Este repositório fornece os códigos e conjuntos de dados dos dois artigos a seguir:
Nosso conjunto de dados SQuARe pode ser encontrado em data/SQuARe/
. Consulte o artigo SQuARe para obter detalhes do conjunto de dados.
Também liberamos o conjunto de dados com as anotações brutas em data/SQuARe/with_raw_annotations
. Como as perguntas e respostas em nosso conjunto de dados são inerentemente subjetivas, acreditamos que as anotações brutas ajudariam a pesquisar ainda mais a discordância entre os anotadores.
Observação : embora tenhamos incluído nosso conjunto de dados traduzido para o inglês, são necessários cuidados ao usá-lo diretamente, uma vez que os tópicos delicados que usamos refletem as idiossincrasias da sociedade coreana. Recomendamos que os pesquisadores construam seu próprio conjunto de dados.
O pipeline para geração do conjunto de dados pode ser encontrado em pipeline/square
.
Nosso conjunto de dados KoSBi pode ser encontrado em data/KosBi/
. Consulte o artigo KoSBi para obter detalhes do conjunto de dados.
Atualização : coletamos mais dados executando uma iteração adicional. Você pode encontrá-los nos arquivos chamados data/KoSBi/kosbi_v2_{train,valid,test}.json
, que incluem os conjuntos de dados KoSBi originais. O número total de pares ( contexto , sentença ) aumentou para quase 68 mil, com 34,2 mil sentenças seguras e 33,8 mil sentenças inseguras.
Semelhante ao SQuARe , o pipeline para geração do conjunto de dados pode ser encontrado em pipeline/kosbi
.
Korean-Safety-Benchmarks
MIT License
Copyright 2023-present NAVER Cloud Corp.
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
@inproceedings{lee2023square,
title={SQuARe: A Large-Scale Dataset of Sensitive Questions and Acceptable Responses Created Through Human-Machine Collaboration},
author={Hwaran Lee and Seokhee Hong and Joonsuk Park and Takyoung Kim and Meeyoung Cha and Yejin Choi and Byoung Pil Kim and Gunhee Kim and Eun-Ju Lee and Yong Lim and Alice Oh and Sangchul Park and Jung-Woo Ha},
booktitle={Proceedings of the 61th Annual Meeting of the Association for Computational Linguistics},
year={2023}
}
@inproceedings{lee2023kosbi,
title={KoSBi: A Dataset for Mitigating Social Bias Risks Towards Safer Large Language Model Application},
author={Hwaran Lee and Seokhee Hong and Joonsuk Park and Takyoung Kim and Gunhee Kim and Jung-Woo Ha},
booktitle={Proceedings of the 61th Annual Meeting of the Association for Computational Linguistics: Industry Track},
year={2023}
}
Se você tiver alguma dúvida sobre nosso conjunto de dados ou códigos, sinta-se à vontade para nos perguntar: Seokhee Hong ([email protected]) ou Hwaran Lee ([email protected])