P-Check: Advancing Personalized Reward Model via Learning to Generate Dynamic Checklist

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Seo, Kwangwook, Lee, Dongha
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866910141467590656
author Seo, Kwangwook
Lee, Dongha
author_facet Seo, Kwangwook
Lee, Dongha
contents Recent approaches in personalized reward modeling have primarily focused on leveraging user interaction history to align model judgments with individual preferences. However, existing approaches largely treat user context as a static or implicit conditioning signal, failing to capture the dynamic and multi-faceted nature of human judgment. In this paper, we propose P-Check, a novel personalized reward modeling framework, designed to train a plug-and-play checklist generator that synthesizes dynamic evaluation criteria for guiding the reward prediction. To better align these checklists with personalized nuances, we introduce Preference-Contrastive Criterion Weighting, a training strategy that assigns saliency scores to criteria based on their discriminative power for personalized judgment. We conduct extensive experiments and demonstrate that P-Check not only improves reward accuracy but also enhances downstream personalized generation, and remains robust in OOD scenarios.
format Preprint
id arxiv_https___arxiv_org_abs_2601_02986
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle P-Check: Advancing Personalized Reward Model via Learning to Generate Dynamic Checklist
Seo, Kwangwook
Lee, Dongha
Computation and Language
Recent approaches in personalized reward modeling have primarily focused on leveraging user interaction history to align model judgments with individual preferences. However, existing approaches largely treat user context as a static or implicit conditioning signal, failing to capture the dynamic and multi-faceted nature of human judgment. In this paper, we propose P-Check, a novel personalized reward modeling framework, designed to train a plug-and-play checklist generator that synthesizes dynamic evaluation criteria for guiding the reward prediction. To better align these checklists with personalized nuances, we introduce Preference-Contrastive Criterion Weighting, a training strategy that assigns saliency scores to criteria based on their discriminative power for personalized judgment. We conduct extensive experiments and demonstrate that P-Check not only improves reward accuracy but also enhances downstream personalized generation, and remains robust in OOD scenarios.
title P-Check: Advancing Personalized Reward Model via Learning to Generate Dynamic Checklist
topic Computation and Language
url https://arxiv.org/abs/2601.02986