TFDS cannot load EMNIST dataset

TensorFlow General Discussion
,
1

Hi,

I have the tensorflow datasets version 4.9.4 installed. I can load the mnist dataset using:

import tensorflow_dataset as tfds

tfds.load("mnist", split=["train"])

But when I try to load the emnist dataset, I got the following error:

NonMatchingChecksumError: Artifact https://www.itl.nist.gov/iaui/vip/cs_links/EMNIST/gzip.zip, downloaded to /root/tensorflow_datasets/downloads/itl.nist.gov_iaui_vip_cs_links_EMNIST_gzipi4VnNviDSrfd9Zju6qv40flc3wr22t8ldulNStS6tmk.zip.tmp.7e59695a085f41a38e3cba964f12d501/itl, has wrong checksum:
* Expected: UrlInfo(size=535.73 MiB, checksum='fb9bb67e33772a9cc0b895e4ecf36d2cf35be8b709693c3564cea2a019fcda8e', filename='gzip.zip')
* Got: UrlInfo(size=110.12 KiB, checksum='b3312c2d21ce4c4ea6e3d632bf885ddca567e35227dc56a6af81ae2620d9885b', filename='itl')
To debug, see: https://www.tensorflow.org/datasets/overview#fixing_nonmatchingchecksumerror

I followed the link, https://www.itl.nist.gov/iaui/vip/cs_links/EMNIST/gzip.zip, and tried to download the file directly from my browser. However, I was redirected to NIST’s homepage.

I think that link is outdated.

3

Hi @davidshen, Thanks for reporting this bug while loading the emnist dataset we have observed the same error. we will work on it and let you know once it is resolved. Thank You.

4

I also created this bug on github canot load EMNIST dataset · Issue #5356 · tensorflow/datasets · GitHub.