Alibaba Cloud open sourced over 100 of its newly-launched AI models, collectively named Qwen 2.5. The company made this announcement during the Apsara Conference, which is its annual conference.
Alibaba Group also states that its cloud computing arm has launched a new full-stack infrastructure that is to cater to the skyrocketing demand for strong AI computing. Based on cloud products and services meant to bring innovation to computing, networking, and data centre architecture, the new infrastructure goes a long way in stimulating the application and development of AI models on a full scale.
Alibaba Cloud Intelligence chairman and CEO, Eddie Wu said: “Alibaba Cloud is investing unprecedented intensity in the research and development of AI technology and building of its global infrastructure. This is an AI infrastructure of the future we intend to build for customers all around the world, unlocking their business potential.
Launched – the new Qwen 2.5 models, with sizes ranging from 0.5 to 72 billion parameters in addition to better knowledge and more significant capabilities in math and coding. Available for over 29 languages, offered for various kinds of AI applications at the edge and in the cloud for numerous industries, including automotive and gaming as well as science research.
Open-source AI models gain traction at Alibaba Cloud
Alibaba Group also states that its cloud computing arm has launched a new full-stack infrastructure that is to cater to the skyrocketing demand for strong AI computing. Based on cloud products and services meant to bring innovation to computing, networking, and data centre architecture, the new infrastructure goes a long way in stimulating the application and development of AI models on a full scale.
Alibaba Cloud Intelligence chairman and CEO, Eddie Wu said: “Alibaba Cloud is investing unprecedented intensity in the research and development of AI technology and building of its global infrastructure. This is an AI infrastructure of the future we intend to build for customers all around the world, unlocking their business potential.
Launched – the new Qwen 2.5 models, with sizes ranging from 0.5 to 72 billion parameters in addition to better knowledge and more significant capabilities in math and coding. Available for over 29 languages, offered for various kinds of AI applications at the edge and in the cloud for numerous industries, including automotive and gaming as well as science research.
Open-source AI models gain traction at Alibaba Cloud
Since its release in April of this year, the Qwen model series has drawn much attention to itself as it has gained 40 million downloads on both Hugging Face and ModelScope. Even more impressive, the models have inspired more than 50,000 derivative models on Hugging Face alone.
This will enable developers and corporations of all sizes to employ AI capabilities and further energize the open-source ecosystem, Alibaba Cloud Intelligence CTO Jingren Zhou noted.
In addition to open-source models, Alibaba Cloud also announced an update of its flagship proprietary model, Qwen-Max, which allegedly supports performance levels comparable to other state-of-the-art models in language understanding, reasoning capability, and mathematics and code authoring.
The company has also developed multimodal capabilities through launching new text-to-video models, which are a member of the Tongyi Wanxiang large model family. Such a model can be able to produce fine-quality videos on topics inspired by Chinese and English language instruction, with styles ranging from realistic scenes to 3D animation.
In addition, Alibaba Cloud launched Qwen2-VL, a new version of the vision language model that can understand videos longer than 20 minutes and support question-answering by video. The company also launched an AI Developer, which is the Qwen-led AI assistant designed to support the programmer by assisting in automated tasks such as requirement analysis, code programming, and bug identification and correction
Infrastructure upgrades announced to support these AI advancements include:
CUBE DC 5.0: Next-Generation Data Center Architecture, Designed for Enhanced Energy and Operational Efficiency
Alibaba Cloud Open Lake: Solutions that Maximize Data Utility for Generative AI Applications
PAI AI Scheduler: Dedicated Alibaba Cloud Native Scheduling Engine with Advanced Computing Resources
DMS: OneMeta+OneOps?-a platform providing unified metadata management capabilities across multiple cloud environments
9th Generation Enterprise Elastic Compute Service (ECS) instance: Offering enhancement on various applications performance.
Upgrades from Alibaba Cloud include the release of more than 100 open-source models to service the target customers and partners in maximising the benefits through latest technology in building more efficient, more sustainable, and inclusive applications of AI.
This will enable developers and corporations of all sizes to employ AI capabilities and further energize the open-source ecosystem, Alibaba Cloud Intelligence CTO Jingren Zhou noted.
In addition to open-source models, Alibaba Cloud also announced an update of its flagship proprietary model, Qwen-Max, which allegedly supports performance levels comparable to other state-of-the-art models in language understanding, reasoning capability, and mathematics and code authoring.
The company has also developed multimodal capabilities through launching new text-to-video models, which are a member of the Tongyi Wanxiang large model family. Such a model can be able to produce fine-quality videos on topics inspired by Chinese and English language instruction, with styles ranging from realistic scenes to 3D animation.
In addition, Alibaba Cloud launched Qwen2-VL, a new version of the vision language model that can understand videos longer than 20 minutes and support question-answering by video. The company also launched an AI Developer, which is the Qwen-led AI assistant designed to support the programmer by assisting in automated tasks such as requirement analysis, code programming, and bug identification and correction
Infrastructure upgrades announced to support these AI advancements include:
CUBE DC 5.0: Next-Generation Data Center Architecture, Designed for Enhanced Energy and Operational Efficiency
Alibaba Cloud Open Lake: Solutions that Maximize Data Utility for Generative AI Applications
PAI AI Scheduler: Dedicated Alibaba Cloud Native Scheduling Engine with Advanced Computing Resources
DMS: OneMeta+OneOps?-a platform providing unified metadata management capabilities across multiple cloud environments
9th Generation Enterprise Elastic Compute Service (ECS) instance: Offering enhancement on various applications performance.
Upgrades from Alibaba Cloud include the release of more than 100 open-source models to service the target customers and partners in maximising the benefits through latest technology in building more efficient, more sustainable, and inclusive applications of AI.
Post a Comment