Zhao Ming: Honor will be able to deploy a 7B device-side AI large model on a 12GB memory phone and ensure a smooth experience

On April 24, Wall Street News reported that Honor CEO Zhao Ming revealed that Honor will be able to deploy a large AI model with a parameter scale of 7B (7 billion parameters) on the end-side of a 12GB memory smartphone in 2024.

Zhao Ming said that Honor’s platform-level AI technology can achieve ultra-high compression rates for large LLM models on the device side. The 7B model can start quickly on 12GB memory devices while ensuring that users’ commonly used application experience is not affected. According to previous reports, at the Honor MagicOS 8.0 conference held in January, Zhao Ming unveiled Honor’s self-developed 7 billion parameter platform-level AI model “Magic Model” on the device side, which was first launched by the Honor Magic 6 mobile phone. At the same time, the Honor Magic Model was announced to be open to developers and launched the Honor Hundred Model Ecological Plan, which aims to “gather the power of China’s AI, provide developers with broader opportunities, and provide users with An unprecedented smart experience.” Zhao Ming said that all existing operating systems are worth redoing with AI. At present, most mobile phone manufacturers have begun to implant large AI models into their own ecosystems, allowing mobile phone AI to understand more functions, so as to continuously learn and optimize, and facilitate users' daily life. With the support of large AI models, Honor mobile phones can achieve simpler video creation with natural semantic understanding and fuzzy semantics Understand more accurate gallery search, drag the schedule, and generate schedules immediately.

The above is the detailed content of Zhao Ming: Honor will be able to deploy a 7B device-side AI large model on a 12GB memory phone and ensure a smooth experience. For more information, please follow other related articles on the PHP Chinese website!