val_loss in keras - for the whole validation dataset or just the last batch?

Question

When training a CNN with keras, the loss will be displayed during training per batch, however the validation loss will only be displayed at the end, when the whole validatio