In this release we provide many additional features, including estimation of the KL divergence between the marginals, different units of measure available and compatibility with the most recent `sklearn`.
Note you will have to adapt your code if you use this new version; the fastest way is to simply replace any `fit` calls with `fit_estimate`, and you should get the same results as before. This is because we split the fitting part of the class with the estimation part, for increased efficiency and flexibility.
Get in touch if you need any help!