“instancenorm” in encode stage

[amberhappy]

The instancenorm para about "track_running_states=True" and "affine=True".

In the training stage, should I set "track_running_states=True" and "affine=True"?

[Rivendile]

In our experiment, we leave these two parameters as False, which is default in PyTorch.
When set track_running_states=True, it will track the mean and variance. When False, it will only use batch statics.
When set affine=True, it will have some learnable parameters in instance norm.
You can try these settings if interested.