silk-road/Wizard-LM-Chinese-instruct-evol
收藏资源简介:
Wizard-LM-Chinese数据集是基于MSRA的Wizard-LM数据集,通过翻译指令并调用GPT获得答案的数据集。该数据集包含了许多难度超过Alpaca的指令。在中文问题翻译过程中,可能会有少量指令注入导致翻译失败的情况。中文回答是根据中文问题再进行问询得到的。
The Wizard-LM-Chinese dataset is based on the original Wizard-LM dataset released by Microsoft Research Asia (MSRA). It is constructed by translating the original instructions into Chinese and generating corresponding answers by invoking GPT. This dataset includes numerous instructions that are more challenging than those in the Alpaca dataset. During the translation process of the original instructions into Chinese questions, a small number of cases may occur where instruction injection leads to translation failures. The Chinese answers are obtained by conducting further queries based on the translated Chinese questions.
数据集概述
数据集名称
- Wizard-LM-Chinese
数据集描述
- 基于MSRA的Wizard-LM数据集,通过翻译指令并调用GPT获得答案。
- 包含难度超过Alpaca的指令。
- 存在少量因指令注入导致的中文翻译失败情况。
- 中文回答是根据中文问题再进行问询得到的。
语言
- 中文 (zh)
- 英文 (en)
任务类别
- 文本生成 (text-generation)
- 问答 (question-answering)
数据集大小
- 10K<n<100K
许可
- CC-BY-4.0
相关数据集计划
- [x] WizardLM的中文翻译已发布
- [ ] Coco Caption的中文翻译
- [ ] CoQA的中文翻译
- [ ] CNewSum的Embedding数据
- [ ] 增广的开放QA数据



