Update model card with paper, code, and metadata

README.md

@@ -1,15 +1,48 @@
 ---
-license: bsd-2-clause
+base_model:
+- zai-org/GLM-Image
 datasets:
 - HuggingFaceFW/finetranslations
 - sojuL/RubricHub_v1
 language:
 - en
 - id
+license: apache-2.0
 metrics:
 - accuracy
-base_model:
-- zai-org/GLM-Image
 tags:
 - art
----
+- rubric
+- reinforcement-learning
+pipeline_tag: text-generation
+---
+
+# RubricHub
+
+This repository contains the model associated with the paper [RubricHub: A Comprehensive and Highly Discriminative Rubric Dataset via Automated Coarse-to-Fine Generation](https://huggingface.co/papers/2601.08430).
+
+## Introduction
+RubricHub introduces a large-scale (~110k) and multi-domain rubric dataset designed to enhance Reinforcement Learning with Verifiable Rewards (RLVR) for open-ended generation. Since open-ended generation often lacks ground truth, RubricHub provides a structured proxy for verification using an automated **Coarse-to-Fine Rubric Generation** framework.
+
+The model in this repository is part of a two-stage post-training pipeline:
+1. **RuFT (Rubric-based Rejection Sampling Fine-Tuning)**: Using rubric scores as filters.
+2. **RuRL (Rubric-based Reinforcement Learning)**: Using rubric scores as dense rewards.
+
+The post-trained Qwen3-14B model using this framework achieves state-of-the-art results on HealthBench, surpassing proprietary models like GPT-5.
+
+## Resources
+- **Paper:** [arXiv:2601.08430](https://arxiv.org/abs/2601.08430)
+- **Code:** [GitHub - teqkilla/RubricHub](https://github.com/teqkilla/RubricHub)
+- **Dataset:** [RubricHub_v1 on Hugging Face](https://huggingface.co/datasets/sojuL/RubricHub_v1)
+
+## Citation
+If you find RubricHub useful for your research, please cite:
+
+```bibtex
+@article{li2026rubrichub,
+  title={RubricHub: A Comprehensive and Highly Discriminative Rubric Dataset via Automated Coarse-to-Fine Generation},
+  author={Li, Sunzhu and Zhao, Jiale and Wei, Miteto and Ren, Huimin and Zhou, Yang and {Jingwen Yang} and Liu, Shunyu and Zhang, Kaike and Chen, Wei},
+  journal={arXiv preprint arXiv:2601.08430},
+  year={2026}
+}
+```