Comparing activation functions II

The code used in the previous exercise has been executed to obtain theactivation_results with the difference that 100 epochs instead of 20 are used. That way you'll have more epochs to further compare how the training evolves per activation function.

For every history callback of each activation function in activation_results:

The history.history['val_loss'] has been extracted.
The history.history['val_acc'] has been extracted.
Both are saved in two dictionaries: val_loss_per_function and val_acc_per_function.

Pandas is also loaded for you to use as pd. Let's plot some quick comparison validation loss and accuracy charts with pandas!

Create a pandas dataframe from val_loss_per_function.
Call plot() on the dataframe.
Create a pandas dataframe from val_acc_per_function.
Once again, call plot() on the dataframe.

Take Hint (-30 XP)

Exercise

Comparing activation functions II

Instructions