This PR improves the model card by:
- Correcting the license from Apache 2.0 to MIT, as stated in the project's GitHub repository.
- Adding the `pipeline_tag: video-feature-extraction` to help users discover this model for relevant tasks.
- Adding an explicit link to the GitHub repository in the model card content.
- Including a sample usage for pretraining the model.
Please review and merge if these changes look good.
**Model Checkpoint:** DISCOVR
**Paper:** *Self-Supervised Learning of Echocardiographic Video Representations via Online Cluster Distillation* (NeurIPS 2025)
📄 [https://arxiv.org/pdf/2506.11777](https://arxiv.org/pdf/2506.11777)
This checkpoint corresponds to the **DISCOVR** model trained for **400 epochs** on **adult and pediatric echocardiography datasets**, including **EchoDynamic**, **RVENet**, and **EchoNet-Pediatric LVH**.
DISCOVR learns spatio-temporal cardiac representations through **self-supervised online cluster distillation**, capturing both fine-grained anatomical details and global temporal dynamics.
The model enables robust, label-efficient feature learning across diverse cardiac populations for downstream tasks such as anomaly detection, classification, and segmentation.
