pretty_name: StereoSet-UK Unlearning
language:
- uk
license: cc-by-sa-4.0
tags:
- stereoset
- ukrainian
- machine-unlearning
- bias
configs:
- config_name: default
data_files:
- split: train
path: data/train.csv
StereoSet-UK Unlearning
StereoSet-UK Unlearning contains 2,101 Ukrainian full-sentence triplets derived from the intrasentence portion of the StereoSet development set. The Ukrainian sentences were translated with the DeepL API and received technical cleanup. English source text is omitted.
Each triplet assigns the stereotype sentence to forget_uk, the anti-stereotype sentence to
retain_uk, and the unrelated sentence to control_uk. Five items with duplicate translated
candidates were excluded.
The 79 source targets use unique, standardized Ukrainian labels. The labels are metadata values and are independent of grammatical inflections inside the full sentences.
The role assignment is an experimental unlearning view created for this release. StereoSet's original scope is stereotypical-bias evaluation. A separate retain corpus is still needed to preserve general language ability and factual knowledge.
Fields
| Field | Description |
|---|---|
item_id |
Original StereoSet item identifier |
bias_type_uk |
Ukrainian bias-category label |
target_uk |
Canonical Ukrainian label for the bias target |
forget_uk |
Full Ukrainian stereotype sentence |
retain_uk |
Full Ukrainian anti-stereotype sentence |
control_uk |
Full Ukrainian unrelated sentence |
This is a provisional machine-translated research dataset. Human validation is still required.
Source and license
This dataset is derived from StereoSet at source commit
ead7d086a64a192a1eca88e0dd2fd163de375218. The source and this translation use the
Creative Commons Attribution-ShareAlike 4.0 International license.
Citation
@inproceedings{nadeem-etal-2021-stereoset,
title = {{StereoSet}: Measuring stereotypical bias in pretrained language models},
author = {Nadeem, Moin and Bethke, Anna and Reddy, Siva},
booktitle = {Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing},
year = {2021},
pages = {5356--5371},
doi = {10.18653/v1/2021.acl-long.416},
url = {https://aclanthology.org/2021.acl-long.416/}
}