New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Output activation and bi-tempered loss #6

Closed

shkarupa-alex opened this issue Sep 2, 2020 · 1 comment

shkarupa-alex commented Sep 2, 2020

Am i right that with bi-tempered loss last model layer activation should be linear? (not sigmoid or softmax as usual case in keras)

My suggestion comes from:

https://github.com/google/bi-tempered-loss/blob/master/index.js#L373
https://github.com/google/bi-tempered-loss/blob/master/tensorflow/loss_test.py#L65 (sigmoid output can't be < 0)

Collaborator

eamid commented Sep 2, 2020

Yes, you should view the activations the same way as the "logits" in the last layer that you pass to the cross entropy loss.

eamid closed this as completed

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment