Matrix dimensions do not match about notes HOT 4 CLOSED

sizes=[4,2,1] example:

from notes.

I guess if you feed x as a feature_size x batch_size vector, then there should be no problem. For example, if you want to feed an input with 4-dimensional feature space, then x should be of size (4, 1). You can easily do this by feeding x.view(feature_size, -1). This will assure that your input is of size (feature_size, batch_size).

from notes.

A detailed version of sizes=[4,2,1]:

from notes.

I think I have found the problem. I was creating inputs as torch.rand(sizes[0]), instead of torch.rand(sizes[0], 1). Second one solves the conflicts.

from notes.