This layer performs the following steps 1) The matrix Psi is by permuting and concatenating the feature vectors in Xf, Xe and Xc 2) Psi is fed through an MLP. This is where all the learnable parameters in the network live. Thi
(self, Xf, Xe, Xc, Kf, Ke, Kc, Ce, Cf, Csf)
source not stored for this graph (policy: none)
nothing calls this directly
no test coverage detected