{"paper":{"title":"P-Check: Advancing Personalized Reward Model via Learning to Generate Dynamic Checklist","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"P-Check trains a checklist generator to produce dynamic criteria that align reward models more closely with individual preferences.","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dongha Lee, Kwangwook Seo","submitted_at":"2026-01-06T12:53:53Z","abstract_excerpt":"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 per"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"P-Check not only improves reward accuracy but also enhances downstream personalized generation, and remains robust in OOD scenarios.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That a learned checklist generator can reliably synthesize dynamic, multi-faceted criteria that capture nuanced personal judgment without introducing artifacts or requiring user-specific data beyond what static baselines already use.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"P-Check advances personalized reward modeling by training a dynamic checklist generator and preference-contrastive weighting to improve reward accuracy, downstream generation, and OOD robustness.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"P-Check trains a checklist generator to produce dynamic criteria that align reward models more closely with individual preferences.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"8bf392728ad17cb37e2e8d35d2905c41f95fd0657f2644385cc74fd86fdf67d1"},"source":{"id":"2601.02986","kind":"arxiv","version":3},"verdict":{"id":"1c8055a6-b917-4fc4-8e6f-1f7d2c38d102","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T17:34:46.181441Z","strongest_claim":"P-Check not only improves reward accuracy but also enhances downstream personalized generation, and remains robust in OOD scenarios.","one_line_summary":"P-Check advances personalized reward modeling by training a dynamic checklist generator and preference-contrastive weighting to improve reward accuracy, downstream generation, and OOD robustness.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That a learned checklist generator can reliably synthesize dynamic, multi-faceted criteria that capture nuanced personal judgment without introducing artifacts or requiring user-specific data beyond what static baselines already use.","pith_extraction_headline":"P-Check trains a checklist generator to produce dynamic criteria that align reward models more closely with individual preferences."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2601.02986/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"bfadb63e5113f8f1aa3bd98227b457c7e82326e48cc315aa2a9f8e5ff2445126"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}