Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
tags:
|
| 4 |
+
- multimodal
|
| 5 |
+
- reasoning
|
| 6 |
+
- sft
|
| 7 |
+
- rl
|
| 8 |
+
- perception
|
| 9 |
+
datasets:
|
| 10 |
+
- multimodal-reasoning-lab/Zebra-CoT
|
| 11 |
+
- ModalityDance/Omni-Bench
|
| 12 |
+
base_model:
|
| 13 |
+
- GAIR/Anole-7b-v0.1
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
# Omni-R1
|
| 17 |
+
|
| 18 |
+
Omni-R1 is trained with multimodal interleaved supervision. It uses PeSFT for stable functional image generation, then PeRPO for RL refinement on unified tasks.
|
| 19 |
+
|
| 20 |
+
<p align="center">
|
| 21 |
+
<a href="https://arxiv.org/abs/2601.09536"><b>Paper</b>👁️</a> ·
|
| 22 |
+
<a href="https://github.com/ModalityDance/Omni-R1"><b>Code</b>🐙</a> ·
|
| 23 |
+
<a href="https://huggingface.co/datasets/ModalityDance/Omni-Bench"><b>Omni-Bench</b>🧪</a>
|
| 24 |
+
</p>
|
| 25 |
+
|
| 26 |
+
## Citation
|
| 27 |
+
```bibtex
|
| 28 |
+
@misc{cheng2026omnir1unifiedgenerativeparadigm,
|
| 29 |
+
title={Omni-R1: Towards the Unified Generative Paradigm for Multimodal Reasoning},
|
| 30 |
+
author={Dongjie Cheng and Yongqi Li and Zhixin Ma and Hongru Cai and Yupeng Hu and Wenjie Wang and Liqiang Nie and Wenjie Li},
|
| 31 |
+
year={2026},
|
| 32 |
+
eprint={2601.09536},
|
| 33 |
+
archivePrefix={arXiv},
|
| 34 |
+
primaryClass={cs.AI},
|
| 35 |
+
url={https://arxiv.org/abs/2601.09536},
|
| 36 |
+
}
|
| 37 |
+
```
|