Questions

1. Given a tensor, a = [1, 2, 3], and another tensor, b = [4, 5, 6], how do you build a tf.data pipeline that would output each value separately, from 1 to 6?
2. According to the documentation of tf.data.Options, how do you make sure that a dataset always returns samples in the same order, run after run?
3. Which domain adaptation methods that we introduced can be used when no target annotations are available for training?
4. What role does the discriminator play in GANs?