Conditional Flows implementation / documentation about normalizing-flows HOT 2 CLOSED

Hi Prof. @cranmer,

there existed some support for flows conditioned on continuous variables scattered around the package, but it was incomplete and, indeed, not very well documented.

I added some functionality now, such as a ConditionalNormalizingFlow model class, a ConditionalDiagGaussian base distribution, where the parameters are encoded by the conditional, and a few flows support now conditional variables. An example for a conditional MAF and a conditional autoregressive NSF is given here; it is also in the documentation.

If there is any particular flow besides those implemented that you want to use with continuous conditional variables, please let me know.

Best regards,
Vincent

PS: Note that these additions are currently only available on the GitHub, but I will update the release on PyPI soon.

from normalizing-flows.

The changes are part of the new version of the package I just released on PyPI. Hence, I'll close this issue. If there are still more features you wish to be included, please create a new issue.

from normalizing-flows.