Datasets:

amanrangapur

/

Fin-Fact

like 6

Hi! You can just delete the script (Fin-Fact.py) and configure from the README header (see https://huggingface.co/docs/hub/datasets-manual-configuration) to point to the data file. Something like:

configs:
- config_name: default
  data_files:
  - split: train
    path: "finfact.json"

Hello, thanks for taking your time to help me resolve this issue. I think this works fine with .csv files. But for json, I am not sure why this is throwing this error.
Error code: FeaturesError
Exception: ValueError
Message: Not able to read records in the JSON file at hf://datasets/amanrangapur/Fin-Fact@bd8018121e6fcc8cce22991ca0e68a7c4cd4cfae/finfact.json.

Not sure. cc @albertvillanova @lhoestq

@albertvillanova @lhoestq Can you please help me out with dataset card loading issue?
Error: Error code: FeaturesError
Exception: ValueError
Message: Not able to read records in the JSON file at hf://datasets/amanrangapur/Fin-Fact@bd8018121e6fcc8cce22991ca0e68a7c4cd4cfae/finfact.json.

Hi ! The Dataset Viewer expects all the columns to have a fixed type.

However the error says cannot mix list and non-list, non-null values: so it looks like at least one field in your JSON file has a mixe type (sometimes it's an object and sometimes it's a list).
Indeed the third sample has 'hrefs': 'http://www.tampabay.com/opinion/columns/crisafulli-why-the-florida-house-opposes-medicaid-expansion/2227220' while in other samples hrefs is a list of strings.

To fix the Viewer you can fix your JSON file to always have lists for hrefs