CORTEXA
← Browse
arxivcs.LGcs.AI2026-06-26

Normalized Rewards for Preference Optimization

Shawn Im, Federico Danieli, Skyler Seto, Barry-John Theobald, Katherine Metcalf

Direct Alignment Algorithms (DAAs) such as DPO have become a common way to post-train and align LLMs with human preferences. However, DAAs have been observed to over-optimize their implicit reward model and decrease the likelihood of preferred responses. This results in a decrease in the total likelihood assigned to responses seen in the preference dataset, potentially resulting in undesirable behavior. To counteract this undesired side-effect of DAAs, we examine the effect of using objectives that add a regularization term to maintain the total length-normalized probabilities of the chosen and rejected responses. To better understand over-optimization, we investigate how response likelihood changes are distributed over the tokens with and without regularization. We find that a significant portion of the likelihood changes are due to a small set of outlier tokens, which explains how DAAs improve generation quality despite decreasing the likelihoods of chosen responses. We apply the proposed regularization to reference-based (DPO) and reference-free (SimPO) methods and find (1) improved trade-offs between generation quality and general benchmark capability and (2) improvements in reward modeling across datasets. For example, on Llama-3.1-8B-Instruct, we see both a >20% relative increase in AlpacaEval2 scores and >9% relative performance gains on general benchmarks. Additionally, we find that the added regularization term effectively mitigates the amount of displacement within preferred responses overall, and for the outlier tokens specifically, by utilizing low-likelihood tokens.

View free PDFSource page

Related papers

arxivcs.CVcs.AIcs.LG2026-07-04

Self-Improving Diffusion Classifiers with Minority Preference Optimization

Hyunsoo Kim, Jungmyung Wi, Soobin Um, Donghyun Kim, Suhyun Kim

Prior studies have demonstrated that diffusion classifiers achieve robust zero-shot classification performance. However, their effectiveness is strongly tied to the pretraining data distribution: they perform well in majority, high-density regions of the data manifold, but are si…

View free PDFSource page
arxivcs.LGcs.AI2026-07-17

Hard Rules, Soft Preferences: Bridging Reasoning, Learning, and Optimization for Personalized Packing Checklist Generation

Himel Dev, Madhusudan Basak, Tanmoy Sen, Paromita Shome, Bashima Islam

Packing for air travel is recurring and error-prone: the checklist must be personal and context-aware, yet feasible under safety rules, item dependencies, and luggage limits. Existing packing assistants are template-driven and generic, or recommendation-driven but unconstrained,…

View free PDFSource page