Create a Data Loader for Lightning Pipeline

Problem Statement:

As part of our PyTorch-based pipeline, we require a data loader that can efficiently load and preprocess our dataset. Currently, we lack a dedicated data loader for this purpose.

Key Features and Requirements:

[ ] Create PyTorch custom dataset class : eeg_dataset.py
[ ] Create Lightning data module for each task : task1_data.py, task2_data.py
[ ] Batch creation and handling
[ ] Windowing
[ ] Documentation and examples on how to use the data loader scripts

Pass condition:

Try to run the baseline in end-to-end fashion on the toy split of the dataset

Additional Context:

This should be implemented as a modular code in a separate file that can be later imported wherever required. Providing information about the data dimensions as output would be great. Understanding whether the data loader has any difference for regression and match-mismatch task.

References:

You can have a look at the implementation by competition organiser in their repository

AlKun25 / WeHear