This section provides detailed instructions for running models using the Mobilint runtime library.
The Mobilint qbruntime tutorial assumes you are working on an inference PC equipped with a Mobilint NPU.
Note: The runtime environment does not need to be the same as the compilation environment. The runtime only requires a system equipped with a Mobilint NPU.
After connecting the hardware, start the Mobilint NPU driver to enable device access. Detailed instructions can be found in the Driver Installation Guide.
If the driver is successfully installed and you are using Docker, you can enable NPU access inside the container using the following flag:
--device /dev/aries0:/dev/aries0Next, install the runtime library. Refer to the Runtime Installation Guide for more information.
To install the runtime library in a Python environment, use the following command:
pip install mobilint-qb-runtimeDepending on your model type, you may need additional Python packages (e.g., torch, numpy, PIL, transformers). Refer to each specific model tutorial for detailed requirements.
Mobilint also provides a utility tool for checking NPU status, verifying MXQ files, and running simple inference tasks. Refer to the Utility Tool Installation Guide for details.
You are now ready to run your models! Explore the tutorials in this directory to run your compiled models on the Mobilint NPU.
