caro-holt/MultiQ
收藏Hugging Face2024-03-02 更新2024-03-04 收录
下载链接：
https://hf-mirror.com/datasets/caro-holt/MultiQ
下载链接
链接失效反馈官方服务：
资源简介：
---
license: cc-by-4.0
language:
- tl
- sm
- mk
- gu
- fi
- mn
- bm
- ta
- ur
- hy
- nl
- tk
- en
- bg
- gd
- pt
- ko
- ga
- eu
- sv
- bs
- co
- fr
- gn
- ro
- it
- dv
- ku
- ak
- eo
- zu
- id
- te
- sl
- lv
- pa
- ru
- si
- ee
- yi
- ny
- az
- sw
- hi
- mt
- sr
- hr
- ka
- ug
- tt
- lg
- kn
- fy
- kk
- ca
- lb
- jv
- et
- la
- tr
- ps
- km
- zh
- uk
- as
- he
- yo
- sq
- da
- gl
- vi
- ay
- is
- ln
- mr
- st
- xh
- cs
- ky
- ml
- ht
- mi
- so
- uz
- el
- ti
- be
- cy
- am
- ig
- or
- fa
- ms
- su
- de
- lo
- ha
- ts
- om
- ar
- my
- es
- qu
- 'no'
- th
- sa
- mg
- pl
- sd
- sk
- bn
- rw
- af
- ne
- lt
- tg
- ja
- sn
- hu
size_categories:
- 10K<n<100K
task_categories:
- question-answering
---
# Dataset Card for MultiQ
This is the dataset corresponding to the paper "Evaluating the Elementary Multilingual Capabilities of Large Language Models with MultiQ".
It is a silver standard benchmark that can be used to evaluate the basic multilingual capabilities of LLMs. It contains 200 open ended questions automatically
translated into 137 typologically diverse languages.
- **Curated by:** Carolin Holtermann, Paul Röttger, Timm Dill, Anne Lauscher
- **Language(s) (NLP):** 137 diverse languages described in detail in our paper
- **License:** CC-BY-4.0 License
### Dataset Sources
- **Repository:** [Github](https://github.com/paul-rottger/multiq)
- **Paper:** TBD
提供机构：
caro-holt
原始信息汇总
数据集概述
基本信息
- 数据集名称： MultiQ
- 相关论文： "Evaluating the Elementary Multilingual Capabilities of Large Language Models with MultiQ"
- 数据集用途： 用于评估大型语言模型的基本多语言能力
数据集内容
- 问题数量： 200个开放式问题
- 语言种类： 137种不同类型的语言
数据集详情
- 语言列表： 137种语言，详细描述见论文
- 许可证： CC-BY-4.0
数据集来源
- 代码仓库： Github
- 相关论文： 待定
