Personalized Adaptation via In-Context Preference Learning

TL;DR

This paper introduces PPT, a transformer-based method that enables scalable, personalized adaptation of language models through in-context learning and online user feedback, outperforming existing approaches.

Contribution

The paper presents a novel two-phase approach combining offline training and online in-context adaptation for personalized language model tuning.

Findings

PPT achieves superior personalization in a contextual bandit setting.

The method reduces computational costs compared to existing personalization techniques.

PPT effectively adapts to individual preferences in real-time.

Abstract

Reinforcement Learning from Human Feedback (RLHF) is widely used to align Language Models (LMs) with human preferences. However, existing approaches often neglect individual user preferences, leading to suboptimal personalization. We present the Preference Pretrained Transformer (PPT), a novel approach for adaptive personalization using online user feedback. PPT leverages the in-context learning capabilities of transformers to dynamically adapt to individual preferences. Our approach consists of two phases: (1) an offline phase where we train a single policy model using a history-dependent loss function, and (2) an online phase where the model adapts to user preferences through in-context learning. We demonstrate PPT's effectiveness in a contextual bandit setting, showing that it achieves personalized adaptation superior to existing methods while significantly reducing the computational costs. Our results suggest the potential of in-context learning for scalable and efficient personalization in large language models.