when training rpnet, division by zero

[Doylegyf]

Traceback (most recent call last): File "rpnet.py", line 405, in model_conv = train_model(model_conv, criterion, optimizer_conv, num_epochs=epochs) File "rpnet.py", line 395, in train_model print ('%s %s %s\n' % (epoch, sum(lossAver) / len(lossAver), time()-start)) ZeroDivisionError: division by zero

It seems lossAver=0.Has anyone encountered such a problem?

[Doylegyf]

Maybe it is because the wr2 network training is not good enough?

[Doylegyf]

The reason is loss.data[0],I changed it to loss.item()

[zhqx511]

Yes，you are right，just because of the torch version

[Yifanzhou-0713]

Same problem. Seems wR2 doesn't own 'model' attribute? Have you solved the problem?