Skip to content

Error Consistency with Confidence Intervals

License

wichmann-lab/error_consistency

Repository files navigation

Imports: isort Code style: black

Quantifying Uncertainty in Error Consistency: Towards Reliable Behavioral Comparison of Classifiers

Requirements

This repo assumes a python version of at least 3.13. All other requirements are listed in requirements.txt and can just be pip-installed.

Contributing

Code Style

We use Black and isort for code formatting, and automate this using pre-commit. We keep notebooks clean using nbstripout.

To install requirements and apply the style:

pip install -r requirements.txt
pre-commit install
pre-commit run --all-files

If the CI fails for a PR you'd like to merge, it means that your changes don't adhere to the style guide. To fix this, run pre-commit run --all-files locally, then commit and push again. If black applies auto-formatting that is really nonsensical, you can force it to ignore a code block like this:

# fmt: off
print("...")
# fmt: on

Testing

We use pytest for unit testing. Install it with pip install pytest and then run pytest from the repository root to run all tests.

About

Error Consistency with Confidence Intervals

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published