BN update
jizongFox opened this issue · 4 comments
Hi @perrying
Thanks for sharing this code.
I saw that you have implemented the track_bn_statistic method in your network by overriding the foward
method. This seems strange and what is the propose of using such way.
I noticed that track_running_stats=False
does not work in this case.
In my implementation, the model is inputted the same data twice in one iteration.
Thus, I'd like to prevent updating the running_mean
and running_var
with the same data.
There may be better ways to implement it, but overriding is a comprehensible way for me.
So in this implementation, you ignore totally the unlabeled data, instead of considering it once.
The unlabeled data are also used to calculate running_mean
and running_var
.
What do you concern?
In this particular setting, there is little difference.