Different weights for calculating probability

[ankit0905]

In the file, “pubmed_inductive_appr2layers.py”, on line 123, you have used the adjacency matrix for the batch of nodes only to calculate the weights to be used for calculating probabilities for each node. While to calculate the same on line 128 (p2), you have used the adjacency matrix for the entire training nodes. Even in the other files, you have p0, as defined on line 107 in this file for the same.

So, shouldn’t it be same throughout and so p1 should be replaced by p0 rather in this file ?