作者位次 #2/3ACL 2026Findings· Findings of ACL
To Intervene or Not: Guiding Inference-time Alignment with Probabilistic Model Blending
摘要
The wide deployment of LLMs has made model alignment necessary to make newly trained models safely and effectively respond to user instructions. Among different methods, inference-time alignment is often cheaper as it intervenes (i.e., offers guidances) only during output generation. Existing pro...