Huawei Prepares to Reveal Latest Advances in Pangu AI Models
Huawei plans to showcase the progress and applications of its Pangu large-scale AI models at an upcoming technology summit. The Pangu series encompasses various AI disciplines and shows strong performance in complex commercial scenarios.
Huawei Cloud's Chief AI Scientist Qi Tian prepares to present the latest progress and applications of the Pangu large-scale models at the AI Large-Scale Model Technology Summit in Hangzhou. This highly anticipated presentation highlights Huawei's ongoing commitment to advancing artificial intelligence technology. The tech giant first launched the Pangu series in April 2021 after beginning its investment in large-scale models back in 2020.
The Pangu portfolio includes a diverse range of AI systems, such as computer vision, natural language processing, scientific computing, voice recognition, and multimodal models. Notably, the NLP model learns from 40TB of Chinese text data to achieve near-human comprehension, while the computer vision model ranks first in small sample learning abilities on ImageNet. Additionally, the scientific computing model actively serves specialized fields like meteorology and biomedicine.
According to a CITIC Securities research report, the practical capabilities of the Pangu models already see verification in sectors like energy consumption and retail finance. By utilizing few-shot learning, transfer learning, and technologies like P-tuning, Huawei tailors these models for complex commercial scenarios rather than just general tasks. Furthermore, Huawei recently demonstrates the adaptability of this technology by miniaturizing a multimodal model to power a new smart search function for smartphone users.