Model.eval() effect on layers contained in nn.Sequential()

autograd
1

Does setting model to model.eval() prior to validation still deactivate appropriate functions (Dropout, batch normalization) if the functions are contained in a nn.Sequential() layer? Like so:

self.classifier = nn.Sequential(
nn.Dropout(p=0.4, inplace=True),
nn.Linear(in_features=888, out_features=64),
nn.ReLU(),
nn.Linear(in_features=64, out_features=num_classes),
)

2

Yes, as the the eval will be called on each child module.
Do you see any weird behavior?