A demo on how to apply continuous delivery principles to train, test and deploy ML models.
Note:
- If you encounter any errors, please refer to FAQs for a list of common errors and how to fix them.
- [Windows users] If you're new to Docker, please use Git Bash to run the commands below
Setup instructions
- Please ensure you've completed the pre-requisite setup
- Fork repository: https://github.com/davified/ci-workshop-app
- Clone repository:
git clone https://github.com/YOUR_USERNAME/ci-workshop-app
- Start Docker on your desktop (Note: Wait for Docker to complete startup before running the subsequent commands. You'll know when startup is completed when the docker icon in your taskbar stops animating)
- Build docker image
# [Mac/Linux users]
docker build . -t ci-workshop-app --build-arg user=$(whoami)
# [Windows users]
MSYS_NO_PATHCONV=1 docker build . -t ci-workshop-app --build-arg user=$(whoami)
- Start docker container
# [Mac/Linux users]
docker run -it -v $(pwd):/home/ci-workshop-app -p 8080:8080 ci-workshop-app bash
# [Windows users]
winpty docker run -it -v C:\\Users\\path\\to\\your\\ci-workshop-app:/home/ci-workshop-app -p 8080:8080 ci-workshop-app bash
# Note: to find the path, you can run `pwd` in git bash, and manually replace forward slashes (/) with double backslashes (\\)
! Pre-workshop setup stops here
### Other useful docker commands ###
# See list of running containers
docker ps
# Start a bash shell in a running container when it’s running
docker exec -it <container-id> bash
Now you're ready to roll!
# Add some color to your terminal
source bin/color_my_terminal.sh
# Run unit tests
nosetests
# Train model
SHOULD_USE_MLFLOW=false python src/train.py
# Start flask app
python src/app.py
# Make requests to your app
# 1. In your browser, visit http://localhost:8080
# 2. In another terminal in the container, run:
bin/predict.sh http://localhost:8080
# You can also use this script to test your deployed application later:
bin/predict.sh http://my-app.herokuapp.com
Please refer to FAQs for instructions on configuring VS Code or PyCharm.
Instructions for setting up your CD pipeline are in docs/CD.md. To keep this example simple, we will deploy to heroku.
Once the CD pipeline is set up, you only need to git add
, git commit
and git push
your code changes, and the CD pipeline will do everything (train, test, deploy) for you.
You can also configure your CD pipeline to deploy using kubernetes instead. See instructions here