Meet Dynamic Individual Preferences: Resolving Conflicting Human Value with Paired Fine-Tuning
This work addresses the problem of adapting LLMs to individual users with evolving and contradictory preferences, which is a practical bottleneck for personalized AI systems.
The paper tackles the challenge of aligning large language models with dynamic and conflicting individual preferences, introducing Preference-Paired Fine-Tuning (PFT) and the Value Conflict Dilemma (VCD) dataset. PFT achieves up to 96.6% accuracy in multi-choice classification and a 44.76% improvement in user-specific preference alignment over single-preference models.
Recent advances in large language models (LLMs) have significantly improved the alignment of models with general human preferences. However, a major challenge remains in adapting LLMs to individual preferences, which are not only diverse but also dynamic. In this paper, we introduce a novel framework, Preference-Paired Fine-Tuning (PFT), designed to align models with contradictory and evolving individual preferences. We present a new dataset, Value Conflict Dilemma (VCD), which includes scenarios that involve conflicting human preferences, facilitating the evaluation of our approach. Our experiments demonstrate that PFT outperforms single-preference training methods, achieving up to 96.6% accuracy in multi-choice classification tasks and the highest open-ended generation score of 8.69. PFT also shows significant improvements over DPO, SFT and some traditional training methods, especially when handling conflicting preferences. Additionally, with limited user history data, models can inferring preference vector rapidly, achieving a 44.76% improvement in user-specific preference alignment in comparison to single-preference models.