Fix: Not Implemented Error: Loading a Dataset Cached in a LocalFileSystem is Not Supported

When loading a local dataset, I encountered the error:
Not Implemented Error: Loading a dataset cached in a LocalFileSystem is not supported

If you check the top of the error message, you’ll find the root cause: the lookup path points to a .cache/ directory (e.g., xxxxxxx/.cache/). This error persists regardless of whether you use an absolute or relative path.

I found others who encountered the same issue, but they attributed it to an outdated datasets library. However, updating the library didn’t resolve the problem for me.

Based on the mention of “cache”, I made an educated guess that the error was caused by repeated data extraction—Jupyter had cached the dataset. The solution turned out to be surprisingly simple: if you run ls -a in the directory of your current notebook, you’ll find a hidden file ._train.json and a hidden directory .ipynb_checkpoints:

$ ls -a
.  ..  .ipynb_checkpoints  ._train.json  train.json  Untitled.ipynb

I wasn’t able to reproduce the issue to confirm which one was the exact cause, so I deleted both:

rm -rf .ipynb_checkpoints ._train.json

After deletion, loading the dataset worked normally:

I hope these will help someone in need~