PaddlePaddle/PaddleRec
A large-scale library of recommendation system algorithms including DeepFM, BERT4Rec, DSSM, and multi-task learning models built with PaddlePaddle.

Velocity · 7d
+1.8
★ / day
Trend
→steady
star history
PaddleRec provides implementations of classic and state-of-the-art recommendation algorithms for both recall and ranking stages, including Wide&Deep, DeepFM, DIEN, MMoE, and GNN-based models. It includes built-in support for popular recommendation datasets like Criteo and MovieLens, and offers training configurations and model hyperparameter tuning tools for production and research use cases.