AI RESEARCH
CARD: Cluster-level Adaptation with Reward-guided Decoding for Personalized Text Generation
arXiv CS.AI
•
ArXi:2601.06352v2 Announce Type: replace Adapting large language models to individual users remains challenging due to the tension between fine-grained personalization and scalable deployment. We present CARD, a hierarchical framework that achieves effective personalization through progressive refinement. CARD first clusters users according to shared stylistic patterns and learns cluster-specific LoRA adapters, enabling robust generalization and strong low-resource performance.