Microsoft's Mu is a new small language model tailored for Windows Settings, enhancing the search function. Designed for hardware optimization, Mu operates on neural processing units, supporting local inferencing. It is primarily trained on the Windows Settings app, enabling direct access to specific functions through user questions. Unlike previous models, Mu does not have public APIs yet, but it is integrated into Windows 11 on selected devices. Its encoder-decoder architecture provides efficiency for performance on limited hardware, distinguishing it from larger models.
Mu is a micro-sized, task-specific language model developed by Microsoft to enhance the Windows Settings search function. It utilizes an encoder-decoder architecture.
Trained specifically on the contents of the Windows Settings app, Mu provides direct links to relevant functions based on user inquiries, streamlining user interactions.
Collection
[
|
...
]