docs: link the arXiv paper and switch the citation to the arXiv BibTeX

README.md

@@ -15,6 +15,7 @@ viewer: false
 # SMILE: CLIP embeddings, mined hard-negative pools and predictions
 
 Artifacts for **The Value of Being Wrong: Self-Mined Visual In-Context Learning from Errors** (NeurIPS 2026 Spotlight).
+Paper: [arXiv:2610.16977](https://arxiv.org/abs/2610.16977).
 Code: [github.com/Tree-Shu-Zhao/SMILE](https://github.com/Tree-Shu-Zhao/SMILE).
 
 ## Contents
@@ -66,10 +67,13 @@ bash scripts/vizwiz/evaluate.sh predictions/vizwiz/Qwen_Qwen2.5-VL-7B-Instruct/v
 ## Citation
 
 ```bibtex
-@inproceedings{zhao2026smile,
+@misc{zhao2026smile,
   title={The Value of Being Wrong: Self-Mined Visual In-Context Learning from Errors},
-  author={Zhao, Shu and Yu, Tan},
-  booktitle={Advances in Neural Information Processing Systems},
-  year={2026}
+  author={Shu Zhao and Tan Yu},
+  year={2026},
+  eprint={2610.16977},
+  archivePrefix={arXiv},
+  primaryClass={cs.CV},
+  url={https://arxiv.org/abs/2610.16977},
 }
 ```