如何在 TensorFlow 数据集中使用 Mnist 数据集？

Question

Hey I want to use data set MNist from TensorFlow-datasets. I used the following code:

import tensorflow as tf
import tensorflow_datasets as tfds
mnist_dataset, mnist_info = tfds.load(name='mnist', with_info=True, as_supervised=True)

I got error:

ImportError: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html

Answer 1

I had the same issue. For some reason TensorFlow Datasets doesn't install the ipywidgets dependency. I tried installing ipywidgets myself and in fact it broke the whole of JupyterLab. For me it that was a problem with the current version of ipywidgets ( 7.5.1 ). I tried using the previous version,

conda install ipywidgets==7.4.2

and now TFDS can download properly.

(Poor code from TensorFlow, by the way. This ipywidgets business is just there for making a pretty loading bar if you're using TFDS in Jupyter; it shouldn't cause everything to fall over when there's a problem. Even after successfully installing the older version I couldn't make the widget rendering work in Jupyter, it prints a bunch of HTML stuff instead, but I don't care, at least it doesn't crash.)