Improve dataset card: Add task category and sample usage from GitHub README

README.md

@@ -1,12 +1,14 @@
 ---
 license: cdla-permissive-2.0
+pretty_name: SWEC iEEG Dataset
 tags:
 - medical
 - biology
 - ieeg
 - seizure
 - epilepsy
-pretty_name: SWEC iEEG Dataset
+task_categories:
+- time-series-forecasting
 ---
 
 ## SWEC iEEG Dataset
@@ -57,6 +59,41 @@ Every file contains attributes `patient` with the patient ID, `channels` with th
 
 ---
 
+## Sample Usage
+
+This section provides basic usage examples for inference and training with MVPFormer, the model associated with this dataset. For more details, refer to the [MVPFormer GitHub repository](https://github.com/IBM/multi-variate-parallel-transformer).
+
+To prepare the environment for running MVPFormer you need a mixture of pip and compilation from source. 
+
+### Install Python Packages
+
+The `requirements.txt` file is provided in the MVPFormer repository. Simply install all requirements with `pip install -r requirements.txt`.
+
+### DeepSpeed
+
+You have to compile [`DeepSpeed`](https://www.deepspeed.ai/tutorials/advanced-install/) manually to activate some necessary extensions. The procedure can vary based on your software and hardware stack, here we report our reference installation steps.
+
+```bash
+DS_BUILD_FUSED_ADAM=1 DS_BUILD_FUSED_LAMB=1 pip install --no-cache-dir deepspeed --global-option="build_ext" --global-option="-j8"
+```
+
+### Inference with MVPFormer
+
+We use `PyTorch Lightning` to distribute reproducible configuration files for our experiments. The example testing configuration file can be found in the `configs` folder of the MVPFormer repository. You can start testing with:
+```bash
+python main.py test --config configs/mvpformer_classification.yaml --model.init_args.base_model '<base_checkpoint_path>' --model.init_args.head_model '<head_checkpoint_path>' --data.init_args.folder '<dataset_path>' --data.init_args.test_patients ['<dataset_subject>']
+```
+
+### Training MVPFormer
+
+We use `PyTorch Lightning` to distribute reproducible configuration files for our experiments. The example testing configuration file can be found in the `configs` folder of the MVPFormer repository. You can start training with:
+```bash
+python main.py fit --config configs/mvpformer_classification.yaml --model.init_args.base_model '<base_checkpoint_path>' --model.init_args.head_model '<head_checkpoint_path>' --data.init_args.folder '<dataset_path>' --data.init_args.train_patients ['<dataset_subject>']
+```
+The example parameters are equivalent to what we have used to train MVPFormer, except in the hardware setup such as the number of GPUs and the number of CPU workers.
+
+---
+
 ## Dataset curation
 
 ### Preparation