ahundt opened this issue 7 years ago · 1 comments
currently when running the ResNet model on the grasp dataset the loss is nan, this must be fixed before training can run correctly.
nan
Fixed in #375