Open Aboelenien opened 3 years ago
Sorry for the confusion. The loss_att is not being used in this implementation. We experimented with a number of composite loss functions but didn't use them in the final implementation. You can remove the "loss_att = 0" and "loss = loss + loss_att"
In this block in training, I don't get the loss_att and its use.
loss_att = 0 loss = criterion(y, target) loss = loss + loss_att