Model authors sometimes use nn.Module.training to change the control flow of their model. This is problematic because we often make the assumption a model in training more or less produces the same result as in eval mode. We should detect when this is not the case and warn the user so they can take appropriate action!
Model authors sometimes use
nn.Module.trainingto change the control flow of their model. This is problematic because we often make the assumption a model in training more or less produces the same result as in eval mode. We should detect when this is not the case and warn the user so they can take appropriate action!