Available information suggests that Huawei may release this software under the name “Pangu Chat”, a name derived from its large-scale learning model. This AI technology is expected to be unveiled at the Huawei Developer Conference (Cloud) to be held on July 7, 2023.
However, additional details indicate that Huawei will only make the software available for business purposes, selling it to government and enterprise customers. Huawei is reported to have tested and verified the operation of Pangu Chat using a large-scale Pangu cloud model with 100 billion parameters.
For the tests, the company used more than 2000 Huawei Ascend 910 processors for more than 2 months for training data. According to industry analysts, more than 4000 GPU/TPU boards are used in large-scale model training per year, and the computational power cost of large-scale models in three years has reached as high as 960 million yuan.
The Pangu model
Officially released in 2021, with version 2.0 to be introduced in April 2022, the Pangu model is contributing to AI development. It is the first Chinese pre-training model with more than 100 billion parameters and the CV model reached 3 billion parameters.
Pangu managed to combine discrimination and generation capabilities, as well as small-sample learning, second-level weather forecasts, and large-scale learning on maps, text, and audio.
The Pangu NLP model uses the Encoder-Decoder architecture, taking into account the large-scale understanding and generation capabilities of the NLP model.
Future of Huawei and AI
Although Huawei has not yet officially announced these news, it is positive to see that the company is using the power of AI to offer better solutions. Launching a rival to ChatGPT is just one example of this.
However, it would be even more interesting if they managed to make this feature available on Huawei smartphones.