I am new to pytorch.
I am trying to perform
sij = Wq([wi ; wj]) where
wj are the word embeddings and trying to concatenate every word with other words in a sentence. I thought of doing it by repeating the word embeddings using the
expand() as described here Expand a 2d tensor to 3d tensor. But the issue is my output is 3D tensor
batch, len, dimension so I am not sure if I should make them 4D tensor.