CI/CD for Machine Learning Development

Event Information

🗓️ This repository was created for a talk at the Keras Community Day by GDG Bandung on September 16, 2023 at BINUS Bandung.

Description

This repository demonstrates a complete machine learning workflow from development to deployment, leveraging GitHub Actions for Continuous Integration and Continuous Deployment (CI/CD).

Features

  • Automated model training and evaluation
  • Comparison of model metrics across different code branches
  • Automated unit and integration tests
  • Auto-commenting of evaluation metrics in pull requests by GitHub Actions

Requirements

To install the project dependencies, run the following command:

pip install -r requirements.txt

Usage

For Training

python main.py --train

For Testing

python main.py --test

For Experimentation with New Code (Training)

python main.py --train --experiment

For Experimentation with New Code (Testing)

python main.py --test --experiment

GitHub Actions Output

The metrics and evaluation will be automatically posted as comments by GitHub Actions in the pull requests.