Towards Disentangled Preference Optimization Dynamics: Suppress the Loser, Preserve the Winner
Preference optimization is widely used to align large language models (LLMs) with human preferences. However, many margin-based methods also suppress the chosen response when they try to suppress the rejected one, and there is no general way to prevent this across different objectives. We address this issue with a unified incentive-score decomposition of preference optimization, revealing that different objectives share the same local update directions and differ only in their scalar weights. This decomposition provides a common framework for analyzing objectives that were previously studied in separate settings. Building on this decomposition, by analyzing the dynamics of the chosen/reje...
| Authors | Wei Chen, Yubing Wu, Junmei Yang et al. |
|---|---|
| Primary category | cs.LG |
| Published | 20 April 2026 |
| Last updated at source | 23 July 2026 |
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.