Hi,

I am trying to replicate your experiment and have noticed that there are three versions of munit such as munit.yaml, tiny_munit.yaml and big_munit.yaml and each is used in training MUNIT on a different dataset. E.g: tiny_munit.yaml is used for ColoredMNIST, whereas, big_munit.yaml is used to for FMoW. Therefore, I was wondering:

1. What is the difference between each? and what is the reason behind selecting munit vs tiny_munit vs big_munit?
2. Which version was used to produce the results reported in your paper?

Thanks.
Sara A. Al-Emadi