1. repeat()
Repeat the training data set number
2. filter
def filter (self, predicate): predicate is determined by the data set for screening.
'''
>>> dataset = tf.data.Dataset.from_tensor_slices([1, 2, 3])
>>> dataset = dataset.filter(lambda x: x < 3)
>>> list(dataset.as_numpy_iterator())
[1, 2]
'''
