On December 28, Wang Haifeng, chief technology officer of Baidu Group and director of the National Engineering Research Center for Deep Learning Technology and Applications, said at the 10th WAVE SUMMIT Deep Learning Developers Conference that since August 31, ERNIE Bot has been open to the society, the number of users has exceeded 100 million, and the daily number of questions has grown rapidly. The number of developers of the deep learning framework platform PaddlePaddle has exceeded 10 million, reaching 10.7 million. Baidu Group unveiled the Star River Community - Big Model Tool Center at the conference to lower the development threshold for big model applications.
Photo by Peng Siyu, on-site reporter at the 10th WAVE SUMMIT Deep Learning Developer Conference
As of the close on December 28th, Baidu Group's Hong Kong stock rose 3.05% to HKD 114.9 per share, with a total market value of HKD 321.3 billion.
The number of ERNIE Bot users has exceeded 100 million
Wang Haifeng shared the latest achievements of Baidu PaddlePaddle and Wenxin at the conference.
Wang Haifeng said that in October, the basic model of ERNIE Bot, a big language model for knowledge enhancement, was upgraded to 4.0, and the four basic AI capabilities of understanding, generation, logic and memory were comprehensively improved. At present, the overall effect of Wenxin Big Model 4.0 has been improved by 32%.
User scale is an important indicator for measuring the prosperity of software applications. Wang Haifeng disclosed on the spot that the number of ERNIE Bot users has exceeded 100 million. Since the opening of ERNIE Bot to provide services to the public on August 31, the number of users' questions has risen all the way, basically in step with the improvement of the effect of Wenxin's big model. More and more users trust and use ERNIE Bot.
In terms of PaddlePaddle platform, according to Wang Haifeng, up to now, PaddlePaddle has gathered 10.7 million developers, served 235000 enterprises and institutions, and created 860000 models based on PaddlePaddle.
Since 2019, Baidu has been deeply involved in the development of pre training models and has released the Wenxin Big Model 1.0. After nearly four years of accumulation, Baidu released ERNIE Bot in March this year.
Wang Haifeng stated that starting from 2022, the addition of large models to deep learning platforms has connected the entire AI industry chain from hardware adaptation, model training, inference deployment, to scenario applications, solidifying the foundation of industrial intelligence. This year, the emergence of big language models has brought dawn to general artificial intelligence.
Development de thresholding
Large models will enter a year of large-scale application
Currently, the development of large-scale model tracks is in full swing both domestically and internationally. On the one hand, the entire industry is accelerating technological exploration, and on the other hand, it is rapidly improving the industrial ecosystem, attracting developers to jointly stimulate the innovation activity of large model applications.
In terms of developer support, Baidu released the Galaxy Community - Big Model Tool Center, which can provide developers with visual interaction and rich tools, and upgraded the ERNIE Bot mechanism to comprehensively reduce the threshold of application development. Wu Tian, vice president of Baidu Group and deputy director of the National Engineering Research Center for Deep Learning Technology and Application, said that these upgrades would solve the problems of the lack of supporting development tools, cumbersome configuration file writing, difficult effect optimization and difficult application promotion existing in the process of using ERNIE Bot.
Ma Yanjun, General Manager of Baidu AI Technology Ecology, stated that with the further optimization of the entire process of the big model suite, the experience of developers calling big models for application development will be greatly improved. 2024 will be a year for large-scale application of big models.
Guoxin Securities believes that as the development of base models enters a new stage, model function iteration and application ecosystem construction are synchronized. Agent platforms such as GPTs are initially attempting to customize personal AI applications for the general public, accelerating the iteration of the process of big language models becoming "operating systems".
CITIC Securities stated that with the continuous progress of algorithms and the gradual implementation of application scenarios, generative AI is expected to continue to become one of the core themes of the development of the technology industry in 2024, and correspondingly affect the total and structure of global enterprise IT expenditure, the demand of the semiconductor and hardware industries, as well as the operational efficiency of internet companies. More intelligent models, multimodality, and reduced inference costs will become the main technological directions for breakthroughs in the AI industry in 2024; The underlying computing power demand structure, product form, market pattern, etc. are also expected to be adjusted; In terms of application scenarios, AI PC/mobile phones, AI+software, autonomous driving, robots, the Internet, etc. are expected to become the main application directions for large models in 2024. Continuously optimistic about short-term, medium and long-term investment opportunities in the global AI industry.
转播
