Error loading 'structured-wikipedia' due to schema mismatch
Hello,
I'm encountering an error when trying to load the 'structured-wikipedia' dataset due to a schema mismatch between the dataset and the expected schema.
I've already tried loading the schema and manually adding it in the load_dataset function, but the issue persists. Here's the code I used:
'from datasets import load_dataset_builder, load_dataset, Features, Value, Sequence
ds_builder = load_dataset_builder('wikimedia/structured-wikipedia', '20240916.en')
features = ds_builder.info.features
dataset = load_dataset('wikimedia/structured-wikipedia', '20240916.en', split='train')
'.
Has anyone solved this problem?
Also seeing this issue. it's currently unusable. Does anyone have any suggestions here?
Hi! Thank you for the feedback and apologies for the delay in responding. We're looking into this and will provide an update next week.
Hi,
First of all thank you for your feedback! We really appreciate interacting with the community.
We're sorry that this isn't working with the hugging face dataset loading by default. We did prioritised the attempt to solve this.
Unfortunately, due to our dataset format and a recursive structure with optional fields, we cant , after several attempts over time, the latest this week, to offer you a off the box solution.
The long term goal is to change from ndjson to other format, parquet possibly, that integrates easily out of the box with ducksdb and hugging face schema parsing rules.
If there is something you think we can do in the short term, to help you use this dataset, let us know and we'll be happy to work with you to do so.
Again our apologies for this inconvenience, we value you and what you can either do or attempt with our dataset.
Thank you