First! Vivo's self-developed AI large model ranks first in C-Eval and CMMLU

Domestic self-developed large AI models once again ushered in significant breakthroughs.

According to news on October 16, vivo will release a matrix of self-developed AI large models, including 5 self-developed large models with three different parameter levels: billion, tens of billions, and hundreds of billions, fully covering core application scenarios. The latest data shows that vivo’s self-developed AI large model ranks first in the global Chinese rankings of C-Eval and CMMLU, and its performance in humanities, social sciences and other fields far exceeds that of large models of the same level.

The C-Eval list is a comprehensive examination evaluation set for Chinese language models jointly constructed by Tsinghua University, Shanghai Jiao Tong University and the University of Edinburgh. It covers 52 different disciplines and has a total of 13,948 multiple-choice questions. It is currently the most comprehensive test evaluation set for Chinese language models. Authoritative Chinese AI large model evaluation list. The CMMLU data set is a comprehensive Chinese evaluation benchmark jointly launched by MBZUAI, Shanghai Jiao Tong University, and Microsoft Research Asia. It is extremely authoritative in evaluating the knowledge and reasoning capabilities of language models in the Chinese context.

According to the relevant person in charge, vivo’s self-developed AI large model will be used for the first time in the upcoming OriginOS 4 system, bringing consumers a more intelligent, convenient and secure mobile phone experience.

At present, AI large model technology is developing rapidly, promoting disruptive changes in social production and lifestyle. In the mobile phone industry, it is also expected to become a key opportunity for manufacturers to accelerate product iteration and open up blue ocean tracks. This time vivo created a self-developed AI large model matrix and applied it to the new system, proving that its exploration of large models has advanced from the technology research and development stage to the application and industrial layout stage. This will not only effectively promote vivo's own business growth and The implementation of the high-end strategy will have a very positive driving effect on the entire industry.

Source: Observer Network

The above is the detailed content of First! Vivo's self-developed AI large model ranks first in C-Eval and CMMLU. For more information, please follow other related articles on the PHP Chinese website!