What is the purpose of the tf.pad operation? Can it be combined with conv2d or max_pool? #13
Description
I notice that tf.pad is used in the code. What is the purpose? Can it be combined with conv2d or max_pool?
For example:
In model.py,
padded = tf.pad(conv1, [[0, 0], [0, 1], [0, 1], [0, 0]], "CONSTANT")
In shufflenet_unit,
padded = tf.pad(shuffled, [[0, 0], [1, 1], [1, 1], [0, 0]], "CONSTANT")
Looking forward to your reply！