About WeightEMA

[YilinLiu97]

Hi @Britefury, in optim_weight_ema.py,
for p, src_p in zip(self.params, self.src_params): p.data[:] = src_p.data[:] this line doesn't seem to really deep copy the student model, so teacher and student may be initialized differently.

[Britefury]

Hi,

Yeah, I think I know what you mean.
Thanks for spotting it.
My mean teacher models in my other projects now use state_dict() and load_state_dict() for this reason.