Can't load reddit_tifu: NonMatchingChecksumError

[bermeitinger-b]

Short description
reddit_tifu cannot be loaded due to mismatching checksum

Environment information

• Operating System: Colab and Linux

• Python version: 3.10, 3.11, 3.12

• tensorflow-datasets/tfds-nightly version: tfds==4.9.6 and tfds-nightly==4.9.7+nightly

• Does the issue still exists with the last tfds-nightly package (pip install --upgrade tfds-nightly) ? yes

Reproduction instructions

import tensorflow_datasets as tfds
ds = tfds.load("reddit_tifu/short", split="train", as_supervised=True)

https://colab.research.google.com/drive/12x9Ch4u-eb5bzYrEW4FM65zbrvRc83Zn?usp=sharing

Link to logs

NonMatchingChecksumError                  Traceback (most recent call last)

[<ipython-input-3-257e87e56f6f>](https://localhost:8080/#) in <cell line: 1>()
----> 1 ds = tfds.load("reddit_tifu/short", split="train", as_supervised=True)

19 frames

[/usr/local/lib/python3.10/dist-packages/tensorflow_datasets/core/download/download_manager.py](https://localhost:8080/#) in _register_or_validate_checksums(self, url, url_info, path)
    519             'https://www.tensorflow.org/datasets/overview#fixing_nonmatchingchecksumerror'
    520         )
--> 521         raise NonMatchingChecksumError(msg)
    522 
    523   def _is_checksum_registered(self, url: str) -> bool:

NonMatchingChecksumError: Artifact https://drive.google.com/uc?export=download&id=1ffWfITKFMJeqjT8loC8aiCLRNJpc_XnF, downloaded to /root/tensorflow_datasets/downloads/reddit_tifu/ucexport_download_id_1ffWfITKFMJeqjT8loC8a_XnFKBkIaeZ0glhNGWombXK7QgN8uq9nDHA-eFtk8ZIIqCA.tmp.aac8a1d89e2b4836802068de76df5ab1/download, has wrong checksum:
* Expected: UrlInfo(size=639.54 MiB, checksum='f175cafe348e0521c2424cd419c934d10c6af613ed8cbe8eaa8cfbaa06377f1a', filename='tifu_all_tokenized_and_filtered.json')
* Got: UrlInfo(size=2.39 KiB, checksum='f9f5d613a6fb71a51c9e1b9622c61b61eea8caf15066bf80007ec8b10b28992e', filename='download')
To debug, see: https://www.tensorflow.org/datasets/overview#fixing_nonmatchingchecksumerror

Expected behavior
The dataset should load.

[fylux]

Seems like this data is stored in Google Drive and Google Drive decided to show a page before the download stating that it cannot check it for viruses. Thus it downloads not the data but the warning page as HTML.

There are 2 potential solutions:

• Data is stored in a place oriented towards public programmatic download like GCS or HuggingFace
• Set this (and other Drive datasets) to manual download, so that the data is not automatically downloaded