Text as a Universal Interface for Transferable Personalization
This addresses the issue of opaque and non-transferable user preferences in LLMs, offering a more interpretable and flexible solution for personalization, though it is incremental in improving existing methods.
The paper tackles the problem of personalization in large language models by proposing natural language as a universal interface for preference representation, leading to interpretable and reusable profiles; experiments on nine benchmarks show that their 8B model achieves state-of-the-art performance, outperforming larger open-source models with strong transferability across tasks and models.
We study the problem of personalization in large language models (LLMs). Prior work predominantly represents user preferences as implicit, model-specific vectors or parameters, yielding opaque ``black-box'' profiles that are difficult to interpret and transfer across models and tasks. In contrast, we advocate natural language as a universal, model- and task-agnostic interface for preference representation. The formulation leads to interpretable and reusable preference descriptions, while naturally supporting continual evolution as new interactions are observed. To learn such representations, we introduce a two-stage training framework that combines supervised fine-tuning on high-quality synthesized data with reinforcement learning to optimize long-term utility and cross-task transferability. Based on this framework, we develop AlignXplore+, a universal preference reasoning model that generates textual preference summaries. Experiments on nine benchmarks show that our 8B model achieves state-of-the-art performanc -- outperforming substantially larger open-source models -- while exhibiting strong transferability across tasks, model families, and interaction formats.