Link the arXiv paper (2609.35767)

README.md

@@ -26,7 +26,7 @@ configs:
 # UMM-Reflection SFT Data
 
 The reflection-SFT data of
-[UMM-Reflection](https://github.com/waltstephen/UMM-Reflection) (*Learning
+[UMM-Reflection](https://github.com/waltstephen/UMM-Reflection) ([paper](https://arxiv.org/abs/2609.35767), *Learning
 Native Reflection in Unified Models*). It trains
 [UMM-Reflection-BAGEL-SFT](https://huggingface.co/YijiaFan/UMM-Reflection-BAGEL-SFT).
 
@@ -124,3 +124,16 @@ ds = load_dataset("YijiaFan/UMM-Reflection-SFT-Data", "trajectories", split="tra
 row = next(iter(ds))
 print(row["user_prompt"], row["think_list"][0])
 ```
+
+## Citation
+
+```bibtex
+@article{ummreflection2026,
+  title   = {Learning Native Reflection in Unified Models
+             with Interleaved Reinforcement Learning},
+  author  = {Fan, Yijia and Huang, Ziqi and Cai, Zhongang and Li, Yan and
+             Wen, Zimo and Yin, Wanqi and Diao, Haiwen and Liu, Ziwei},
+  journal = {arXiv preprint arXiv:2609.35767},
+  year    = {2026}
+}
+```