Custom dataset for ECG

class Custom_dataset(Dataset):
    def __init__(self, data_dir, transform=ToTensor()):
      self.all_files = []
      files = glob.glob(data_dir + '/*.csv')
      for f in files:
        temp_df = pd.read_csv(f,sep=" ")
        self.all_files.append(temp_df)

    def __len__(self):
      return len(self.all_files)

    def __getitem__(self,idx):
      all_input_tensors = []
      all_output_tensors = []
      #turn list of dataframes into Tensor
      for files in self.all_files:
        temp_df=pd.read_csv(f,sep=" ")
        input_tensor=df.iloc[:,1]
        input_tensor = torch.tensor(input_tensor)
        all_input_tensors.append(input_tensor)
        #output_tensor=df.iloc[:,2:3]
        output_tensor=df.iloc[:,2:9].values
        output_tensor=torch.tensor(output_tensor)
        all_output_tensors.append(output_tensor)
      X=torch.stack(all_input_tensors)
      y=torch.stack(all_output_tensors)
      Sample = X,y
      return Sample

if __name__ == "__main__":
  data_dir = "/content/CSV_data"
  dataset = Custom_dataset(data_dir)