Datasets:

stanfordnlp

/

imdb

like 223

I was having the same issue with this dataset. In the end, it seems to be something with fsspec module. There's a bug on 2023.10.0.

I've solved this by installing the 2023.9.2 and restarting the kernel.

1 - Install the 2023.9.2 version

!pip install fsspec==2023.9.2

2 - Restart your kernel
3 - Download it again

from datasets import load_dataset

imdb_dataset = load_dataset('imdb', download_mode = "force_redownload")

Here is the original solution I've followed: https://stackoverflow.com/questions/77433096/notimplementederror-loading-a-dataset-cached-in-a-localfilesystem-is-not-suppor

I found this solution because I had several random errors while trying different ways to download this dataset. I was trying to set a specific path or ignore the "unsupervised" split validation by the datasets module.

I hope it helps you.

Hi @stefanbschneider ,

You need to update datasets:

pip install -U datasets

Thanks @BrunoBitencourt that solved the issue for me too!
@albertvillanova simply updating or even manually deleting and reinstalling the datasets package did not lead to any success for me.

Hi @saibot01141 ,

If you are running your code in a notebook and you have already imported an old version of datasets:

• You have to update datasets
• And then you have to restart your kernel because otherwise, you will not be using the recently installed version, but the previously imported version

Thanks, updating datasets solved the issue for me! :)