search

BioroboticsLab / IBA

Information Bottlenecks for Attribution
MIT License
75 stars 9 forks source link

When training readout, loss is Nan #41

Open lipanUestc opened 2 years ago

lipanUestc commented 2 years ago

First of all, thank you very much for your open source code! Your work is great! But when I ran the code in "pytorch_IBA_trainreadout.ipynb", I found information Loss is all Nan after the first epoch. Is there something wrong with me? Could you give me some advice?

lipanUestc commented 2 years ago

I found that the reason for this problem was that I changed the dataset. Is this problem caused by the estimator's evaluation of mean and variance? What should I do when I want to change the dataset?