JudgeLRM: Large Reasoning Models as a Judge

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Nuo, Hu, Zhiyuan, Zou, Qingyun, Wu, Jiaying, Wang, Qian, Hooi, Bryan, He, Bingsheng
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914131749109760
author Chen, Nuo
Hu, Zhiyuan
Zou, Qingyun
Wu, Jiaying
Wang, Qian
Hooi, Bryan
He, Bingsheng
author_facet Chen, Nuo
Hu, Zhiyuan
Zou, Qingyun
Wu, Jiaying
Wang, Qian
Hooi, Bryan
He, Bingsheng
contents Large Language Models (LLMs) are increasingly adopted as evaluators, offering a scalable alternative to human annotation. However, existing supervised fine-tuning (SFT) approaches often fall short in domains that demand complex reasoning. Judgment is inherently reasoning-intensive: beyond surface-level scoring, it requires verifying evidence, identifying errors, and justifying decisions. Through the analysis of evaluation tasks, we find a negative correlation between SFT performance gains and the proportion of reasoning-demanding samples, revealing the limits of SFT in such scenarios. To address this, we introduce JudgeLRM, a family of judgment-oriented LLMs, trained using reinforcement learning (RL) with judge-wise, outcome-driven rewards to activate reasoning capabilities. JudgeLRM consistently outperform SFT-tuned baselines in the same size, as well as other RL and SFT variants, and even surpass state-of-the-art reasoning models: notably, JudgeLRM-3B/4B exceeds GPT-4, while JudgeLRM-7B/8B/14B outperforms DeepSeek-R1 by over 2% in F1 score, with particularly strong gains on reasoning-heavy tasks. Our findings underscore the value of RL in unlocking reasoning-aligned LLM judges.
format Preprint
id arxiv_https___arxiv_org_abs_2504_00050
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle JudgeLRM: Large Reasoning Models as a Judge
Chen, Nuo
Hu, Zhiyuan
Zou, Qingyun
Wu, Jiaying
Wang, Qian
Hooi, Bryan
He, Bingsheng
Computation and Language
Artificial Intelligence
Large Language Models (LLMs) are increasingly adopted as evaluators, offering a scalable alternative to human annotation. However, existing supervised fine-tuning (SFT) approaches often fall short in domains that demand complex reasoning. Judgment is inherently reasoning-intensive: beyond surface-level scoring, it requires verifying evidence, identifying errors, and justifying decisions. Through the analysis of evaluation tasks, we find a negative correlation between SFT performance gains and the proportion of reasoning-demanding samples, revealing the limits of SFT in such scenarios. To address this, we introduce JudgeLRM, a family of judgment-oriented LLMs, trained using reinforcement learning (RL) with judge-wise, outcome-driven rewards to activate reasoning capabilities. JudgeLRM consistently outperform SFT-tuned baselines in the same size, as well as other RL and SFT variants, and even surpass state-of-the-art reasoning models: notably, JudgeLRM-3B/4B exceeds GPT-4, while JudgeLRM-7B/8B/14B outperforms DeepSeek-R1 by over 2% in F1 score, with particularly strong gains on reasoning-heavy tasks. Our findings underscore the value of RL in unlocking reasoning-aligned LLM judges.
title JudgeLRM: Large Reasoning Models as a Judge
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2504.00050