SVCCA on LSTM output

Hi everyone, I am kind of a newbie to PyTorch and ML. I have been trying to check for nodes correlation using the svcca module from google here. The examples work well and if I try to use random tensors converted into np arrays I’m able to get the svcca values. I then tried to get the activations from a layer of an LSTM net by adding a self.activations variable to the model, which saves the activations of the LSTM layer output. def forward(self, text): embedded = self.embedding(...