Free Inference Speed Gain with ONNX

ONNX Runtime is a cross-platform inference and training machine-learning accelerator. Learn more about it here.

This repo demonstrates how converting a model saved in H5 format to .onnx for a free speed gain. The model used is a simple neural network trained on fashion_mnist dataset.

Table of Contents

• Dependency
• Python Packages
• Test It Out

Dependency

1. Python 3.8+

Python Packages

pip3 install -r requirements.txt

Test It Out

Run all cells in onnx.ipynb