Alibaba Cloud Unveils New AI Models
- By Paul Mah
- September 25, 2024
At its annual Apsara conference last week, Alibaba Cloud announced the release of its newly-launched large language model, Qwen 2.5, to the global open-source community.
Over 100 models will be made open-source, including base models, instruct models, and quantized models ranging from 0.5 to 72 billion parameters in size – and of different modalities such as language, audio, and vision, along with specialized code and mathematical models.
New AI models
Alibaba Cloud also announced an upgrade to its proprietary flagship model Qwen-Max. In a comparison chart furnished by Alibaba Cloud, the enhanced Qwen-Max model was shown on par with GPT-4o in areas such as language comprehension and reasoning, math, and coding.
The Qwen model has achieved substantial traction since its debut in April 2023 and has surpassed 40 million downloads across platforms such as Hugging Face and ModelScope, an open-source community initiative by Alibaba. These models have inspired the creation of over 50,000 models on Hugging Face.
“Today marks a significant milestone as we launch our most expansive open-source initiative to date. This initiative is set to empower developers and corporations of all sizes, enhancing their ability to leverage AI technologies and further stimulating the growth of the open-source community,” said Jingren Zhou, CTO of Alibaba Cloud Intelligence.
AI infrastructure upgrade
In addition, Alibaba Cloud unveiled a revamped full-stack infrastructure designed to meet the growing demands for AI computing. This new infrastructure includes innovative cloud products and services that enhance computing, networking, and data center architecture, designed to support the development and application of AI models.
This includes a Alibaba Cloud Open Lake service and the PAI AI Scheduler with integrated model training and inference.
On its part, the 9th Generation Elastic Compute Service (ECS) instance offers a 30% increase in search recommendation speed and a 17% improvement in read and write queries for database products compared to the previous generation.
Finally, Alibaba Cloud revealed its latest data center architecture it calls CUBE DC 5.0. It offers air-liquid hybrid cooling with a direct current power distribution architecture for energy efficiency. Deployment time is reduced by as much as 50% due to a modular, prefabricated components.
“Alibaba Cloud is investing, with unprecedented intensity, in the research and development of AI technology and the building of its global infrastructure. We aim to establish an AI infrastructure of the future to serve our global customers and unlock their business potential,” said Eddie Wu, Chairman and CEO of Alibaba Cloud Intelligence.
Paul Mah
Paul Mah is the editor of DSAITrends, where he report on the latest developments in data science and AI. A former system administrator, programmer, and IT lecturer, he enjoys writing both code and prose.