Margin Ranking loss in the wild

[rhsimplex (Ryan Henderson)]

Hi all,

I’m a big fan of margin ranking loss for regression-ish problems. I couldn’t find any writeups on torch’s MarginRegressionLoss function outside of the context of contrastive embeddings, so I wrote my own:

A Neural Network Trick for Ranking Problems

Posting here in the hopes it helps someone out.

Thanks!

[ptrblck]

Thanks for sharing the post!