Set generator to eval mode before sampling?

[prabaey]

Environment details

If you are already running CTGAN, please indicate the following details about the environment in
which you are running it:

• CTGAN version: 0.7.4
• Python version: 3.10.12
• Operating System: Linux

Problem description

I noticed there is no line of code that sets the self._generator module in evaluation mode before sampling. I would expect this to be necessary, since there's Batchnorm layers in the generator that operate differently under training mode vs. evaluation mode. Something like this at the start of the CTGAN.sample() method:

self._generator.eval()

Why is this not needed? Am I missing something here? Thanks!