Paper
Rationale Matters: Learning Transferable Rubrics via Proxy-Guided Critique for VLMReward Models
Authors
Weijie Qiu, Dai Guan, Junxin Wang, Zhihang Li, Yongbo Gai, Mengyu Zhou, Erchao Zhao, Xiaoxi Jiang, Guanjun Jiang
Abstract
Generative reward models (GRMs) for vision-language models (VLMs) often evaluate outputs via a three-stage pipeline: rubric generation, criterion-based scoring, and a final verdict. However, the intermediate rubric is rarely optimized directly. Prior work typically either treats rubrics as incidental or relies on expensive LLM-as-judge checks that provide no differentiable signal and limited training-time guidance. We propose Proxy-GRM, which introduces proxy-guided rubric verification into Reinforcement Learning (RL) to explicitly enhance rubric quality. Concretely, we train lightweight proxy agents (Proxy-SFT and Proxy-RL) that take a candidate rubric together with the original query and preference pair, and then predict the preference ordering using only the rubric as evidence. The proxy's prediction accuracy serves as a rubric-quality reward, incentivizing the model to produce rubrics that are internally consistent and transferable. With ~50k data samples, Proxy-GRM reaches state-of-the-art results on the VL-Reward Bench, Multimodal Reward Bench, and MM-RLHF-Reward Bench, outperforming the methods trained on four times the data. Ablations show Proxy-SFT is a stronger verifier than Proxy-RL, and implicit reward aggregation performs best. Crucially, the learned rubrics transfer to unseen evaluators, improving reward accuracy at test time without additional training. Our code is available at https://github.com/Qwen-Applications/Proxy-GRM.
Metadata
Related papers
Vibe Coding XR: Accelerating AI + XR Prototyping with XR Blocks and Gemini
Ruofei Du, Benjamin Hersh, David Li, Nels Numan, Xun Qian, Yanhe Chen, Zhongy... • 2026-03-25
Comparing Developer and LLM Biases in Code Evaluation
Aditya Mittal, Ryan Shar, Zichu Wu, Shyam Agarwal, Tongshuang Wu, Chris Donah... • 2026-03-25
The Stochastic Gap: A Markovian Framework for Pre-Deployment Reliability and Oversight-Cost Auditing in Agentic Artificial Intelligence
Biplab Pal, Santanu Bhattacharya • 2026-03-25
Retrieval Improvements Do Not Guarantee Better Answers: A Study of RAG for AI Policy QA
Saahil Mathur, Ryan David Rittner, Vedant Ajit Thakur, Daniel Stuart Schiff, ... • 2026-03-25
MARCH: Multi-Agent Reinforced Self-Check for LLM Hallucination
Zhuo Li, Yupeng Zhang, Pengyu Cheng, Jiajun Song, Mengyu Zhou, Hao Li, Shujie... • 2026-03-25
Raw Data (Debug)
{
"raw_xml": "<entry>\n <id>http://arxiv.org/abs/2603.16600v1</id>\n <title>Rationale Matters: Learning Transferable Rubrics via Proxy-Guided Critique for VLMReward Models</title>\n <updated>2026-03-17T14:45:49Z</updated>\n <link href='https://arxiv.org/abs/2603.16600v1' rel='alternate' type='text/html'/>\n <link href='https://arxiv.org/pdf/2603.16600v1' rel='related' title='pdf' type='application/pdf'/>\n <summary>Generative reward models (GRMs) for vision-language models (VLMs) often evaluate outputs via a three-stage pipeline: rubric generation, criterion-based scoring, and a final verdict. However, the intermediate rubric is rarely optimized directly. Prior work typically either treats rubrics as incidental or relies on expensive LLM-as-judge checks that provide no differentiable signal and limited training-time guidance. We propose Proxy-GRM, which introduces proxy-guided rubric verification into Reinforcement Learning (RL) to explicitly enhance rubric quality. Concretely, we train lightweight proxy agents (Proxy-SFT and Proxy-RL) that take a candidate rubric together with the original query and preference pair, and then predict the preference ordering using only the rubric as evidence. The proxy's prediction accuracy serves as a rubric-quality reward, incentivizing the model to produce rubrics that are internally consistent and transferable. With ~50k data samples, Proxy-GRM reaches state-of-the-art results on the VL-Reward Bench, Multimodal Reward Bench, and MM-RLHF-Reward Bench, outperforming the methods trained on four times the data. Ablations show Proxy-SFT is a stronger verifier than Proxy-RL, and implicit reward aggregation performs best. Crucially, the learned rubrics transfer to unseen evaluators, improving reward accuracy at test time without additional training. Our code is available at https://github.com/Qwen-Applications/Proxy-GRM.</summary>\n <category scheme='http://arxiv.org/schemas/atom' term='cs.CV'/>\n <published>2026-03-17T14:45:49Z</published>\n <arxiv:comment>25 pages, 10 figures,</arxiv:comment>\n <arxiv:primary_category term='cs.CV'/>\n <author>\n <name>Weijie Qiu</name>\n </author>\n <author>\n <name>Dai Guan</name>\n </author>\n <author>\n <name>Junxin Wang</name>\n </author>\n <author>\n <name>Zhihang Li</name>\n </author>\n <author>\n <name>Yongbo Gai</name>\n </author>\n <author>\n <name>Mengyu Zhou</name>\n </author>\n <author>\n <name>Erchao Zhao</name>\n </author>\n <author>\n <name>Xiaoxi Jiang</name>\n </author>\n <author>\n <name>Guanjun Jiang</name>\n </author>\n </entry>"
}