A Question about Research mi_lira_2021: why set training=True at Inference

[Shua-Kang]

Hi, I have a question about the code of Research mi_lira_2021.
It set training=True at inference:

privacy/research/mi_lira_2021/inference.py

Line 72 in 4dd8d0f

logits = model.model(this_x, training=True)

But since the model contains batchnorm layers, it should be training=True to avoid updating the parameters in the inference.
So why training=True here?