Torchtext Iterator omit the last batch

nlp
1

It seems like Torchtext Iterator doesn’t omit the last batch that usually doesn’t conforms to the batch_size parameter.

If there is a way to omit the last batch from the Iterator?

2

Are you feeding this Iterator to a DataLoader?
If so, you could set drop_last=True.
I’m not that deeply familiar with Torchtext so let me know, if that would work for you.

1 Like