alfredodeza / huggingface-onnx

Example notebook to work with ONNX using Hugging Face libraries

Using 🤗 Hugging Face with ONNX

This repository contains example notebooks to work with ONNX using 🤗 Hugging Face libraries and tools. It is part of the Duke University MLOps Coursera Specialization

Exercise

This repository is ready to use with GitHub Codespaces. All dependencies will be installed and the notebooks will be ready to use. If you are running this locally and without Codespaces, you'll need to follow these steps:

1. Create a new Conda environment and install the dependencies listed in environment.yml
2. Open the Jupyter Notebooks and select the Python from the Conda environment created

With the environment in place, go through the notebooks and use the examples that already work and then try these challenges:

1. Use a different model to port over to Hugging Face
2. Try a specific feature for a different model
3. Load the resulting ONNX model with the ONNX Runtime and try using it

Objectives

This introductory repository will expose you to learn how does ONNX and Hugging Face works together. It shows what and how you install the necessary libraries to create ONNX models.

Learning Resources

There are a few sources that are useful to dive deeper into working with ONNX and Hugging Face. Here are a few you can explore:

• The ONNX Model Zoo
• Exporting to ONNX - Hugging Face documentation
• Main ONNX landing page
• ONNX documentation

Related Coursera courses

• Linux and Bash for Data Engineering
• Open Source Platforms for MLOps
• Python Essentials for MLOps
• Web Applications and Command-Line tools for Data Engineering
• Python and Pandas for Data Engineering
• Scripting with Python and SQL for Data Engineering