Custom Data augmentation Function for sequence data

Hi everyone

I have a class to load data from files and normalize them.

I would like to know how to write my own custom augmentation functions in pytorch for RNN (For sequence in general).
The code where a load the dataset is something like this:

train_dataset = PartDataset(root = ‘directory_path’, classification = True)
train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=batch_size, shuffle=True, num_workers=4)