Invertible conditional flows #84
-
Issue Description (feature request)Does flowtorch already support invertible conditional flows? That is, learning p(x | z). And given x and z, we get back the value sampled from the base distribution. Is there a reasonable way to do this already and if not, is it planned be supported any time soon? Thanks! |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 1 reply
-
@kirtanp thanks for the great question! First I can say that adding conditional flows is one of the highest items on our implementation agenda. We had to take it out after a major refactor and haven't gotten around to putting it back in yet. Separately, you can get the value sampled from the base distribution for analytically invertible flows with the |
Beta Was this translation helpful? Give feedback.
-
Hey, thanks for the great package! Really appreciate the effort As a followup on this thread I was wondering whether to this date conditional flows have been implemented or if there are plans for doing it in the near term. |
Beta Was this translation helpful? Give feedback.
@kirtanp thanks for the great question! First I can say that adding conditional flows is one of the highest items on our implementation agenda. We had to take it out after a major refactor and haven't gotten around to putting it back in yet.
Separately, you can get the value sampled from the base distribution for analytically invertible flows with the
.normalize()
and.rnormalize()
methods ofFlow
, which work independently from conditionality. See here: https://github.com/facebookincubator/flowtorch/blob/main/flowtorch/distributions/flow.py