Asked about Ranking models and how to measure them
What is a two-tower model?
An answer that works
Separate networks embed users and items, and relevance is their similarity. Item embeddings are precomputed, so retrieval is a fast nearest-neighbour search. It is the same design as dense retrieval in RAG.
Reading an answer is not being able to give one. The section that teaches this is Ranking models and how to measure them, 4 hours of Recommendation systems.
Learn this properlyAlso asked about Recommendation systems