Update model card with metadata, paper, project, and code links

This PR enhances the model card by:
- Adding the `pipeline_tag: robotics`, which helps users discover the model for relevant tasks.
- Adding `library_name: lerobot`, as evidenced by the `lerobot-style` policy checkpoints mentioned in the GitHub README and `wandb.project: lerobot` in the training config. This enables the automated usage widget on the Hub.
- Including a link to the paper: [Real-to-Sim Robot Policy Evaluation with Gaussian Splatting Simulation of Soft-Body Interactions](https://huggingface.co/papers/2511.04665).
- Providing links to the project page (https://real2sim-eval.github.io/) and the GitHub repository (https://github.com/kywind/real2sim-eval).
- Adding the paper's abstract and BibTeX citation for comprehensive information.

Please review and merge if these updates align with the model's purpose.

README.md

@@ -1,3 +1,30 @@
----
-license: mit
----
+---
+license: mit
+pipeline_tag: robotics
+library_name: lerobot
+---
+
+# Real-to-Sim Robot Policy Evaluation with Gaussian Splatting Simulation of Soft-Body Interactions
+
+This repository contains the official implementation of the **Real-to-Sim Robot Policy Evaluation with Gaussian Splatting Simulation of Soft-Body Interactions** framework. This work was presented in the paper [Real-to-Sim Robot Policy Evaluation with Gaussian Splatting Simulation of Soft-Body Interactions](https://huggingface.co/papers/2511.04665).
+
+## Abstract
+Robotic manipulation policies are advancing rapidly, but their direct evaluation in the real world remains costly, time-consuming, and difficult to reproduce, particularly for tasks involving deformable objects. Simulation provides a scalable and systematic alternative, yet existing simulators often fail to capture the coupled visual and physical complexity of soft-body interactions. We present a real-to-sim policy evaluation framework that constructs soft-body digital twins from real-world videos and renders robots, objects, and environments with photorealistic fidelity using 3D Gaussian Splatting. We validate our approach on representative deformable manipulation tasks, including plush toy packing, rope routing, and T-block pushing, demonstrating that simulated rollouts correlate strongly with real-world execution performance and reveal key behavioral patterns of learned policies. Our results suggest that combining physics-informed reconstruction with high-quality rendering enables reproducible, scalable, and accurate evaluation of robotic manipulation policies.
+
+## Links
+- **Paper**: https://huggingface.co/papers/2511.04665
+- **Project Page**: https://real2sim-eval.github.io/
+- **Code**: https://github.com/kywind/real2sim-eval
+
+For detailed installation instructions, how to load assets and checkpoints, and how to run policy evaluation, keyboard interactive control, motion replay, and policy training, please refer to the [official GitHub repository](https://github.com/kywind/real2sim-eval).
+
+## Citation
+If you find this repo useful, please consider citing the paper:
+```bibtex
+@article{zhang2025real,
+    title={Real-to-Sim Robot Policy Evaluation with Gaussian Splatting Simulation of Soft-Body Interactions},
+    author={Zhang, Kaifeng and Sha, Shuo and Jiang, Hanxiao and Loper, Matthew and Song, Hyunjong and Cai, Guangyan and Xu, Zhuo and Hu, Xiaochen and Zheng, Changxi and Li, Yunzhu},
+    journal={arXiv preprint arXiv:2511.04665},
+    year={2025}
+}
+```