# Online LLM Personalization from Pairwise Preferences Across Users

arXiv: https://arxiv.org/abs/2610.17203 · HTML: https://mlsanity.com/paper/2610.17203/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 43 of its 61 references that are in the map.
- First posted on arXiv 12 Oct 2026.

## Builds on (most cited first, 30 of 43)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Large Language Models are not Fair Evaluators](https://mlsanity.com/paper/2305.17926/) (2023)
- [Length-Controlled AlpacaEval: A Simple Way to Debias Automatic Evaluators](https://mlsanity.com/paper/2404.04475/) (2024)
- [Whose Opinions Do Language Models Reflect?](https://mlsanity.com/paper/2303.17548/) (2023)
- [Out of One, Many: Using Language Models to Simulate Human Samples](https://mlsanity.com/paper/2209.06899/) (2022)
- [AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedback](https://mlsanity.com/paper/2305.14387/) (2023)
- [Using Large Language Models to Simulate Multiple Humans and Replicate Human Subject Studies](https://mlsanity.com/paper/2208.10264/) (2022)
- [Large Language Models Are Not Robust Multiple Choice Selectors](https://mlsanity.com/paper/2309.03882/) (2023)
- [LaMP: When Large Language Models Meet Personalization](https://mlsanity.com/paper/2304.11406/) (2023)
- [Iterative Preference Learning from Human Feedback: Bridging Theory and Practice for RLHF under KL-Constraint](https://mlsanity.com/paper/2312.11456/) (2023)
- [Personalized Soups: Personalized Large Language Model Alignment via Post-hoc Parameter Merging](https://mlsanity.com/paper/2310.11564/) (2023)
- [Principled Reinforcement Learning with Human Feedback from Pairwise or K-wise Comparisons](https://mlsanity.com/paper/2301.11270/) (2023)
- [PersonaLLM: Investigating the Ability of Large Language Models to Express Personality Traits](https://mlsanity.com/paper/2305.02547/) (2023)
- [Personalizing Reinforcement Learning from Human Feedback with Variational Preference Learning](https://mlsanity.com/paper/2408.10075/) (2024)
- [The PRISM Alignment Dataset: What Participatory, Representative and Individualised Human Feedback Reveals About the Subjective and Multicultural Alignment of Large Language Models](https://mlsanity.com/paper/2404.16019/) (2024)
- [Personalized Language Modeling from Personalized Human Feedback](https://mlsanity.com/paper/2402.05133/) (2024)
- [Online Clustering of Bandits](https://mlsanity.com/paper/1401.8257/) (2014)
- [Preference-based Online Learning with Dueling Bandits: A Survey](https://mlsanity.com/paper/1807.11398/) (2018)
- [RLHF from Heterogeneous Feedback via Personalization and Preference Aggregation](https://mlsanity.com/paper/2405.00254/) (2024)
- [PERSONA: A Reproducible Testbed for Pluralistic Alignment](https://mlsanity.com/paper/2407.17387/) (2024)
- ["My Answer is C": First-Token Probabilities Do Not Match Text Answers in Instruction-Tuned Language Models](https://mlsanity.com/paper/2402.14499/) (2024)
- [Meta-Thompson Sampling](https://mlsanity.com/paper/2102.06129/) (2021)
- [Active Preference Optimization for Sample Efficient RLHF](https://mlsanity.com/paper/2402.10500/) (2024)
- [PersonalLLM: Tailoring LLMs to Individual Preferences](https://mlsanity.com/paper/2409.20296/) (2024)
- [Generative Students: Using LLM-Simulated Student Profiles to Support Question Item Evaluation](https://mlsanity.com/paper/2405.11591/) (2024)
- [When "A Helpful Assistant" Is Not Really Helpful: Personas in System Prompts Do Not Improve Performances of Large Language Models](https://mlsanity.com/paper/2311.10054/) (2023)
- [Meta-learning with Stochastic Linear Bandits](https://mlsanity.com/paper/2005.08531/) (2020)
- [Latent Bandits Revisited](https://mlsanity.com/paper/2006.08714/) (2020)
- [Bayesian decision-making under misspecified priors with applications to meta-learning](https://mlsanity.com/paper/2107.01509/) (2021)
- [LoRe: Personalizing LLMs via Low-Rank Reward Modeling](https://mlsanity.com/paper/2504.14439/) (2025)

## Abstract (arXiv)

Large language models can answer the same question in different styles, but users' preferred styles are hidden and vary across users and questions. We study inference-time personalization through online selection from a fixed set of prompts, each representing a persona. Users arrive sequentially, each asks a short sequence of questions, and the system displays two responses per question and observes one pairwise choice. We formulate this interaction as a variant of the contextual dueling bandit problem in which the learner starts without preference data, adapts within each session, and exploits information shared across users. The objective is to minimize cumulative strong regret, summing the utility gaps of both displayed responses relative to the user's best response among the available options. We make two contributions. First, we introduce the problem and release a reproducible preference benchmark and dataset built around a tutoring case study, containing educational questions, LLM-generated answers in six teaching styles, and pairwise judgments from LLM-simulated students with hidden preference profiles. We fit Bradley-Terry models to these judgments to obtain reference utilities for regret evaluation. Second, we propose a baseline that learns a mixture prior over users from completed sessions and adapts to each new user within the session. On students and topics never used in development, it removes 44-48% of random pairs' excess regret and achieves over half the performance of a per-user linear oracle fitted in hindsight. We release code, prompts, and the benchmark.