Netflix Trains a Foundation Model for Personalized Recommendations

Explore how Netflix develops a foundation model for personalized recommendations in this 32-minute video by Aladdin Persson. Learn about the motivation behind foundation models, Netflix's data-centric approach, available data sources, and tokenization using BPE analogy. Discover techniques for improving training efficiency, understand the model's objectives, and examine unique challenges faced during development. The video also covers scaling strategies for recommendation models and concludes with key takeaways. For machine learning consulting inquiries, contact the presenter at [email protected].