Introduction

Mobilint® Model Compiler (i.e., Compiler) is a tool that converts models from deep learning frameworks (ONNX, PyTorch, Keras, TensorFlow, etc…) into Mobilint® Model eXeCUtable (i.e., MXQ), a format executable by Mobilint® Neural Processing Unit (NPU). This is the qbcompiler, Mobilint’s SDK documentation. In this documentation, you can learn how to use it, what frameworks it supports, etc. A set of functions you can use to interact with the SDK will be given below. An input to qbcompiler is a trained deep learning model, its input shape, and calibration data. It will return MXQ (compiled model) as an output.