[China , Dongguan , July 7 , 2023 ] The Huawei Developer Conference 2023 ( Cloud) officially kicked off in Dongguan, China on July 7, and at the same time held more than 30 cities in more than 10 countries around the world. In the sub-venue, global developers are invited to gather together to exchange and share new industrial opportunities and new technological practices under the AI wave.
In the keynote speech of the conference held on the afternoon of the 7th, Zhang Pingan, executive director of Huawei and CEO of Huawei Cloud, released Pangu Large Model 3.0 and Shengteng AI cloud service. Among them, Pangu Large Model 3.0 focuses on the three innovation directions of "industry reshaping", "technology rooting" and "opening up and flying together" to continuously build core competitiveness and provide better services for industry customers, partners and developers. Ascend AI cloud service single cluster provides 2000P Flops computing power, and the long-term stability rate reaches 90% after 30 days of kilocalorie training, providing stable and reliable AI computing power for the industry, and making large models within easy reach.
Zhang Pingan, Executive Director of Huawei and CEO of Huawei Cloud, released the Pangu Model 3.0
Zhang Ping'an said : "The Pangea Big Model will allow every industry, every company, and everyone to have their own expert assistants to make work more efficient and easier. We always adhere to the strategy of AI for Industries, and continue to develop on the road of deepening the industry. Go forward. I firmly believe that big models will reshape thousands of industries, and every developer will be a hero who changes the world."
Pangu Big Model 3.0 : Reshaping Thousands of Industries
Huawei Cloud Pangu Large Model has focused on serving the industry from the very beginning. The Pangu Large Model 3.0 released at this conference is also an industry-oriented large model series, including the "5+N+X" three-tier architecture:
The L 0 layer includes five basic large models of natural language, vision, multimodality, forecasting, and scientific computing, providing a variety of skills to meet the needs of industry scenarios. Pangu 3.0 provides customers with a series of basic large models with 10 billion parameters, 38 billion parameters, 71 billion parameters and 100 billion parameters, matching the diversified needs of customers in different scenarios, different time delays, and different response speeds. At the same time, it provides a new set of capabilities, including knowledge question answering, copy generation, and code generation for large NLP models, as well as image generation and image understanding for multi-modal large models. These skills can be directly used by customers and partner companies. Regardless of the size of the large model, Pangu provides a consistent set of capabilities.
Layer L1 is N large industry models. HUAWEI CLOUD can provide general industry large models trained using industry public data, including government affairs, finance, manufacturing, mining, meteorological and other large models; it can also be based on industry customers' own data , On the L0 and L1 layers of the Pangu large model, train your own proprietary large model for customers.
The L2 layer provides customers with more detailed scene models, and is more focused on specific industry applications or specific business scenarios such as government hotlines, outlet assistants, leading drug screening, foreign object detection on conveyor belts, and typhoon path prediction. Ready-to-use" model service.
The Pangu large model adopts a complete layered decoupling design, which can be quickly adapted and quickly met the changing needs of the industry. Customers can load independent datasets for their own large models, and can also upgrade the base model or capability set separately. On the basis of L0 and L1 large models, HUAWEI CLOUD also provides customers with large-scale industry development kits. Through secondary training on customer-owned data, customers can have their own exclusive industry large models. At the same time, according to different data security and compliance requirements of customers, Pangea Large Model also provides public cloud, large model cloud area, and diversified deployment forms of hybrid cloud.
Zhang Pingan, Executive Director of Huawei and CEO of Huawei Cloud
"Pangu was born for the industry, so we must think about the industry and better serve customers from all walks of life," Zhang Ping'an said. Today, the Pangea large model has played a huge role in many industries such as finance, finance, manufacturing, pharmaceutical research and development, coal mines, and railways.
In the field of government affairs, HUAWEI CLOUD and Shenzhen Futian District Government Service Data Management Bureau have launched Xiaofu, a Futian government affairs smart assistant based on the Pangu government affairs model. The language that works for the government makes the city more loving. The Pangu government affairs large model fine-tunes more than 200,000 pieces of government affairs data, including 12345 hotlines, policy documents, government affairs encyclopedias, etc., and has mastered a wealth of industry knowledge such as laws and regulations, handling procedures, etc. Xiaofu, based on the large model of Pangu's government affairs, embodies industry, professionalism, leading and benefiting the people in the one-stop business.
You Peng, Director of Huawei Cloud EI Service Product Department, introduced the Pangu government affairs model
In the field of coal mines, the Pangu Mine large-scale model has been used in 8 mines across the country. A large-scale model can cover more than 1,000 subdivided scenarios under business processes such as mining, excavation, machinery, transportation, transportation, and washing of coal mines, allowing more More coal miners can work on the ground, which not only makes the working environment of coal miners more comfortable, but also greatly reduces safety accidents.
In the field of railways, the large model of Pangu Railway can accurately identify 67 types of freight cars and more than 430 types of faults running on the current network, and the screening rate of fault-free pictures is as high as 95%. It is liberated from the detection of millions of "picture seas" every day.
In the field of meteorology, the Pangu Meteorological Large Model is the first AI forecasting model whose accuracy exceeds traditional numerical forecasting methods, and the forecasting speed has also been greatly improved. It turns out that predicting the path of a typhoon in the next 10 days requires 5 hours of simulation on a high-performance computer cluster with 3,000 servers. Now, based on the pre-trained Pangu meteorological large model, through AI reasoning, researchers only need to configure a single card on a single server, and can obtain more accurate prediction results within 10 seconds.
Tian Qi, Chief Scientist of HUAWEI CLOUD Artificial Intelligence, introduced how to refine a basic model into an industry model
In the financial field, the Pangu financial large model can pre-train various bank operations, policies, and case documents, and can automatically generate procedures and operation guidance for counter staff according to customer problems, reducing the average number of operations that originally required 5 times For one time, the closing time is shortened by more than 5 minutes. The Pangu model allows hundreds of thousands of tellers to have their own smart assistants.
In the field of manufacturing, in the past, it often took more than 3 hours to complete the production plan for one day when making a device allocation plan for a single production line. After learning the various device data, business processes and rules of Huawei's production line, the Pangu manufacturing model can accurately understand the business needs, and call the Tianchou AI solver plug-in to make the next three days in one minute. Production Plan.
In the field of drug research and development, the original research and development of a new drug takes an average of 10 years and costs 1 billion US dollars. The large molecular model of Pangu drugs helped the team of Professor Liu Bing of the First Affiliated Hospital of Xi’an Jiaotong University discover the world’s first new target and new class of antibiotics in 40 years, and shorten the lead drug development cycle to one month and reduce the development cost by 70%.
Rooted in technology and extreme performance, creating another pole of AI in the world
The innovation of the large model is not only the innovation of the model itself, but also depends on the innovation of various root technologies of AI. Huawei has built an AI computing power cloud platform based on Kunpeng and Ascend at the bottom layer, as well as the heterogeneous computing architecture CANN, the full-scenario AI framework MindSpore, and the AI development production line ModelArts, etc., to provide distributed computing power for the development and operation of large models. Key capabilities such as parallel acceleration, operator and compilation optimization, and cluster-level communication optimization. Based on Huawei's AI root technology, the performance of large model training can be adjusted to 1.1 times that of mainstream GPUs in the industry.
Yao Jun, Director of Huawei's Noah's Ark Laboratory, introduced Pangu's basic large-scale model full-stack innovation
Computing power is the basis for training large models. At this conference, Zhang Ping'an announced that the Ascend AI cloud service with a single cluster 2000P Flops computing power will be launched simultaneously in Ulanqab and Gui'an AI computing power centers of Huawei Cloud . In addition to supporting Huawei's full-scenario AI framework Shengsi MindSpore, Ascend AI Cloud Service also supports mainstream AI frameworks such as Pytorch and Tensorflow. At the same time, 90% of operators in these frameworks can be smoothly migrated to the Ascend platform through Huawei's end-to-end migration tool. For example, Meitu migrated 70 models to Ascend in just 30 days. At the same time, HUAWEI CLOUD and the Meitu team jointly optimized more than 30 operators and accelerated the process in parallel. The AI performance was improved by 30% compared with the original solution. .
In addition, GPU failures are often encountered during large model training, and developers have to restart training frequently, which takes a long time and costs a lot. Ascend AI cloud service can provide longer and more stable AI computing power services. The 30-day long-term stability rate of kilocalorie training reaches 90%, and the breakpoint recovery time does not exceed 10 minutes.
"In order to help global customers, partners, and developers train and use large models, we are committed to creating another pole of the world's AI for global customers and providing new choices for all AI developers," Zhang Pingan said.
In-depth integration, Pangea large model reconstruction product innovation
In addition to the practice in thousands of industries, the HUAWEI CLOUD Pangu model has also been deeply integrated into HUAWEI CLOUD's product services to restructure product innovation.
For example, in the data service, the copywriting and code generation technology of Pangu Large Model can improve the efficiency of data writing and front-end code writing, and greatly shorten the cycle of new product launch and empowerment. In the cloud customer service, through the dialogue Q&A embedded in the industry knowledge base and intent mining ability, the whole process of AI priority answering is realized, and the efficiency of customer service is improved by 30%. In BI, through NL2SQL and AutoGraph intelligent routing, automatic recommendation from SQL to visual charts is realized, and through multiple rounds of natural language interaction, everyone can easily gain insight into business details from data. In cloud search, through multi-modal Embedding and NL2API technology, it realizes video, text, map and other wide-ranging scene searches. With the help of powerful semantic understanding and generalization capabilities, the search accuracy rate is increased by 15%.
At the same time, HUAWEI CLOUD combined the CodeArts R&D tools with the Pangu large model, and officially released CodeArts Snap, an intelligent programming assistant for developers . The tool has trained 76 billion lines of selected codes and 13 million technical documents. It has three core functions of intelligent generation, intelligent question and answer, and intelligent collaboration. It can realize code generation in one sentence of dialogue, and automatic annotation and generation of test cases with one click. One command can be deployed intelligently, so that every software developer has his own programming assistant.
Xu Feng, Director of Huawei Cloud PaaS Service Product Department, introduced CodeArts to reshape software development
In addition, in order to enable enterprises to build stronger content innovation capabilities in the AI era, HUAWEI CLOUD empowers the MetaStudio digital content production line through the Pangu basic large model, creates the Pangu digital NPC model, provides two major services of model generation and model drive, and has been used 200,000 hours of audio and video data were used for pre-training. Based on these two major services, developers can quickly generate and drive digital human models, empowering online education, entertainment live broadcast, corporate conferences and other industry applications, so that every enterprise employee can realize "digital human freedom". For example, users only need to upload a 20-second personal video on the service page of HUAWEI CLOUD MetaStudio to quickly generate a personalized digital human explanation video. The work completed by three R&D personnel in three days in the past can now be completed in only three minutes .
Lu Yangming, Director of Huawei Cloud Media Service Product Department, introduced MetaStudio to reshape digital content production
Open to the same flight, join hands with customers and partners to win the large-scale model ecology
HUAWEI CLOUD is committed to building an open and win-win global ecosystem centered on developers. Currently, the number of HUAWEI CLOUD developers worldwide has exceeded 4.6 million, and the number of products on the cloud store has reached more than 10,000. HUAWEI CLOUD provides an easy-to-use and reliable toolkit for large models, Kaitian aPaaS that gathers a large number of multi-industry scene APIs, and an exclusive community for large models that includes rich and high-quality courses and technical certifications. We hope to explore Pangu together with developers and partners. The innovation path of combining models and industries.
Kang Ning, President of HUAWEI CLOUD Global Ecosystem Department, demonstrates the large-scale model ecological plan
The development of large models is inseparable from the continuous input of high-quality data. HUAWEI CLOUD, together with the China Public Relations Association, the Cultural Big Data Industry Committee, and a number of partners, jointly initiated the establishment of a large-scale high-quality data alliance. The alliance will gather open data from various member units, create high-quality data sets covering thousands of industries, and promote the vigorous development of industry models.
Developers are the core force driving digital innovation. At this conference, the 2023 Huawei Developer Contest with the theme of "Unlimited Creativity" was officially launched. As the top event in Huawei's ICT field, the Huawei Developer Contest has set up two major tracks, cloud base and industry, covering China, Asia Pacific, Europe, North Africa, and Latin America, allowing developers to show their talents in the field of artificial intelligence and realize innovative application development .