Unable to access Wikitext2 and Wikitext 103 datasets!

nlp
1

Hi facing with the below error we I try to load the wikitext datasets from torchtext

from torchtext.data.functional import to_map_style_dataset
from torchtext import datasets
data = datasets.WikiText103(root='/data',split='train')
data_set = to_map_style_dataset(data)

Facing with the below error on colab

TTPError: 403 Client Error: Forbidden for url: https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-103-v1.zip
This exception is thrown by __iter__ of HTTPReaderIterDataPipe(skip_on_error=False, source_datapipe=OnDiskCacheHolderIterDataPipe, timeout=None)

My torchtext version is 0.17.1

1 Like
2

I face same issue here, anyone can help?

3

@ptrblck Hi can you please help us?

4

That’s not really an PyTorch issue, I think. When you just click the link

https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-103-v1.zip

I don’t know who manages this dataset, but most likely not the PyTorch devs etc.

5

Yes, the link needs to be changed to a working one.

Anyways I downloaded the same dataset from huggingface datasets.