- The URL for the validation set was previously pointed toward to the training set. Fixed to correct URL. (563edc818b5f855ea451913ac4a2b92cc0162bb4)
- Update checksums and sizes for the new files (d22e4f7c16dad1ab83621aab87ffe7f81e5c2023)
- Bump versions (85f680a6c074232322808a2b71ddca61fd98bf34)
- Update README.md according to validation split (706b69e1efe01f8bf55feb8e2b6bf531f7ed99ae)
Co-authored-by: Or Sharir <orsharir@users.noreply.huggingface.co>
Convert dataset sizes from base 2 to base 10 in the dataset card (#2)
* language(s) and license(s) for consistency
* `licenses:` => `license:`
* `languages:` => `language:`
* backward-compat
* make style
* set default values for deprecated languages and licenses
Co-authored-by: Quentin Lhoest <lhoest.q@gmail.com>
Commit from https://github.com/huggingface/datasets/commit/a5192964dc4b76ee5c03593c11ee56f29bbd688d
Remove a copy-paste sentence in dataset cards (#4281)