Alibaba Cloud open sources two generative AI models for researchers

TL;DR:

  • Alibaba Cloud unveils open-source language models, Qwen-7B and Qwen-7B-Chat, each trained on 7 billion parameters.
  • Models and accompanying resources are available through ModelScope and Hugging Face for global researchers.
  • Companies with fewer than 100 million monthly active users can deploy open-source models for commercial use free of charge.
  • Meta’s Llama 2 integrated into Alibaba’s ModelScope platform.
  • Alibaba Cloud’s dedication to inclusive technologies and empowerment of developers and SMEs.

Main AI News:

In the ever-intensifying race of generative AI advancements, Chinese tech titan Alibaba Cloud has taken a bold step by announcing the open-sourcing of not one but two of its powerful language models (LLMs). The groundbreaking models, named Qwen-7B and Qwen-7B-Chat, have undergone extensive training on a staggering 7 billion parameters, marking a remarkable leap in the field of artificial intelligence.

According to reports from the reputable South China Morning Post, Alibaba Cloud is now making these AI marvels and their accompanying codes and documentation readily accessible to the global community of scholars, researchers, and commercial institutions through their AI model repository ModelScope. Additionally, the collaborative AI platform Hugging Face, a prominent player in the AI research community, will also host these invaluable resources.

This strategic move by Alibaba Cloud aims to empower developers and smaller enterprises, fostering an inclusive approach to cutting-edge technologies. By providing free-of-charge deployment of the open-source models for commercial use to companies with fewer than 100 million monthly active users, Alibaba Cloud seeks to democratize the benefits of generative AI across the industry landscape.

Notably, Alibaba Cloud’s commitment to the research community and the advancement of AI technologies does not end here. In a recent development, Meta’s renowned open-source LLM, Llama 2, has been seamlessly integrated into Alibaba’s ModelScope platform, bolstering the array of resources available to researchers and developers.

This groundbreaking initiative aligns with Alibaba Cloud’s broader vision, as the company is set to be spun off from its parent company Alibaba Group in the upcoming year. The pursuit of inclusive technologies remains at the core of their mission, as Zhou Jingren, the Chief Technology Officer of Alibaba Cloud Intelligence, was quoted as saying, “We aim to promote inclusive technologies and enable more developers and small and medium-sized enterprises to reap the benefits of generative AI.”

While the Chinese internet giant faced recent reports of significant job cuts in its Cloud unit, with around 7 percent of its workforce affected, Alibaba remains undeterred in its commitment to growth and innovation. With plans for separate IPOs for its various business groups, including a restructuring of its Cloud division, Alibaba CEO Daniel Zhang’s strategic vision is clear – to chart a path of progress and excellence in the dynamic world of AI and cloud computing.

Conclusion:

Alibaba Cloud’s decision to open-source its cutting-edge language models signifies a major breakthrough in the generative AI market. By making these models freely accessible to researchers and developers worldwide, Alibaba Cloud is fostering collaboration and driving innovation. The deployment of open-source models for smaller companies without cost barriers will encourage widespread adoption and experimentation with generative AI, potentially leading to groundbreaking applications and advancements across various industries. This move cements Alibaba Cloud’s position as a key player in the AI research community and is likely to shape the future of the market, promoting further democratization and technological progress. Businesses in the AI space should closely monitor these developments and consider leveraging open-source AI resources to stay competitive in an increasingly dynamic market.

Source