Download_and_extract() file missing, but only for one split

🤗Datasets
1

I’m encountering a strange problem when trying to load my dataset (NSynth.py · jg583/NSynth at main). The download links (The NSynth Dataset) contain tar.gz archives that, when unzipped, each yield a folder containing:

  1. An ‘audio’ folder with .wav files
  2. An ‘examples.json’ file

When I try to load the dataset, the train and validation splits are generated without issue. However, I get a FileNotFoundError on the test split, because the ‘examples.json’ file doesn’t show up in the extracted folder. I’ve manually downloaded from the source, and it still contains the ‘examples.json’ file. Does anyone have any idea what might be causing this error (and how to fix it)? I would really appreciate any help!

PS: This is a secondary concern, but what steps would I need to take to get Dataset Viewer compatibility?

EDIT: The issue seems to be resolved. Still not sure why it was happening.

2

Hi ! To get the Dataset Viewer to work, you need to remove the loading script.

Since your data seems to be in a format supported by the Viewer (Parquet) and the README already contains the info of which column is what type, it should work directly without further modifications