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

Indonesian

Pembelajaran penguatan MMBen untuk pengeditan gambar yang dipandu instruksi membutuhkan model penghargaan yang memberikan umpan balik yang dapat diandalkan, akurat, dan efisien. Prediksi skalar langsung mengompres preferensi pengeditan multi-dimensi, sedangkan pengevaluator generatif dapat mengekspos penilaian mereka melalui penilaian eksplisit. Namun, kriteria penilaian yang luas tidak menentukan semua kondisi pengeditan pada persyaratan pengeditan dan pemeliharaan independen, dan menghasilkan trajektori pertimbangan panjang yang mahal untuk optimalisasi online. Kami memperkenalkan CheckDistill, kerangka pemodelan penghargaan pointwise yang menginternalisasikan daftar penilaian instruksi-adaptif. Kami memecah setiap kondisi pengeditan menjadi kondisi atomik, secara independen memverifikasi persyaratan, dan membangun label penilaian empat dimensi bersama dengan bukti verif

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