Huawei Consumer Business Group, Smart Business Unit, and Huawei Cloud CEO announced that a prominent cloud product, Pangu is about to return to production.
According to Huawei, after two years of silence, the AI model of the Pangu series will be officially launched soon. Currently, the NLP large model, CV large model, and scientific computing large model in the Pangu large model has been marked as coming online.
Pangu NLP large model uses the Encoder-Decoder architecture for the first time, taking into account the understanding and generation capabilities of the NLP large model, ensuring the flexibility of embedding the model in different systems.
In downstream applications, only a small number of samples and learnable parameters are needed to complete the rapid fine-tuning and downstream adaptation of a large-scale model of 100 billion. This model has a good performance in intelligent public opinion and intelligent marketing.
The Pangu CV large model is the industry’s largest CV large model that establishes on-demand model extraction for the first time.
For the first time, it realizes both discrimination and generation capabilities, and can adaptively extract models of different scales based on model size and running speed requirements, and AI application development can be quickly implemented.
This enables it to obtain better separability in shallow features, significantly improves the ability of small sample learning, ranking first in the industry, and has a good performance in intelligent inspection and intelligent logistics.
As for the Pangu scientific computing large model, with the help of the innovative 3DEST network structure and layered time aggregation algorithm. The accuracy of the key elements of the weather forecast and the commonly used time range exceeds the current most advanced forecast method. Also, the speed is 1000 times faster than the traditional method. above.
In addition, the model supports a wide range of downstream forecasting schemes. For example, in the typhoon track prediction task, compared with traditional numerical weather forecasting methods, the Pangea meteorological large model can reduce the position error by more than 20%.