CARD: Cluster-level Adaptation with Reward-guided Decoding for Personalized Text Generation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911622237257728 |
|---|---|
| author | Song, Yutong Wu, Jiang Zhang, Weijia Shen, Chengze Yuan, Shaofan Lu, Weitao Wang, Jian Wang, Yu Dutt, Nikil Rahmani, Amir M. |
| author_facet | Song, Yutong Wu, Jiang Zhang, Weijia Shen, Chengze Yuan, Shaofan Lu, Weitao Wang, Jian Wang, Yu Dutt, Nikil Rahmani, Amir M. |
| contents | 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. To capture individual differences within each cluster, we propose an implicit preference learning mechanism that contrasts user-authored text with cluster-level generations, allowing the model to infer user-specific style preferences without manual annotation. At inference time, CARD injects personalization exclusively at decoding via lightweight user preference vectors and low-rank logit corrections, while keeping the base model frozen. Experiments on the LaMP and LongLaMP benchmarks show that CARD achieves competitive or superior generation quality compared to state-of-the-art baselines, while significantly improving efficiency and scalability for practical personalized text generation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_06352 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | CARD: Cluster-level Adaptation with Reward-guided Decoding for Personalized Text Generation Song, Yutong Wu, Jiang Zhang, Weijia Shen, Chengze Yuan, Shaofan Lu, Weitao Wang, Jian Wang, Yu Dutt, Nikil Rahmani, Amir M. Artificial Intelligence 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. To capture individual differences within each cluster, we propose an implicit preference learning mechanism that contrasts user-authored text with cluster-level generations, allowing the model to infer user-specific style preferences without manual annotation. At inference time, CARD injects personalization exclusively at decoding via lightweight user preference vectors and low-rank logit corrections, while keeping the base model frozen. Experiments on the LaMP and LongLaMP benchmarks show that CARD achieves competitive or superior generation quality compared to state-of-the-art baselines, while significantly improving efficiency and scalability for practical personalized text generation. |
| title | CARD: Cluster-level Adaptation with Reward-guided Decoding for Personalized Text Generation |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2601.06352 |