timeseries_dataset_from_array why is dimension added to data but not target

Question

As the title states, why doesn't timeseries_dataset_from_array() add an extra dimension to the output?

print("data shape" , data[0].shape)
print("target shape" , targets[0].shape)

ds = tf.keras.preprocessing.timeseries_dataset_from_array(
    data=data,
    targets=targets,
    sequence_length=5,
    sequence_stride=1,
    shuffle=True,
    batch_size=32,)
print("ds shape ", ds.element_spec)

Output:

data shape (224, 304, 2)
target shape (12,)
ds shape (TensorSpec(shape=(None, None, 224, 304, 2), dtype=tf.float32, name=None), TensorSpec(shape=(None, 12), dtype=tf.float32, name=None))