Baidu Releases Wenxin Large Model 4.0 Turbo and PaddlePaddle Framework 3.0 at WAVESUMMIT 2024

TapTechNews June 28th, at the WAVESUMMIT Deep Learning Developers Conference 2024 held today, Baidu officially released the latest technologies such as the Wenxin Large Model 4.0 Turbo and the PaddlePaddle framework 3.0.

Wang Haifeng, Baidu's chief technology officer and director of the National Engineering Research Center for Deep Learning Technology and Application, said that as of now, the PaddlePaddle Wenxin ecosystem has attracted 14.65 million developers and serves 370,000 enterprises and institutions, creating 950,000 models.

Wang Haifeng today released the new generation of the PaddlePaddle framework - the PaddlePaddle framework 3.0, which has core technologies such as unified dynamic and static automatic parallelism, compiler automatic optimization, large model multi-hardware adaptation, and large model training and inference integration, supporting better results and superior performance for large models.

According to the introduction, the PaddlePaddle framework 3.0 also provides a fully functional and low-cost solution for large model hardware adaptation, and has built R & D infrastructure such as code merging, continuous integration, and model regression testing for hardware manufacturers, providing a complete guarantee for hardware adaptation.

TapTechNews found that the PaddlePaddle framework 3.0 Beta version has currently been launched on the official website, and it is recommended for developers with deep learning development experience, source code and security requirements.