With the residual block in residual neural networks, is the addition at the end of the block true element addition or is it concatenation?
For example, would addition([1, 2], [3, 4])
produce [1, 2, 3, 4]
or [4, 6]
?
With the residual block in residual neural networks, is the addition at the end of the block true element addition or is it concatenation?
For example, would addition([1, 2], [3, 4])
produce [1, 2, 3, 4]
or [4, 6]
?
It would result in [4, 6], and you can find out more in this paper