README.md

---
title: RISE RL
emoji: 🧠
colorFrom: yellow
colorTo: indigo
sdk: static
pinned: false
license: apache-2.0
short_description: Rubric-Informed Selective Exploration
---

# RISE-RL: Rubric-Informed Selective Exploration for Open-Ended Reinforcement Learning

Official project page for **RISE-RL**.

- **Paper:** https://arxiv.org/abs/2608.09123
- **PDF:** https://arxiv.org/pdf/2608.09123
- **Project Page:** https://huggingface.co/spaces/houkunkun/RISE-RL


## Authors

Jinkun Hou, Zhuo Liu, Huimin Ren, Hongsheng Xin, Pan Zhou, Kun Zhan

## Abstract

RISE-RL is a rubric-informed selective exploration method for open-ended reinforcement learning. It uses repeatedly missed rubric
criteria to elicit privileged trajectories that are difficult to discover through unguided exploration alone.

For the complete abstract and experimental results, see our paper:

https://arxiv.org/abs/2608.09123

## Citation

```bibtex
@article{hou2026riserl,
  title   = {RISE-RL: Rubric-Informed Selective Exploration for Open-Ended Reinforcement Learning},
  author  = {Hou, Jinkun and Liu, Zhuo and Ren, Huimin and Xin, Hongsheng and Zhou, Pan and Zhan, Kun},
  journal = {arXiv preprint arXiv:2608.09123},
  year    = {2026}
}