paper-with-me

홈 › Papers

RUBRIC-ARROW: Alternating Pointwise Rubric Reward Modeling for LLM Post-training in Non-verifiable Domains

2026-05-27 · Haoxiang Jiang, Zihan Dong, Tianci Liu, Wanying Wang, Ran Xu, Tony Yu, Linjun Zhang, Haoyu Wang arxiv

Pointwise reward modeling offers critical signals for LLM post-training, yet struggles with absolute scoring in subjective, non-verifiable settings. Rubric-based methods address this by decomposing evaluation into explicit criteria, but existing approaches typically depend on frontier LLMs and suffer from ties caused by hard Boolean aggregation. We present RUBRIC-ARROW, an alternating framework that jointly trains a rubric generator and a rubric-conditioned judge, with its RL stage using only pairwise preference data. Our method couples a probability-based scoring rule that reduces ties with phase-specific preference-based rewards and an alternating GRPO scheme that together train the pointwise evaluator. Extensive experiments show that RUBRIC-ARROW achieves competitive reward-modeling accuracy and yields consistent gains for downstream policy post-training.

📄 PDF Abstract BibTeX arXiv:2605.29156

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Open Rubric System: Scaling Reinforcement Learning with Pairwise Adaptive Rubric

2026-02-15 · Ruipeng Jia, Yunyi Yang, Yuxin Wu, Yongbo Gai 외 arxiv

Scalar reward models compress multi-dimensional human preferences into a single opaque score, creating an information bottleneck that often leads to brittleness and reward hacking in open-ended alignment. We argue that r…

Reinforcement Learning

Generating and Refining Dynamic Evaluation Rubrics for LLM-as-a-Judge

2026-05-28 · Zijie Wang, Eduardo Blanco arxiv

LLM-as-a-Judge is a scalable alternative to human evaluation, yet existing rubric-based methods rely on human-annotated data such as reference answers or expert-crafted rubrics. We propose to automatically generate fine-…

Alternating Reinforcement Learning for Rubric-Based Reward Modeling in Non-Verifiable LLM Post-Training

2026-02-02 · Ran Xu, Tianci Liu, Zihan Dong, Tony Yu 외 arxiv

Standard reward models typically predict scalar scores that fail to capture the multifaceted nature of response quality in non-verifiable domains, such as creative writing or open-ended instruction following. To address …

Reinforcement LearningInstruction Following

Small Language Models as Judges for Rubric-Based Reinforcement Learning

2026-08-30 · Fengyu Xie, Yilun Zhao, Bingsen Chen, Arman Cohan 외 hf

Rubric-based reinforcement learning extends RL beyond tasks with exact answers or rule-based verifiers by scoring responses against instance-specific criteria. However, this makes reward computation expensive: training r…

Reinforcement Learning

Alternating Reinforcement Learning with Contextual Rubric Rewards: Beyond the Scalarization Strategy

2026-03-04 · Guangchen Lan, Lian Xiong, Xin Zhou, Hejie Cui 외 arxiv

Reinforcement Learning with Rubric Rewards (RLRR) is a framework that extends conventional reinforcement learning from human feedback (RLHF) and verifiable rewards (RLVR) by replacing scalar preference signals with struc…

Reinforcement Learning