2026-10-26 · America/Los_Angeles · 论文 · #20

DuoLM: Fast and Effective Personalized Recommendation with Language Models

Language models can make personalized recommendations from rich item content, user history, and feedback. However, evaluating many candidates with full-text inputs incurs substantial online computation and limits serving throughput. We introduce DuoLM, a framework for fast and effective personalized recommendation that integrates a compact recommendation model with an efficient serving runtime. An encoder LLM learns compact item representations for offline reuse, while a personalized ranker LLM combines them with user history and feedback for recommendation. We train both models end to end…

热度 38.0 / 100；排名与评分保留该期记录。