...
plot_model(model, show_shapes=True, show_layer_names=True)
# The student can have his/her model saved: model_file = 'ANN_model.h5' ##Call functions implementation to monitor the chosen metrics checkpoint = keras.callbacks.ModelCheckpoint(filepath = model_file, monitor = 'val_loss', mode='min', save_best_only = True) #Stop training when a monitored metric has stopped improving early_stop = keras.callbacks.EarlyStopping(monitor = 'val_loss', mode='min',# quantity that has to be monitored(to be minimized in this case) patience = 50, # Number of epochs with no improvement after which training will be stopped. min_delta = 1e-7, restore_best_weights = True) # update the model with the best-seen weights #Reduce learning rate when a metric has stopped improving reduce_LR = keras.callbacks.ReduceLROnPlateau( monitor = 'val_loss', mode='min',# quantity that has to be monitored min_delta=1e-7, factor = 0.1, # factor by which LR has to be reduced... patience = 10, #...after waiting this number of epochs with no improvements #on monitored quantity min_lr= 0.00001 ) callback_list = [reduce_LR, early_stop, checkpoint] #callback_list = [checkpoint] #callback_list = [ early_stop, checkpoint]
...