Sumit on Mythryl: “Personalized LLM Recommendation with Semantic IDs: Autoregressive…”

Personalized LLM Recommendation with Semantic IDs: Autoregressive Beam Search vs. Multi-Token Prediction at Industry Scale Netflix compares beam search & multi-token prediction for generating Semantic IDs in LLM recommenders. Beam search wins on ranking. 📝 arxiv.org/abs/2610.15864

arxiv.orgPersonalized LLM Recommendation with Semantic IDs: Autoregressive Beam Search vs. Multi-Token PredictionLarge-scale personalization is shifting from embedding-table retrieval to LLM-based recommenders that generate the Semantic ID (SID) of the item to recommend, where each SID is a short sequence of dis...

14/10/26, 8:46 am0 reposts0 likes

Post