August 7, 2023
By Sharan Kaur Phillora
In an advancement that marks a significant stride in the AI landscape, Alibaba Group, China’s tech and e-commerce powerhouse, declared the introduction of two open-sourced artificial intelligence models, Qwen-7B and Qwen-7B-Chat, on August 3rd. These models, each comprising 7 billion parameters, are trimmed versions of the Tongyi Qiawen, which Alibaba launched earlier in April.
Here’s what we know:
Alibaba’s latest release aims to democratize AI for small and medium-sized enterprises (SMEs), allowing them to leverage advanced AI technology in their operations. Various features of Qwen-7B and Qwen-7B-Chat are poised to be attractive to businesses, including code, model weights, and documentation, which will be freely available to scholars, researchers, and commercial entities globally.
Notably, these large language models (LLMs) are the first to be open-sourced by a Chinese tech corporation. A caveat is that businesses exceeding 100 million monthly active users must obtain a license.
This announcement comes on the heels of Alibaba’s recent enhancements to its AnalyticDB data warehousing service, via an update incorporating a vector engine. This allows corporate clients to expeditiously craft custom generative AI applications, thereby fortifying their technological capabilities.
The timing of the release follows Meta’s unveiling of its open-sourced LLM, Llama 2, in collaboration with Microsoft on July 16th. With 70 billion parameters, Llama 2 is claimed to have been trained on 40% more public data, with double the processing context of its prior version. Much like Alibaba’s recent models, Llama 2 mandates licensing for companies with over 700 million monthly users.
Additionally, on July 26th, Alibaba announced its creation of the first training and deployment solution in China for the entire Llama2 series, further emphasizing its determination to be a leader in AI-powered solutions.
Alibaba’s latest move signals an escalating competition in the world of AI and particularly showcases China’s growing influence in the AI community. The open-sourcing of these models is a monumental step towards fostering collaboration and innovation, all while providing a new set of tools for enterprises and researchers.
About the author
Sharan Kaur Phillora’s thirst for knowledge has led her to study many different subjects, including NFTs and Blockchain technology – two emerging technologies that will change how we interact with each other in the future. When she isn’t exploring a new idea or concept, she enjoys reading literary masterpieces.
