search

hexiangnan / adversarial_personalized_ranking

Adversarial Learning, Matrix Factorization, Recommendation
https://github.com/hexiangnan/adversarial_personalized_ranking
215 stars 62 forks source link

fix bug on reg #8

Closed AaronHeee closed 5 years ago

AaronHeee commented 5 years ago

Fix bugs on L2 regularization implementation.

Add reg on embedding_p, embedding_q and embedding_q_neg respectively, instead of the entire embedding matrix.