This PR improves the model card by:
- Setting the correct `pipeline_tag` to `graph-ml`.
- Linking the model to the research paper [Protein Fold Classification at Scale: Benchmarking and Pretraining](https://arxiv.org/abs/2605.18552).
- Adding a link to the official [TEDBench GitHub repository](https://github.com/BorgwardtLab/TEDBench).
- Maintaining the existing library metadata and usage instructions.
