Datasets:
Dataset Viewer
The dataset viewer is not available for this dataset.
Unexpected token '<', "<html>
<h"... is not valid JSON
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
piqa-ka
Georgian translation of the PIQA (Physical Interaction QA) benchmark.
Dataset Summary
| Property | Value |
|---|---|
| Examples | 1,720 |
| Splits | validation |
| Languages | Georgian, English |
| Task | Physical Commonsense Reasoning |
Data Fields
goal: Goal description (English)sol1: Solution 1 (English)sol2: Solution 2 (English)label: Correct solution indexgoal_ka: Goal description (Georgian)sol1_ka: Solution 1 (Georgian)sol2_ka: Solution 2 (Georgian)
Translation Methodology
- Translation model generates initial Georgian translation
- Human contractor verifies and fixes translation if needed
- Human validator confirms translation quality
Usage
from datasets import load_dataset
ds = load_dataset("tbilisi-ai-lab/piqa-ka")
Citation
@misc{tbilisi2025piqaka,
title = {piqa-ka: Georgian PIQA Benchmark},
author = {Tbilisi AI Lab Team},
year = {2025},
publisher = {Hugging Face},
howpublished = {\url{https://huggingface.co/datasets/tbilisi-ai-lab/piqa-ka}}
}
License
Contact
- Organization: Tbilisi AI Lab
- Website: ailab.ge
- Downloads last month
- 28