Model Compression Toolkit User Guide

Overview

Model Compression Toolkit (MCT) is an open source project for neural networks optimization that enables users to compress and quantize models. This project enables researchers, developers and engineers an easily way to optimized and quantized state-of-the-art neural network.

Currently, MCT supports hardware-friendly post training quantization (HPTQ) with Tensorflow 2 [1].

MCT project is developed by researchers and engineers working in Sony Semiconductors Israel.

Install

See the MCT install guide for the pip package, and build from source.

From Source:

git clone https://github.com/sony/model_optimization.git
python setup.py install

From PyPi:

pip install model-compression-toolkit

A nightly version is also available:

pip install mct-nightly

Supported Features

Quantization:

  • Hardware-friendly Post Training Quantization [1]

  • Gradient base post training using knowledge distillation (Experimental)

Visualization:

Quickstart

Take a look of how you can start using MCT in just a few minutes

API Documentation

Please visit the MCT API documentation here

References

[1] Habi, H.V., Peretz, R., Cohen, E., Dikstein, L., Dror, O., Diamant, I., Jennings, R.H. and Netzer, A., 2021. HPTQ: Hardware-Friendly Post Training Quantization. arXiv preprint.