Installation¶
The app runs in the browser at faircombine.streamlit.app without any installation. To run it on your own computer, e.g., to assess unpublished models without uploading them, install the python package.
From PyPI¶
The package needs python 3.14. With uv the app is installed as a command line tool in an environment of its own:
fairvis starts the app and opens it in the browser, by default at http://localhost:8501. Options are passed to streamlit run, e.g., another port:
The defaults are the ones of the deployed app: no usage statistics, the light theme, uploads of at most 1 MB and errors shown without traceback. An option overrides its default, e.g., fairvis --client.showErrorDetails=full for the full traceback.
Without uv, install with pip into a virtual environment:
python -m venv .venv
source .venv/bin/activate # Linux and macOS
.venv\Scripts\activate # Windows
pip install fair-ca-visualization
fairvis
python -m fairvis starts the app as well.
The export of figures needs Chrome or Chromium, see Exporting a figure. Without one, download a Chrome for the export into the data directory of your user, where the app finds it:
uv tool run --from fair-ca-visualization plotly_get_chrome # or plotly_get_chrome in the activated virtual environment
From a checkout¶
To run the latest development version:
git clone https://github.com/matthiaskoenig/fair-ca-visualization.git
cd fair-ca-visualization
uv sync
uv run streamlit run src/fairvis/app.py
./run_app.sh is a shortcut for the last command. See Contributing for the development setup.