Alibaba Cloud has entered the competitive world of generative AI by open-sourcing two of its large language models (LLMs), Qwen-7B and Qwen-7B-Chat. Trained on 7 billion parameters each, these models are now freely available to researchers, scholars, and commercial institutions worldwide through Alibaba Cloud’s AI model repository ModelScope and the US collaborative AI platform, Hugging Face.
To encourage broad adoption, companies with less than 100 million monthly active users can deploy these open-source models for commercial use at no cost. This move aims to make generative AI accessible to developers and small to medium-sized enterprises, fostering inclusive technologies.
Zhou Jingren, Chief Technology Officer of Alibaba Cloud Intelligence, emphasized their commitment to democratizing AI access and empowering the research community. Additionally, Alibaba’s ModelScope platform has integrated Llama 2, an open-source LLM created by Meta.
Amidst these developments, Alibaba Cloud is preparing for a spin-off from its parent company, Alibaba Group. The company has been undergoing restructuring and planning separate public listings for its various business groups, leading to some job cuts in its Cloud unit.
In conclusion, Alibaba Cloud’s decision to open-source these AI models signifies a significant step in advancing generative AI and promoting collaboration within the research community.