What is a good training loss?

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