XNLI dataset: NonMatchingChecksumError

[xiey1]

Hi,
I tried to download "xnli" dataset in colab using
xnli = load_dataset(path='xnli')
but got 'NonMatchingChecksumError' error

`NonMatchingChecksumError Traceback (most recent call last)
in ()
----> 1 xnli = load_dataset(path='xnli')

3 frames
/usr/local/lib/python3.6/dist-packages/datasets/utils/info_utils.py in verify_checksums(expected_checksums, recorded_checksums, verification_name)
37 if len(bad_urls) > 0:
38 error_msg = "Checksums didn't match" + for_verification_name + ":\n"
---> 39 raise NonMatchingChecksumError(error_msg + str(bad_urls))
40 logger.info("All the checksums matched successfully" + for_verification_name)
41

NonMatchingChecksumError: Checksums didn't match for dataset source files:
['https://www.nyu.edu/projects/bowman/xnli/XNLI-1.0.zip']`

The same code worked well several days ago in colab but stopped working now. Thanks!

[lhoestq]

Thanks for reporting.
The data file must have been updated by the host.
I'll update the checksum with the new one.

[lhoestq]

Well actually it looks like the link isn't working anymore :(

[lhoestq]

The new link is https://cims.nyu.edu/~sbowman/xnli/XNLI-1.0.zip
I'll update the dataset script

[lhoestq]

I'll do a release in the next few days to make the fix available for everyone.
In the meantime you can load xnli with

xnli = load_dataset('xnli', script_version="master")

This will use the latest version of the xnli script (available on master branch), instead of the old one.

[xiey1]

That's awesome! Thanks a lot!