Alibaba Cloud CTO Zhou Jingren: Creating the Most Open Cloud in the AI Era

On October 31, at the 2023 Cloud Habitat Conference, Alibaba Cloud CTO Zhou Jingren stated that in the era of intelligence, Alibaba Cloud will upgrade its cloud computing system through full stack technology innovation from underlying computing power to AI platforms and model services, creating the most open cloud in the AI era. On site, Zhou Jingren announced the latest progress in cloud computing infrastructure, upgraded the artificial intelligence platform, and released the large-scale model Tongyi Qianwen 2

On October 31, at the 2023 Cloud Habitat Conference, Alibaba Cloud CTO Zhou Jingren stated that in the era of intelligence, Alibaba Cloud will upgrade its cloud computing system through full stack technology innovation from underlying computing power to AI platforms and model services, creating the most open cloud in the AI era. On site, Zhou Jingren announced the latest progress in cloud computing infrastructure, upgraded the artificial intelligence platform, and released the large-scale model Tongyi Qianwen 2.0 with hundreds of billions of parameter scales, as well as the one-stop model application development platform Alibaba Cloud Bailian. Alibaba Cloud has initially built a full stack cloud computing system in the AI era.

At present, half of China's major model enterprises are running on Alibaba Cloud, and 2.8 million AI developers are active in the Alibaba Cloud magic platform community. In the future, Alibaba Cloud will work together with Qianxing Baiye to promote AI innovation and share technological dividends, "said Zhou Jingren.

Comprehensive upgrade of AI infrastructure

Large models are the core technology of this wave of AI, and the quality of basic models largely determines the prospects for AI industrialization. Training large models is a systematic engineering that encompasses complex technologies such as computing power bases, networks, storage, big data, AI frameworks, AI models, etc. Only a powerful cloud computing system can train high-quality large models.

Zhou Jingren stated that Alibaba Cloud proposed the concept of "data center is a computer" in 2009, and today, the AI era requires such a technical system even more. As a supercomputer, cloud computing can efficiently connect heterogeneous computing resources, break through the bottleneck of a single performance chip, and collaborate to complete large-scale intelligent computing tasks.

In order to ensure stable interconnection and efficient parallel computing of large model training, Alibaba Cloud has upgraded its artificial intelligence platform PAI. The bottom layer of PAI adopts the HPN7.0 new generation AI cluster network architecture, supporting scalable clusters of up to 100000 cards, and the acceleration ratio of large-scale distributed training is up to 96%, far exceeding the industry level; In large model training tasks, it can save over 50% of computing resources and lead the world in performance.

The Alibaba Cloud Tongyi Big Model series is based on artificial intelligence platform PAI training. In addition to the Tongyi Big Model, half of China's major model companies are running on Alibaba Cloud, while a large number of leading enterprises and institutions such as Baichuan Intelligence, Zhipu AI, Zero One Universe, Kunlun Wanwei, vivo, and Fudan University are all training big models on Alibaba Cloud.

Wang Xiaochuan, founder and CEO of Baichuan Intelligent, stated that "Baichuan has released 7 large models since its establishment only half a year ago, and the support of cloud computing is indispensable for rapid iteration." Baichuan Intelligent and Alibaba Cloud have conducted in-depth cooperation, and with the joint efforts of both parties, Baichuan has successfully completed the training task of 1000 card large models, effectively reducing model inference costs and improving model deployment efficiency.

Alibaba Cloud has become a public AI computing base for China's big models. As of now, many leading mainstream large models in China have provided API services through Alibaba Cloud, including the Tongyi series, Baichuan series, Zhipu AIChatGLM series, and Jiang Ziya universal large models.

With the gradual deepening of AI industrialization, there will inevitably be an explosion of large-scale demand for intelligent computing power. Alibaba Cloud has built 89 cloud computing data centers in 30 regions around the world, providing more than 3000 edge computing nodes. The advantages of low latency and high elasticity of cloud computing will be brought into full play. This year, Alibaba Cloud successfully supported the popular Miaoyang camera across the entire network, which experienced a short-term and high-intensity traffic explosion.

Moving towards the Cloud of Autonomous Driving

Zhou Jingren said, "With the integration of big model technology and cloud computing itself, we hope that in the future, the cloud can be autonomous like a car, greatly improving the experience of developers using the cloud

It is reported that over 30 cloud products on Alibaba Cloud have integrated big model capabilities. For example, the Alibaba Cloud big data governance platform DataWorks has added a new interactive form - Copilot. Users only need to input in natural language to generate SQL and automatically perform corresponding data ETL operations. Overall development and analysis can improve efficiency by more than 30%, comparable to "autonomous driving".

Alibaba Cloud containers, databases, and other products also provide similar development experiences, which can achieve functions such as NL2SQL and SQL annotation generation/error correction/optimization. In the future, these capabilities will be integrated into other Alibaba Cloud products.

Better big models to accelerate application innovation

At the 2023 Yunqi Conference, Zhou Jingren announced the latest progress of Alibaba's self-developed large model and released the Tongyi Qianwen 2.0 with a parameter scale of hundreds of billions. Tongyi Qianwen 2.0 has comprehensively surpassed GPT-3.5 and Llama2 in 10 authoritative evaluations, accelerating its pursuit of GPT-4.

Qianxing and Baiye all want to use big models to achieve changes in production and service methods, but the high threshold of using big models has kept most people out of the technological wave. Whether it is customizing exclusive large models or building innovative applications based on large models, there are high requirements for talent, technology, and funding.

At the Yunqi Conference, Zhou Jingren released a one-stop large model application development platform - Alibaba Cloud Bailian. This platform integrates mainstream high-quality large models both domestically and internationally, providing services such as model selection, fine-tuning training, security kits, model deployment, and full link application development tools, simplifying complex tasks such as low-level computing deployment, model pre training, and tool development for users. Developers can develop a large model application within 5 minutes and "refine" an enterprise exclusive model within a few hours, allowing them to focus more on application innovation.

To promote the easier integration and implementation of big models in various industries, Alibaba Cloud has "sampled" 8 industry big models based on Tongyi, and announced their progress on site: personalized character creation platform Tongyi Xingchen, intelligent investment research assistant Tongyi Dianjin, AI reading assistant Tongyi Zhiwen, etc. have made their debut; The intelligent coding assistant Tongyi Lingcode has been widely adopted within Alibaba Cloud and has received high praise; The work learning AI assistant Tongyi Tingwu processes over 50000 audio and video recordings every day, accumulating over 1 million users.

The big model is triggering a new round of innovation in Qianxing Baiye. Currently, companies such as CCTV, Longxin Technology, and Asia Information Technology have taken the lead in developing exclusive models and applications on Alibaba Cloud Bailian. Longxin Technology has trained a power exclusive big model on the cloud, developed "Power Bill Interpretation Intelligent Assistant" and "Power Industry Policy Analysis/Data Analysis Assistant", improving customer reception efficiency by 50% and reducing complaints by 70%.

Promoting Ecological Prosperity with a Firm Openness

Promoting the prosperity of China's AI ecosystem is Alibaba Cloud's primary goal. Alibaba Cloud will firmly build the most open big model platform in the AI era. We welcome all big models to join Alibaba Cloud's Bailian and provide AI services to developers together, "said Zhou Jingren.

Alibaba Cloud is the earliest technology company in China to open source and self-developed large models, sparking a wave of open source for large models in China. Currently, Alibaba Cloud has opened up versions 7B and 14B of Tongyi Qianwen, with over a million downloads. At the scene, Zhou Jingren announced that the Tongyi Qianwen 72B model was about to be open source, making it the largest open-source model with the largest parameters in China.

In addition to sharing self-developed new technologies with developers, Alibaba Cloud also strongly supports the development of third-party large models. On the Alibaba Cloud Magic Building community, top players in the industry such as Baichuan Intelligence, Zhipu AI, Shanghai Artificial Intelligence Laboratory, and IDEA Research Institute have all released their core big models through open source. Alibaba Cloud provides free GPU computing power for developers to "try out" big models, which has exceeded 30 million hours so far.

Zhou Jingren revealed that the Magic Building Community has gathered over 2300 AI models, attracting 2.8 million AI developers. The download volume of AI models has exceeded 100 million, making it the largest and most active AI community in China.

At the 2023 Yunqi Conference, Alibaba Cloud announced a heavyweight plan: the "Cloud Industry Development Plan", which will give every college student in China a cloud server. In addition, Alibaba Cloud will also provide larger scale computing resource support for contracted universities, helping Chinese young scholars and students climb the scientific research peak. At present, Tsinghua University, Peking University, Zhejiang University, Shanghai Jiao Tong University, University of Science and Technology of China, South China University of Technology and other universities have reached the first batch of cooperation.

Disclaimer: The content of this article is sourced from the internet. The copyright of the text, images, and other materials belongs to the original author. The platform reprints the materials for the purpose of conveying more information. The content of the article is for reference and learning only, and should not be used for commercial purposes. If it infringes on your legitimate rights and interests, please contact us promptly and we will handle it as soon as possible! We respect copyright and are committed to protecting it. Thank you for sharing.(Email:[email protected])