huggingface / huggingface/datasets
support squad structure datasets using a YAML parameter
- Dominant language
- Python
- Stars
- 22k
- Forks
- 3.4k
- Avg merge
- 5d 7h
- Merged PRs (30d)
- 17
Description
### Feature request
Since the squad structure is widely used, I think it could be beneficial to support it using a YAML parameter.
could you implement automatic data loading of squad-like data using squad JSON format, to read it from JSON files and view it in the correct squad structure.
The dataset structure should be like this:
https://huggingface.co/datasets/squad
Columns:id,title,context,question,answers
### Motivation
Dataset repo requires arbitrary Python code execution
### Your contribution
The dataset structure should be like this:
https://huggingface.co/datasets/squad
Columns:id,title,context,question,answers
train and dev sets in squad structure JSON files
Contributor guide
Assessment
This issue has not been assessed yet.