Welcome to cardinal’s documentation!¶
cardinal is a python package that allows simple and scikit-learn compatible active learning experiments. It provides the building blocks necessary to perform simple or more complex query sampling strategies.
The code of the project is on Github: cardinal
cardinal can be installed through Pypi using:
$ pip install "cardinal[extra]"
Extras bring additional features to cardinal:
sklearnallows to use the KMeansSampler and RankedBatchModeSampler
examplesintalls all packages necessary to run the examples
submodularinstalls apricot-select, and allows using SubmodularSampler
docinstalls sphinx-gallery to be able to generate the documentation
allis an alias on all of the above.
You can also fetch the code and install the package from your local repository. Again, the preferred way is to use pip.
$ git clone https://github.com/dataiku-research/cardinal $ cd cardinal $ pip install -e ".[extra]"