DataPipe batch vs Dataloader batch

data
1

Hi guys,

I am trying to refactor my code with DataPipes instead of Datasets + Dataloader. One of the problems I am running into is when I tried to handle batching with DataPipes instead of the Dataloader, the results are different:

image

As you can see the example batch is different depending on the method used. Could someone explain how to achieve the same functionality that DataLoader(batch_size = n) but with DataPipes?

Thanks.