The preprocessing of sequences can easily be a bottleneck in the Deep Learning pipeline. And tf.data.Dataset provides us ways to optimize the input pipeline. But there are l