Tag
This paper introduces CLIPer, a method for personalizing LLM outputs at inference time using a classifier, avoiding the computational cost of extensive fine-tuning.