This repository contains the data generated by Karya as part of PARIKSHA.
We conducted two kinds of LLM evaluations: (1) pairwise (2) individual. Data of both the evaluations is present in this repository as two different csv files. For more details please refer to our work.
If you use our data or find our insights in the paper relevant to your work, please consider citing us!
@misc{watts2024parikshalargescaleinvestigationhumanllm,
title={PARIKSHA: A Large-Scale Investigation of Human-LLM Evaluator Agreement on Multilingual and Multi-Cultural Data},
author={Ishaan Watts and Varun Gumma and Aditya Yadavalli and Vivek Seshadri and Manohar Swaminathan and Sunayana Sitaram},
year={2024},
eprint={2406.15053},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2406.15053},
}