The models, Qwen-7B, and Qwen-7B-Chat are equipped with 7 billion parameters each, making them valuable assets in the field of generative artificial intelligence.
Describing the two models, Alibaba Cloud said that Qwen-7B-Chat, which is a fine-tuned version of Qwen-7B, is capable of engaging in human-like conversations.
It added that the release of these open-source models aims to democratize AI technology and support LLM start-ups.
Furthermore, the cloud computing firms said that its AI model repository, ModelScope, and the collaborative AI platform, Hugging Face, will provide free access to the models’ code, documentation, and internal mechanisms for scholars, researchers, and commercial institutions worldwide.
Alibaba Cloud said that companies with less than 100 million monthly active users can freely deploy the models, larger companies require a license from Alibaba Cloud.
The launch by Alibaba Cloud is coming a few weeks after Meta released its Llama 2 model, which is accessible to developers in China via Alibaba’s ModelScope platform.
Alibaba and AI developments
Alibaba Cloud, as a computing firm, has been intensifying its efforts in generative AI in a drive to meet up with the fast pace of operations of OpenAI, owners of ChatGPT.
This effort led to the launch of Tongyi Wanxiang, a Midjourney-style image-generating model, at the World Artificial Intelligence Conference in Shanghai.
In addition, it introduced ModelScopeGPT, a versatile framework aimed at helping users tackle complex AI tasks in language, vision, and speech domains.
By utilizing various AI models available on ModelScope, an open-source Model-as-a-Service (MaaS) platform with over 900 AI models, users can access comprehensive support for specialized AI applications.
This initiative highlights Alibaba’s commitment to advancing AI technology and providing developers with accessible tools to achieve their AI goals.