Marco Pegoraro

My Docker-based AI/ML Training Environment

mpeg -

Without further ado, here is the link to the GitHub project I've created to stretch my muscles into Machine Learning, Artificial Intelligence, and Python:

GitHub - marcopeg/learning-python: A containerized project to play around with Python and Jupiter Notebook
A containerized project to play around with Python and Jupiter Notebook - marcopeg/learning-python
GitHubmarcopeg

Run that on your computer, or spin it up as a GitHub Space. It works both ways.

You get:

  • A JupyterLAB server with its web interface on localhost:8888 running Python 3.9 and NodeJS 20.x kernels
  • A Postgres DB with pgvector already installed
  • A Docker Compose project that spins up containers
  • A Makefile interface to operate the project

You need:

Here are a few high-level commands to operate with the environment:

# Start the services
make start

# Stop the services
make stop

👉 Once the environment start, open your browser to http://localhost:8888

Managing Dependencies

As this thing works inside a Docker container, you will need to modify the relative Dockerfile in order to install dependencies for Python or NodeJS.

  • Open .jupiter/Dockerfile with your favourite editor.
  • Search for "Python Dependencies" or "Node Dependencies"
  • Add what you need
  • run the following command:
# Install the dependencies and restart the environment
make reset