Abstract:Direct Preference Optimization (DPO) and related methods align large language models from pairwise preferences by regularizing updates against a fixed reference policy. As the policy drifts, a static reference, however, can become increasingly miscalibrated, leading to distributional mismatch and amplifying spurious preference signals under noisy supervision. Conversely, reference-free variants avoid mismatch but often suffer from unconstrained reward drift. We propose Geometric Anchor Preference Optimization (GAPO), which replaces the fixed reference with a dynamic, geometry-aware anchor: an adversarial local perturbation of the current policy within a small radius that serves as a pessimistic baseline. This anchor enables an adaptive reweighting mechanism, modulating the importance of each preference pair based on its local sensitivity. We further introduce the Anchor Gap, the reward discrepancy between the policy and its anchor, and show under smoothness conditions that it approximates worst-case local margin degradation. Optimizing a logistic objective weighted by this gap downweights geometrically brittle instances while emphasizing robust preference signals. Across diverse noise settings, GAPO consistently improves robustness while matching or improving performance on standard LLM alignment and reasoning benchmarks.
| Comments: | Under Review |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2602.04909 [cs.LG] |
| (or arXiv:2602.04909v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2602.04909 arXiv-issued DOI via DataCite |
From: Youngjae Cho [view email]
[v1]
Wed, 4 Feb 2026 00:40:21 UTC (315 KB)
[v2]
Tue, 10 Feb 2026 01:32:46 UTC (312 KB)
[v3]
Fri, 15 May 2026 07:22:08 UTC (264 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。