Skip to content
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.

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

Dividing contrastive loss by number of replicas -- correct? #202

Open
piotrmwojcik opened this issue Jul 31, 2022 · 0 comments
Open

Dividing contrastive loss by number of replicas -- correct? #202

piotrmwojcik opened this issue Jul 31, 2022 · 0 comments

Comments

@piotrmwojcik
Copy link

I think that in case when we are using only contrastive loss, dividing it by the number of replicas isn't doing what it is supposed to do. For every replica, the numerator consists of positive pairs for one replica-batch and the denominator contains every negative pair from the enlarged batch.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant