Newest papers

CheckDistill: Adaptive Checklist Distillation for Efficient Distributional Reward Modeling in Image Editing

Yaqi Cai, Zuan Gao, Mengting Chen, Wei Li, Bo Rui Zheng · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26740 · cited by 0 works (OpenAlex) · open-access PDF

Korean

지침을 지시한 이미지 편집을 위한 체크 벤 벤 확률 학습은 신뢰할 수 있고 정확하고 효율적인 피드백을 제공하는 보상 모델을 필요로합니다. 직접적인 스케일러 예측은 다차원 편집 선호도를 압축하는 반면, 생성 평가자는 명시적 인 추론을 통해 자신의 판단을 노출 할 수 있습니다. 그러나 광범위한 점수 기준은 신뢰할 수있는 인스턴스에 의존하는 편집 및 보존 제약에 대한 모든 조건을 지정하지 않으며 긴 추론 경로를 생성하는 것은 온라인 최적화에 비용이 많이 드는 것입니다. 우리는 CheckDistill을 도입합니다. 지침에 적응하는 체크 리스트를 내부화하는 포인트별 보상 모델링 프레임 워크. 우리는 각 편집 조건

Machine translation (Meta M2M100); the English original follows.

Abstract

Reinforcement learning for instruction-guided image editing requires reward models that provide reliable, accurate, and efficient feedback. Direct scalar prediction compresses multidimensional editing preferences, whereas generative evaluators can expose their judgment through explicit reasoning. However, broad scoring criteria do not specify all instance-dependent editing and preservation constraints, and generating long reasoning trajectories is costly for online optimization. We introduce CheckDistill, a pointwise reward modeling framework that internalizes instruction-adaptive checklists. We decompose each editing condition into atomic, independently verifiable requirements and construct four-dimensional scoring labels together with structured verification evidence. After supervised score initialization, a privileged teacher conditions on checklists, item-level judgments, and dimension-level rationales to supervise the student's own scoring trajectories. A reliability-weighted on-policy self-distillation objective adjusts supervision at score tokens according to agreement between the teacher's expected scores and reference labels, while retaining candidate probability weights in a truncated token support. The deployed student directly predicts four score distributions without checklist inputs or explicit reasoning and aggregates their expectations into a scalar reward. CheckDistill achieves aggregate accuracies of 72.30%, 56.67% and 82.90% on MMRB2, MER-Bench and EditReward-Bench, respectively, exceeding the compared open-source baselines on all three metrics.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Verifiable secret sharing, Machine learning, Artificial intelligence, Image editing, Security token

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange