Why can't the subset be both simultaneously when generating the dataset?

Question

When I'm generating the dataset through tf.keras.preprocessing.image_dataset_from_directory, I'm getting the error. I'm following this source for learning Image classification from scratch](https://www.stackoverflow.com/) computer vision.

image_size = (180, 180)
batch_size = 128

train_ds, val_ds = tf.keras.preprocessing.image_dataset_from_directory(
    "PetImages",
    validation_split=0.2,
    subset="both",
    seed=1337,
    image_size=image_size,
    batch_size=batch_size,
)

Error:

ValueError: `subset` must be either "training" or "validation", received: both

Source