Forrest: ByteDance's Recommendation Architecture for a Feed That Cannot Be Stale
How ByteDance's Forrest-style recommendation stack splits retrieval, ranking, and real-time features so a For You feed can update as user behavior arrives.
3 min read
Tag
2 articles · 6 min of reading
How ByteDance's Forrest-style recommendation stack splits retrieval, ranking, and real-time features so a For You feed can update as user behavior arrives.
How ByteDance's Monolith system trained sparse recommendation models online, with collisionless embeddings and a sync path from trainer to serving.