Online LLM Personalization from Pairwise Preferences Across Users

arXiv:2610.17203 · cs.LG, cs.CL, stat.ML · Submitted 2026-10-12 · Read on arXiv

cs.LG, cs.CL, stat.ML

Submitted: 2026-10-12

Updated: 2026-10-12

Comments: 45 pages, 5 figures. Website: https://rogy-labs.github.io/tutorduel/ Code: https://github.com/RoGy-Labs/tutorduel-code Data: https://huggingface.co/datasets/romainpuech/tutorduel

Code: https://github.com/RoGy-Labs/tutorduel-code

Project page: https://rogy-labs.github.io/tutorduel

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Personalization — 1× in this paper · explained in 2 episodes

Sources

• Personalized Soups: Personalized Large Language Model Alignment via Post-hoc Parameter Merging
• Personalized Language Modeling from Personalized Human Feedback
• GPT-4 Technical Report
• RLHF from Heterogeneous Feedback via Personalization and Preference Aggregation
• LLM Agents Grounded in Self-Reports Enable General-Purpose Simulation of Individuals

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks