search

marco-rudolph / differnet

This is the official repository to the WACV 2021 paper "Same Same But DifferNet: Semi-Supervised Defect Detection with Normalizing Flows" by Marco Rudolph, Bastian Wandt and Bodo Rosenhahn.
215 stars 67 forks source link

train loss #42

Closed emiyashiroz closed 1 year ago

emiyashiroz commented 1 year ago

Is it normal for the training loss to be less than 0?

marco-rudolph commented 1 year ago

Yes. The loss reflects the negative log likelihood that can be less than zero.