Multi30k torchtext datatset splits Error

nlp
1

Actually, I am trying to code a Seq2Seq Machine Translation using Pytorch and TorchText Libraray using Multi30k Datasets.

I was trying to perform train,valid,test=Multi30k.splits() , but got this error -

AttributeError                            Traceback (most recent call last)
<ipython-input-17-cc515cd9fb6e> in <module>()
----> 1 train_data, valid_data, test_data = Multi30k.splits(
      2     exts=(".de", ".en"), fields=(german, english)
      3 )

AttributeError: 'function' object has no attribute 'splits'

Can someone please help me in solving that.
I am also attaching the Google Colab Notebook for further reference.

2

The TorchText module has undergone a major overhaul over the last year, changing how you instantiate things. The suggested way to use Multi30k from the release notes is

from torchtext.datasets import Multi30k
train_data, valid_data, test_data = Multi30k()

More parameters are given in the documentation.

Best regards

Thomas

3

Oh yes, it worked successfully.
Thanks for your help…