Advanced Tutorials

The following tutorials are captured in Jupyter notebooks and are distributed as part of the dioptra git repository in the examples folder. These tutorials use the Python client and can be run interactively from the provided notebook files.

View the Tutorials

Click any of the links below to view the Jupyter notebook tutorials on GitHub.

Name	Description
Hello World with the Dioptra Python Client	This tutorial demonstrates how interact with Diopta via the Python Client. It walks through registering a queue, importing plugins and entrypoints, creating an experiment, and finally executing jobs.
Adversarial ML with OPTIC	This tutorial demonstrates how the OPTIC (Open Perturbation Testing for Image Classifers) plugin can be used to evaluate adversarial attacks and defenses on a model trained to recognize handwritten digits.

Run the Tutorials

To run the tutorials in a local deployment, you’ll need to do the following pre-requisites:

• Install Dioptra: Ensure you have access to a Dioptra deployment

• Setup the Python Client: Set up the Python Client

• Add a Dataset: Download and configure any datasets used by the tutorial

Then create a new virtual environment for running the tutorials and start Jupyter:

# Move into the examples folder of cloned repo
cd /path/to/dioptra/examples

# Create a new virtual environment at /path/to/dioptra/examples/.venv
uv venv

# Activate the virtual environment
source .venv/bin/activate

# Install the dependencies
uv pip install -r examples-setup-requirements.txt

# Run this in the examples/ folder
jupyter notebook

Once Jupyter starts up and is visible in your web browser, use the file explorer to navigate to open the tutorial you want to try.