Good news! IDC released the first AI large model evaluation report in China, Baidu Wenxin model version 3.5 has been available for internal testing, and the comprehensive ability of Wenxin model 3.5 is the first in China! As early as May, Baidu chairman and CEO Robin Li revealed the news.
As the world's first large-language model to benchmark ChatGPT, Wenxin Model integrates knowledge enhancement, retrieval enhancement and dialogue enhancement technology innovation, and has achieved outstanding results beyond ChatGPT 3.5 in a number of public tests. IDC released the first AI big model evaluation report in China, according to which the first brother of China's big model is Baidu Wenxin, with 7 full scores, and the only full score for algorithm models.
The evaluation shows the advantages and disadvantages of the large model in terms of ability, and also has a strong guiding effect on the iterative development of the model. In the Chinese AGIEval evaluation, the Wen Xin large model scored 3.5 and scored 64.37, ranking first. In the English part of the AGIEval evaluation, the Wen Xin model scored 50.59 points of 3.5, second only to GPT-4. It can be seen that Wenxin model version 3.5 has outstanding Chinese ability, and even exceeds the performance of GPT-4, far ahead of other open source large models.
Kai-Fu Lee, Chairman and CEO of Innovation Works, said: "China has a rich Chinese corpus and a huge market, and through the development of AI models, China can promote the development of innovation industries and realize the dual dividends of science and technology and economy. #百度文心大模型多维度获国内第一#
