What classifies as a good training loss for RESNET based classification?
Have a look at this post.
Even though the code is in Keras, the explications are quite good and you can easily find the corresponding losses in the Pytorch Docs
1 Like