Now, finally, the NPU units in Intel processors will become useful? The company has opened the source code of its NPU Acceleration Library

The developers need this

Intel has open sourced its NPU Acceleration Library, allowing compact AI models to run directly on the NPU units in Meteor Lake processors. 

The library is created in Python and is designed to help developers improve their products, provided they run on the NPU units of the latest Intel processors. The library allows you to run compact language models like Gemma-2b or TinyLlama directly on the NPU, without requiring any cloud server. Actually, this is exactly why NPU hardware units are needed in processors at all, and what is currently missing.  

Recall that NPU units are found in Intel Meteor Lake and AMD Ryzen 7040/8040 processors. For now, these blocks are completely useless for most users.