Paper 2601.08430
RubricHub: A Comprehensive and Highly Discriminative Rubric Dataset via Automated Coarse-to-Fine Generation
- Published
- Jan 2026
- Research lab
- Independent
- Citations
- 29
- GitHub
- 66 stars
01 In brief
Summary
The paper introduces RubricHub, a large-scale (~110k) multi-domain rubric dataset, generated via an automated Coarse-to-Fine framework to address limitations in existing rubric-based evaluation for open-ended tasks.
The framework comprises three stages: principle-guided and response-grounded generation, multi-model aggregation, and difficulty evolution, producing fine-grained, highly discriminative criteria.
RubricHub covers five domains: Science, Instruction Following, Writing, Medical, and Chat.
The authors validate its utility through a two-stage post-training pipeline: Rubric-based Rejection Sampling Fine-Tuning (RuFT) and Rubric-based Reinforcement Learning (RuRL).
Post-trained Qwen3-14B achieves state-of-the-art results on HealthBench (69.3), surpassing GPT-5 (67.2).
Ablations confirm each framework component adds value, and positive-only criteria outperform those with negative penalties.
The study also shows that rubric quality improves with grader model scale up to 30B, after which performance saturates.
02 From the paper
Abstract
Reinforcement Learning with Verifiable Rewards (RLVR) has driven substantial progress in reasoning-intensive domains like mathematics. However, optimizing open-ended generation remains challenging due to the lack of ground truth. While rubric-based evaluation offers a structured proxy for verification, existing methods suffer from scalability bottlenecks and coarse criteria, resulting in a supervision ceiling effect. To address this, we propose an automated Coarse-to-Fine Rubric Generation framework. By synergizing principle-guided synthesis, multi-model aggregation, and difficulty evolution, our approach produces comprehensive and highly discriminative criteria capable of capturing the subtle nuances. Based on this framework, we introduce RubricHub, a large-scale ($\sim$110k) and multi-domain dataset. We validate its utility through a two-stage post-training pipeline comprising Rubric-based Rejection Sampling Fine-Tuning (RuFT) and Reinforcement Learning (RuRL). Experimental results demonstrate that RubricHub unlocks significant performance gains: our post-trained Qwen3-14B achieves state-of-the-art (SOTA) results on HealthBench (69.3), surpassing proprietary frontier models such as GPT-5. Our code is available at \href{https://github.com/teqkilla/RubricHub}{ this URL}.