August 3, 2023 – Alibaba Cloud, one of China’s leading technology giants, has taken a groundbreaking step in the world of language models. Today, they have open-sourced their cutting-edge models, Qwen-7B and Qwen-7B-Chat, boasting a staggering 7 billion parameters. This move marks a significant milestone as Alibaba Cloud becomes the first major tech company in China to embrace the open-source community with such large-scale language models.
Back in April, Alibaba Cloud unveiled its latest language model, “Qianwen,” which garnered much attention. At that time, Alibaba Cloud’s CTO, Jingren Zhou, expressed their commitment to open up the capabilities of Qianwen to assist businesses in creating their personalized large-scale models. It seems they are true to their word, as today’s announcement showcases their dedication to this vision.
Notably, Alibaba plans to integrate Qianwen into all of its products, leading the charge with platforms like DingTalk and Tmall Genie, which are already undergoing testing. After thorough evaluation and certification, these platforms will officially launch the new functionalities powered by Qianwen.
Looking ahead, Zhou envisions a future where every enterprise hosted on Alibaba Cloud can tap into the full potential of Qianwen while leveraging their own industry knowledge and specific use cases to train customized models. This promises to revolutionize how software and services can be enhanced with large-scale language models, opening up exciting possibilities across various industries.
Reports indicate that several products and over 40 scenarios within DingTalk have already integrated Qianwen. This integration allows users to input natural language queries using a simple slash “/”, generating outputs like PowerPoint presentations, mind maps, data analysis charts, and project boards. Such features are just the tip of the iceberg, hinting at the vast potential of large-scale language models in transforming workplace productivity and user experiences.