Update README.md

README.md

@@ -153,11 +153,14 @@ for jsonl_file in test_file.parent.glob("*.jsonl"):
 If you use this dataset, please cite:
 
 ```bibtex
-@misc{vlmgym2024,
-  title={VLM-Gym: A Benchmark for Vision-Language Models in Interactive Environments},
-  author={VLM-Gym Team},
-  year={2024},
-  url={https://huggingface.co/datasets/VisGym/inference-dataset}
+@misc{wang2026visgymdiversecustomizablescalable,
+      title={VisGym: Diverse, Customizable, Scalable Environments for Multimodal Agents}, 
+      author={Zirui Wang and Junyi Zhang and Jiaxin Ge and Long Lian and Letian Fu and Lisa Dunlap and Ken Goldberg and XuDong Wang and Ion Stoica and David M. Chan and Sewon Min and Joseph E. Gonzalez},
+      year={2026},
+      eprint={2601.16973},
+      archivePrefix={arXiv},
+      primaryClass={cs.CV},
+      url={https://arxiv.org/abs/2601.16973}, 
 }
 ```