CARD combines cluster-level LoRA adapters and reward-guided decoding for scalable LLM personalization
Yutong Song · hf · 2026-09-28
CARD is a hierarchical framework addressing the tension between fine-grained personalization and scalable deployment of LLMs.
- Cluster adapters: users are clustered by shared stylistic patterns, with group-specific LoRA adapters for robust generalization and low-resource performance.
- Implicit preference learning: contrasts user-authored text with cluster-level generations to infer user-specific style preferences without manual annotation.
- Decoding-time injection: the base model stays frozen; personalization is injected only at decoding via lightweight preference vectors and low-rank logit corrections.
On LaMP and LongMP benchmarks, CARD beats baselines in generation quality while markedly improving efficiency and scalability.
More from Research
- Yoav Goldberg: 29 arXiv papers on JEV just two weeks after its limited release is 'not healthy' — yoavgo · 2026-09-28
- Why Data Efficiency Is Humans' Last Edge — Yet Massive Data Feeding Stays Cheapest — himanshustwts · 2026-09-28
- Harness-Zero: PKU, Google and HKUST Distill Agent Harnesses into Model Weights — AxSaucedo · 2026-09-28
- MultiFlow: coupled flow matching predicts single-cell perturbation responses in unseen cellular contexts — burny_tech · 2026-09-28
- Self-Replicating Programs Emerge from Random Noise in Open-Source Browser Remake of BRLabs' Computational Life — zzznah · 2026-09-28
- Pop-Corn preprint targets unseen-perturbation cell composition shifts in Perturb-seq — burny_tech · 2026-09-28