[2509.23371] Alignment through Meta-Weighted Online Sampling: Bridging the Gap between Data Generation and Preference Optimization

Computer Science > Computation and Language arXiv:2509.23371 (cs) [Submitted on 27 Sep 2025 (v1), last revised 27 Feb 2026 (this version, v2)] Title:Alignment through Meta-Weighted Online Sampling: Bridging the Gap between Data Generation and Preference Optimization Authors:Junming Yang, Ning Xu, Biao Liu, Shiqi Qiao, Xin Geng View a PDF of the paper titled Alignment through Meta-Weighted Online Sampling: Bridging the Gap between Data Generation and Preference Optimization, by Junming Yang and 4 other authors View PDF HTML (experimental) Abstract:Preference optimization is crucial for aligning large language models (LLMs) with human values and intentions. A significant challenge in this process is the distribution mismatch between pre-collected offline preference data and the evolving model policy. Existing methods attempt to reduce this gap using static heuristics or decoupled online sampling strategies, but they often fail to adapt to the model's dynamic learning state. To bridge this gap, we propose Meta-Weighted Adaptive Preference Optimization (MetaAPO), a novel framework that dynamically couples data generation with model training. MetaAPO employs a lightweight meta-learner, as an "alignment gap estimator", to evaluate the potential benefits of on-policy sampling in relation to offline data. This guides targeted online generation and assigns sample-wise meta-weights to the optimization objective, dynamically balancing the quality and distribution of online and offlin...