You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Implement the loss function described here, which achieves competitive results with more cited ones (symmetric negative cosine similarity & contrastive loss) while providing better interpretability.
The text was updated successfully, but these errors were encountered:
Thanks for facilitating this; ive made a fork on my account and made a feature branch there branching from dev. Any further details on how youd like to work towards a PR?
Since there are no dependencies with this issue, you can simply create a PR once you feel comfortable with your implementation and then we'll take a look. It would be good to rebase your feature branch onto our develop before you open the PR.
Add Alignment and Uniformity Loss
Following the discussion in #249 and as suggested by @EelcoHoogendoorn.
Implement the loss function described here, which achieves competitive results with more cited ones (symmetric negative cosine similarity & contrastive loss) while providing better interpretability.
The text was updated successfully, but these errors were encountered: