Hugging Face
Models
Datasets
Spaces
Posts
Docs
Enterprise
Pricing
Log In
Sign Up
Datasets:
CohereForAI
/
aya_collection_language_split
like
84
Follow
Cohere For AI
644
Modalities:
Tabular
Text
Formats:
parquet
Languages:
Achinese
Afrikaans
Amharic
+ 112
Size:
100M - 1B
ArXiv:
arxiv:
2402.06619
Libraries:
Datasets
Dask
Croissant
+ 1
License:
apache-2.0
Dataset card
Viewer
Files
Files and versions
Community
6
8fabf5a
aya_collection_language_split
/
traditional_chinese
3 contributors
History:
3 commits
shivi
uploading traditional_chinese test split
9618c45
verified
9 months ago
test-00000-of-00001.parquet
53 MB
LFS
uploading traditional_chinese test split
9 months ago
train-00000-of-00001.parquet
548 MB
LFS
uploading traditional_chinese train split
9 months ago
validation-00000-of-00001.parquet
46 MB
LFS
uploading traditional_chinese validation split
9 months ago