In the notebook covering batch normalization, the second half of the experiment was missing a model.eval() in val_loss function.
Due to which the graphs were not accurate.
The graph presented in page 147 shown below
should actually be
In the notebook covering batch normalization, the second half of the experiment was missing a model.eval() in val_loss function.
Due to which the graphs were not accurate.
The graph presented in page 147 shown below
should actually be