Improve dataset card: Add metadata, links, citation, and license

README.md

@@ -1,7 +1,24 @@
+---
+task_categories:
+- image-text-to-text
+license: cc-by-nc-nd-4.0
+tags:
+- vqa
+- cultural-heritage
+- ancient-greek-pottery
+- multimodal
+---
+
 # VASEVQA Dataset
 
 A Visual Question Answering dataset for ancient vase analysis and description.
 
+VaseVQA is a comprehensive benchmark comprising 31,773 images designed to probe deep understanding of cultural-heritage artifacts, specifically ancient Greek pottery. It aims to equip Multimodal Large Language Models (MLLMs) with robust, expert-level reasoning for tasks like authentication, historical attribution, style classification, and compositional reasoning within this specialized domain.
+
+**Paper:** [VaseVQA: Multimodal Agent and Benchmark for Ancient Greek Pottery](https://huggingface.co/papers/2509.17191)
+**Code:** [https://github.com/AIGeeksGroup/VaseVQA](https://github.com/AIGeeksGroup/VaseVQA)
+**Project Page:** [https://github.com/AIGeeksGroup/VaseVQA](https://github.com/AIGeeksGroup/VaseVQA)
+
 ## Dataset Overview
 
 - **Total Images**: 31,773 vase images
@@ -41,7 +58,8 @@ Each vase is annotated with questions about:
   "conversations": [
     {
       "from": "human",
-      "value": "<image>\nWhat is the fabric of the vase?"
+      "value": "<image>\
+What is the fabric of the vase?"
     },
     {
       "from": "gpt", 
@@ -56,3 +74,19 @@ Each vase is annotated with questions about:
 - `images/`: Directory containing 31,773 vase images
 - `descriptions.csv`: Original metadata source
 - `fix_multi_data.py`: Script for adding image tokens to multi-view data
+
+## Citation
+
+If you use any content of this repo for your work, please cite the following paper:
+```bibtex
+@article{ge2025vasevqa,
+  title={VaseVQA: Multimodal Agent and Benchmark for Ancient Greek Pottery},
+  author={Ge, Jinchao and Cheng, Tengfei and Wu, Biao and Zhang, Zeyu and Huang, Shiya and Bishop, Judith and Shepherd, Gillian and Fang, Meng and Chen, Ling and Zhao, Yang},
+  journal={arXiv preprint arXiv:2509.17191},
+  year={2025}
+}
+```
+
+## License
+
+Our data is licensed under the [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) license. This means it is for Non-Commercial use, and no derivatives or modifications are allowed.