模很大,但该模型可以针对特定的企业用例进行高度定制和调整。SambaNova的举措表明AI芯片初创公司正在加大投入,推出规模更大、性能更强的企业级AI解决方案。这也预示着AI芯片和模型之间的竞争可能会更加激烈。
Title: AI chip startup launches trillion-parameter enterprise-level AI model
Keywords: AI chips, enterprise-level models, parameter scale
News content: AI chip startup SambaNova recently launched an enterprise-level AI model named Samba-1 with one trillion parameters, which is likely one of the largest models released so far. Rather than a single model, Samba-1 combines over 50 high-quality AI models using SambaNova’s “Composition of Experts” architecture. Although the overall scale of Samba-1 is massive, the model can be highly customized and tuned for specific enterprise use cases. SambaNova’s move signals that AI chip startups are stepping up investments to launch larger-scale, higher-performance enterprise AI solutions. It also foreshadows intensifying competition between AI chips and models.
【来源】https://venturebeat.com/ai/sambanova-debuts-1-trillion-parameter-composition-of-experts-model-for-enterprise-gen-ai/
Views: 1