4

Baidu Wenxin Model is the best rated large language model, IDC report says - Pin...

 1 year ago
source link: https://en.pingwest.com/w/11976
Go to the source link to view the article. You can view the picture content, updated content and better typesetting reading experience. If the link is broken, please click the button below to view the snapshot at that time.
neoserver,ios ssh client

Baidu Wenxin Model is the best rated large language model, IDC report says

Baidu Wenxin Model is the best rated large language model, IDC report says

July 19, 2023 9:58 am

Baidu Wenxin Big Model recognized as the most powerful China-made large language model (LLM) in terms of comprehensive technical capabilities, according to IT market research and consulting company IDC.

IDC recently released the “AI Large Model Technical Capability Assessment Report, 2023”, in which the company conducted an assessment of the comprehensive technical capabilities of major language models using 12 indicators.

Baidu Wenxin Large Model has achieved a perfect score in 7 out of 12 indicators and secured the top position in the overall evaluation among. Particularly, its algorithm model and general capabilities excel in the fundamental technological aspects, leading the industry. The algorithm model received the only perfect score among all evaluated models.

The report highlights Baidu's strengths in model capabilities, tool platforms, ecological layout, and industry coverage, underscoring its efforts in commercialization and exploration.

Baidu has been deeply involved in the development of pre-training models since 2019 and has successively released the Knowledge-Enhanced Wenxin series of models. Recently, Baidu officially launched the Wenxin Large Model 3.5 version, improving technological capabilities of Wenxin Large Model such as base models, knowledge enhancement, and retrieval enhancement.

According to Baidu's Chief Technology Officer, Wang Haifeng, since the release of the Wenxin Large Model 1.0 version in March 2019, Baidu has undergone four years of technical refinement and research iterations. During this period, they achieved upgrades to the base model, resulting in a 50% improvement in model performance, a 2x increase in training speed, and a 30x increase in inference speed.

Baidu also has a self-developed deep learning platform Flying Paddle, which strongly supports the efficient training and reasoning of large models.

During the speech at the 2023 World Artificial Intelligence Conference, Wang Haifeng revealed that as of now, Flying Paddle (also known as "飞桨" in Chinese) has gathered 7.5 million developers.

SHARE


About Joyk


Aggregate valuable and interesting links.
Joyk means Joy of geeK