Doubly Robust Alignment for Large Language Models

arXiv:2506.01183v3 Announce Type: replace-cross Abstract: This paper studies reinforcement learning from human feedback (RLHF) for aligning large language models with human preferences. While RLHF has demonstrated promising results, many algorithms are highly sensitive to misspecifications in the underlying preference model (e.g., the Bradley-Terry model), the reference policy, or the reward function, resulting in undesirable fine-tuning. To address model misspecification, we propose a doubly robust preference optimization algorithm that remains consistent when either the preference model or the reference policy is correctly specified (without requiring both). Our proposal demonstrates superior and more robust performance than state-of-the-art algorithms, both in theory and in practice. The code is available at https://github.com/DRPO4LLM/DRPO4LLM

In-Context Reward Adaptation for Robust Preference Modeling
Distributionally Robust Reinforcement Learning with Human Feedback
Difficulty-Based Preference Data Selection by DPO Implicit Reward Gap
f-Divergence Regularized RLHF: Two Tales of Sampling and Unified Analyses
Factored Causal Representation Learning for Robust Reward Modeling in RLHF

Source: arXiv cs.AI | 2026-10-28

Related