Evol-Instruct-Arabic-GPT4

The dataset is created by 1. translating English questions of [Evol-instruct-70k](https://huggingface.co/datasets/WizardLM/WizardLM_evol_instruct_70k) into Arabic using **GPT4**, and 2. requesting **GPT4** to generate responses in Arabic. For more details, please refer to: - **Repository**: - https://github.com/FreedomIntelligence/AceGPT - https://github.com/FreedomIntelligence/LLMZoo - **Paper**: - [AceGPT, Localizing Large Language Models in Arabic](https://arxiv.org/abs/2309.12053) - [Phoenix: Democratizing ChatGPT across Languages](https://arxiv.org/abs/2304.10453) ### BibTeX entry and citation info ```bibtex @article{huang2023acegpt, title={AceGPT, Localizing Large Language Models in Arabic}, author={Huang, Huang and Yu, Fei and Zhu, Jianqing and Sun, Xuening and Cheng, Hao and Song, Dingjie and Chen, Zhihong and Alharthi, Abdulmohsen and An, Bang and Liu, Ziche and others}, journal={arXiv preprint arXiv:2309.12053}, year={2023} } @article{chen2023phoenix, title={Phoenix: Democratizing chatgpt across languages}, author={Chen, Zhihong and Jiang, Feng and Chen, Junying and Wang, Tiannan and Yu, Fei and Chen, Guiming and Zhang, Hongbo and Liang, Juhao and Zhang, Chen and Zhang, Zhiyi and others}, journal={arXiv preprint arXiv:2304.10453}, year={2023} } ```本数据集的构建流程如下： 1. 通过GPT-4（GPT4）将[Evol-instruct-70k](https://huggingface.co/datasets/WizardLM/WizardLM_evol_instruct_70k)中的英文问题翻译为阿拉伯语； 2. 调用GPT-4（GPT4）生成对应的阿拉伯语回答。 如需了解更多细节，请参阅： - **代码仓库**： - https://github.com/FreedomIntelligence/AceGPT - https://github.com/FreedomIntelligence/LLMZoo - **相关论文**： - 《AceGPT：面向阿拉伯语的大语言模型本地化》（https://arxiv.org/abs/2309.12053） - 《Phoenix：实现跨语言ChatGPT民主化》（https://arxiv.org/abs/2304.10453） ### BibTeX引用信息 bibtex @article{huang2023acegpt, title={AceGPT, Localizing Large Language Models in Arabic}, author={Huang, Huang and Yu, Fei and Zhu, Jianqing and Sun, Xuening and Cheng, Hao and Song, Dingjie and Chen, Zhihong and Alharthi, Abdulmohsen and An, Bang and Liu, Ziche and others}, journal={arXiv preprint arXiv:2309.12053}, year={2023} } @article{chen2023phoenix, title={Phoenix: Democratizing chatgpt across languages}, author={Chen, Zhihong and Jiang, Feng and Chen, Junying and Wang, Tiannan and Yu, Fei and Chen, Guiming and Zhang, Hongbo and Liang, Juhao and Zhang, Chen and Zhang, Zhiyi and others}, journal={arXiv preprint arXiv:2304.10453}, year={2023} }

搜集汇总

背景与挑战

背景概述

该数据集通过GPT4将Evol-instruct-70k的英文问题翻译为阿拉伯文，并生成相应的阿拉伯文回复。其开发与AceGPT和Phoenix等阿拉伯语大语言模型研究相关。