MTA: A Merge-then-Adapt Framework for Personalized Large Language Models

Personalized Large Language Models (PLLMs) aim to align model outputs with individual user preferences, a crucial capability for usercentric applications. However, the prevalent approach of fine-tuning a separate module for each user faces two major limitations: (1) storage costs scale linearly with the number of users, rendering the method unscalable; and (2) fine-tuning a static model from scratch often yields suboptimal performance for users with sparse data. To address these challenges, we propose MTA, a Merge-then-Adapt framework for PLLMs. MTA comprises three key stages. First, we construct a shared Meta-LoRA Bank by selecting anchor users and pre-training meta-personalization traits within meta-LoRA modules. Second, to ensure scalability and enable dynamic personalization combination beyond static models, we introduce an Adaptive LoRA Fusion stage. This stage retrieves and dynamically merges the most relevant anchor meta-LoRAs to synthesize a user-specific adapter on the fly, thereby removing the need to maintain a dedicated, persistently stored per-user adapter for each user and enabling flexible personalization. Third, we propose a LoRA Stacking for Few-Shot Personalization stage, which optionally applies an additional ultra-low-rank, lightweight residual LoRA module on top of the merged LoRA. This stacked module captures user-specific residual signals under few-shot settings. Extensive experiments on the LaMP benchmark demonstrate that our approach outperforms existing SOTA methods across multiple tasks. Our code is available at https://github. com/Applied-Machine-Learning-Lab/ ACL2026_MTA .