GitHub - recommenders-team/recommenders: Best Practices on Recommendation Systems

Abstraction: Microsoft open-source repository of recommendation algorithm best practices

Key points:

• 20,000+ GitHub stars; under Linux Foundation of AI and Data; v1.2.1 as of snapshot
• Provides Jupyter notebook examples covering 30+ algorithms including ALS, NCF, xDeepFM, BiVAE, BPR, LightGCN, SAR, SVD, and Transformer-based SASRec/SSEPT
• Covers five tasks: data prep, modeling, evaluation, hyperparameter tuning, and Azure operationalization
• Benchmark on MovieLens 100k: BiVAE achieves MAP=0.146, BPR MAP=0.132, NCF MAP=0.108 (top performers)
• Algorithms marked with * are Microsoft inventions/contributions (SAR, xDeepFM, DKN, GeoIMC, etc.)
• Recommended tooling: uv for environment management, VS Code for development

Connections: Microsoft · Github · Recommendation Systems · Collaborative Filtering

Source: https://github.com/microsoft/recommenders