Relevant news indicates that Xiaomi is expanding the construction of its own GPU computing cluster and expects to invest in building its own large-scale natural language models to promote the development of its artificial intelligence application services.
Relevant sources said that Xiaomi had already invested in this project several months ago, and that it was led by Xiaomi CEO Lei Jun. However, Xiaomi did not respond to the rumors.
Prior to this, Xiaomi had already invested relevant resources in the development of artificial intelligence applications, including the early promotion of Xiao Ai, the establishment of an artificial intelligence team in 2016, and the later investment in the development of autonomous driving technology in vehicles. In April 2023, it established the Xiaomi Artificial Intelligence Laboratory's large natural language model team and recruited many technical talents to join, including Luan Jian, who was formerly a researcher at Toshiba China Research Institute, an advanced speech scientist at Microsoft China Academy of Engineering, and the chief language scientist of Microsoft Xiaoice, as the head of Xiaomi's large natural language model team.
Xiaomi has currently established MiLM-13B, a large-scale natural language model with 1.3 billion parameters on mobile phones. In specific scenarios, its computing performance is close to that of MiLM-60B, a large-scale natural language model with 6 billion parameters deployed in the cloud. The large-scale natural language model has also been applied to the new version of Xiao Ai, enabling it to interact in a more natural conversational way.
Xiaomi has undergone six expansions in the past seven years, and the number of relevant technical personnel has increased to more than 7. It has also expanded its artificial intelligence technology applications to cover vision, acoustics, voice, natural language processing, knowledge graphs, machine learning, large natural language models, and multimodal operations, and is also applied in mobile phones, smart IoT devices, robots, vehicles and other scenarios.
