At the Wave Summit 2023 Deep Learning Developer Conference held on August 16, Baidu Chief Technology Officer and Director of the National Engineering Research Center for Deep Learning Technology and ApplicationsWang Haifeng said, The large language model has four core basic capabilities of understanding, generation, logic, and memory, bringing the dawn of general artificial intelligence.
In Wang Haifeng’s view, the typical abilities of artificial intelligence, such as creation, programming, problem solving, planning, etc., all rely to varying degrees on the four core basic abilities of understanding, generation, logic, and memory. . Take problem solving as an example. From reading the question, to answering the question, to finally writing the answer, artificial intelligence is nothing more than a combination of the four major abilities of understanding, memory, logic, and generation.
How to obtain the four core basic capabilities of artificial intelligence? First, we integrate learning from trillions of data and hundreds of billions of knowledge, and develop a large pre-trained model. On this basis, we implement reinforcement learning through fine-tuning and feedback, and finally let the large model obtain Knowledge enhancement, retrieval enhancement, dialogue enhancement. Deep learning technology is the only way to achieve general artificial intelligence.
In the exploration of deep learning framework technology, Baidu launched the open source and open flying paddle deep learning platform. Flying Paddle Deep Learning Platform Wenxin large model, mutually promote and promote each other, forming the basis of Baidu's artificial intelligence strategy.
According to Wang Haifeng, after years of development, Baidu Fei Paddle Ecosystem has gathered 8 million developers, served 220,000 enterprises and institutions, and created a total of 800,000 models based on Fei Paddle.
At the Wave Summit conference, Baidu officially introduced the Chinese name "Galaxy Community" to the Flying Paddle developer community AI Studio. Wang Haifeng said that Baidu is willing to work with developers to build the Galaxy community and go to the starry sea of general artificial intelligence with the blessing of Feipiao deep learning platform Wenxin Big Model.
In the era of artificial intelligence, the IT technology stack has undergone fundamental changes, from a three-layer architecture of chip-operating system-application to a four-layer architecture of chip-framework-model-application. Among them, the deep learning framework can be called the operating system in the era of artificial intelligence, connecting downwards to various chips and cloud computing platforms, and upwards to undertake various AI business models and industry applications.
In the national "14th Five-Year Plan", the deep learning framework is included in the category of "new generation artificial intelligence" and has become a cutting-edge technology that is focused on support.
In 2010, Baidu announced All in AI, a full-stack artificial intelligence strategy. After investing hundreds of billions of dollars in R&D, Baidu has gradually made breakthroughs in the field of artificial intelligence: at the chip level, there is the Kunlun core; at the framework level, there is the Feipiao deep learning platform; at the model level, there is the Wenxin large model; in the application The layer has intelligent search and automatic driving.
The open-source Fei Paddle deep learning platform in 2016 is a major breakthrough for Baidu in artificial intelligence framework layer technology. It has become one of the preferred platforms for domestic artificial intelligence developers along with TensorFlow and PyTorch, the two major foreign deep learning frameworks. one.
At the Wave Summit conference, Ma Yanjun, general manager of Baidu AI technology ecology, officially released the Flying Paddle open source framework v2.5 version. The infrastructure of Flying Paddle has been fully upgraded, the differential function has been improved, and the performance optimization has been universal. . At the same time, PaddleScience v1.0, a scientific computing tool component, was officially released, and PaddleHelix released the HelixDock protein-compound conformational affinity prediction large model.
After opening up the Feipiao deep learning platform and the Wenxin large model, Baidu is working hard to build the Feipiao Wenxin ecosystem. Software and hardware adaptation and the construction of the Galaxy community have become the focus of development.
In terms of software and hardware adaptation, China Electronics Standardization Institute took the lead, and Baidu, Sugon, Feiteng, and Inspur jointly drafted the national standard "Artificial Intelligence Deep Learning Framework Multi-Hardware Platform Adaptation Technical Specification." Based on this national standard, Baidu Feipiao and Wenxin Large Model cooperate with NVIDIA, Sugon, Kunlun Core, Cambrian, Huawei, Tianshu Zhixin, Biren Technology, Qualcomm, Intel, MediaTek, Moore Thread, and ARM Technology 12 hardware partners have carried out adaptation, covering a variety of cloud and device-side hardware types.
In terms of community, AI Studio (Galaxy Community) is already the largest AI community in China, gathering 6.09 million development projects. With the advancement of large model technology, large model developers are also rapidly gathering in the Galaxy community. At the Wave Summit conference, Wu Tian, vice president of Baidu Group and deputy director of the National Engineering Research Center for Deep Learning Technology and Applications, announced the latest upgrade of the Galaxy community and officially launched the Galaxy large model community. According to reports, the Galaxy large model community has accumulated more than 300 large model creative applications.
At the conference, Wu Tian also announced the Wenxin Big Model Galaxy co-creation plan. Baidu will provide ecological and industrial resources to jointly promote artificial intelligence application innovation with developers.
On March 16, Baidu officially released the knowledge-enhanced large model Wen Xinyiyan. In less than half a year, Wenxinyiyan, which has accelerated its evolution and iteration, has launched version 3.5.
Wenxin Yiyan’s rapid progress is due to the mutual promotion of Feipiao’s industrial-grade deep learning open source open platform and Wenxin’s large model. According to Wu Tian, in terms of efficiency, through Fei Paddle’s end-to-end adaptive hybrid parallel training technology and collaborative optimization of compression, inference, and service deployment, the training speed of Wenxin’s large model has reached 3 times and the inference speed has reached more than 30 times.
At the Wave Summit conference, Wu Tian released five native plug-ins of Wen Xinyiyan: Baidu Search, Browsing Documents (document-based interaction), Eyan Yitu (data insight chart generation), Shuo Diagram painting (image-based interaction), Yijingliuying (text to video). Wen Xin Yi Yan supports the use of three plug-ins at the same time, which not only expands the capabilities of large models, but also better adapts to scene needs.
With the help of plug-ins, Wen Yanyiyan demonstrated the black technology of producing commercials in seconds and getting code in a few sentences.
Baidu Chief Technology Officer Wang Haifeng said that in the future, Baidu will build a plug-in ecosystem with developers and share the results of technological innovation. Technology and ecology are grasped with both hands, and the application of flying paddle Wenxin has been implemented and is accelerating.
Text: Bugle-X / Data Monkey
The above is the detailed content of Toward general artificial intelligence, Baidu Feipiao + Wenxin create the largest AI community in China. For more information, please follow other related articles on the PHP Chinese website!
![[Web front-end] Node.js quick start](https://img.php.cn/upload/course/000/000/067/662b5d34ba7c0227.png)
