Deep Learning for Recommendations

Two-Tower Architecture

Dual model with separate towers to encode user and item features in a common embedding space. This architecture scales efficiently for millions of items thanks to pre-computed item embeddings.

← Geri