Add pipeline tag and citation to model card

README.md

@@ -1,14 +1,33 @@
 ---
 license: apache-2.0
+pipeline_tag: video-text-to-text
 ---
 
 # SVI-Bench
 
 **SVI-Bench** is a benchmark for **Strategic Video Intelligence**: evaluating whether multimodal models can move from seeing what happens in complex multi-agent sports video to explaining why it happens, simulating alternatives, and autonomously integrating evidence for higher-level analysis.
 
-## Where to go
+It leverages team sports (basketball, soccer, and hockey) as a dynamic microworld to evaluate four cognitive pillars: Perception, Reasoning, Simulation, and Agency.
 
-- 📚 **Dataset:** [huggingface.co/datasets/MVP-Group/SVI-Bench](https://huggingface.co/datasets/MVP-Group/SVI-Bench)
+## Resources
+
+- 📚 **Dataset:** [MVP-Group/SVI-Bench](https://huggingface.co/datasets/MVP-Group/SVI-Bench)
 - 🌐 **Project page:** [svi-bench.github.io](https://svi-bench.github.io/)
 - 📄 **Paper (arXiv):** [arxiv.org/abs/2605.31529](https://arxiv.org/abs/2605.31529)
 - 💻 **Code:** [github.com/Texaser/SVI-Bench](https://github.com/Texaser/SVI-Bench)
+
+## Citation
+
+If you use SVI-Bench in your research, please cite the following paper:
+
+```bibtex
+@misc{pan2026svibenchdynamicmicroworldstrategic,
+      title={SVI-Bench: A Dynamic Microworld for Strategic Video Intelligence}, 
+      author={Yulu Pan and Han Yi and Seongsu Ha and Md Mohaiminul Islam and Benjamin Zhang and Lorenzo Torresani and Gedas Bertasius},
+      year={2026},
+      eprint={2605.31529},
+      archivePrefix={arXiv},
+      primaryClass={cs.CV},
+      url={https://arxiv.org/abs/2605.31529}, 
+}
+```