(Adapted from Rui Vieira's https://github.com/ruivieira/trustyai-odh-notebook-demo)
This notebook demonstrates the core features of the TrustyAI python package from a Jupyter Notebook. Namely, this demo covers:
- An intro to AI explainability algorithms
- Applying these algorithms to a real world model
- Tyrus, the TrustyAI explainability assistant and dashboard
- Model bias metrics
Building from source:
docker build . -t YOUR/TAG
docker run -it -p 8888:8888 YOUR/TAG
Run a pre-built version:
docker run -it -p 8888:8888 quay.io/trustyai/trustyai-notebook-demo
This will launch a Jupyter notebook server within a docker image; look for a message in the output to find the URL to the notebook server:
To access the notebook, open this file in a browser:
file:///[PATH].html
Or copy and paste one of these URLs:
http://[URL]:8888/?token=$token
or http://127.0.0.1:8888/?token=$token
Within the server, run the demo.ipynb
notebook to access the demo.