Week 3 Assignment 2 Exercise 3

It says conv2d expects one tensor but received 2. All previous exercises pass grader.

What is the Input(input_size) function as well.

You need to use “cblock?[0]” to “chain the first element of the output of the previous block”

Substitute the block number for the ? in each line.


That creates the input layer.

Thank you!. It worked

Right. Notice that the output of the conv_block is a 2-tuple. The first tensor is the input to the next layer of the “contracting” path and the second one is the input for the shortcut path that will feed into the parallel layer on the “expanding path”.

But notice that on the “upsampling” path, the output is just one tensor, so you don’t have to play the “indexing” game with the output, as you do with the outputs of the contracting path.


Thanks for the explanation @paulinpaloalto , because I wondered about the lack of symmetry in the arguments there.