Hugging Face's logo

Datasets:
uonlp
/
CulturaX

Tasks:
Text Generation
Fill-Mask
Sub-tasks:
language-modeling
masked-language-modeling
Languages:
Afrikaans
Tosk Albanian
Amharic
Multilinguality:
multilingual
Size Categories:
n<1K
1K<n<10K
10K<n<100K
Language Creators:
found
Annotations Creators:
no-annotation
Source Datasets:
original
ArXiv:
Tags:
Dataset card Viewer Files Community
19

mc4 undefined portion included?

#10
by acul3 - opened
Discussion
acul3

thanks for releasing dataset

i have question, did you guys include mc4 undefined portion to culturax ?
i mean reclassify language of undefined mc4 and then filter it

sample:
https://huggingface.co/datasets/allenai/c4/blob/mC4310/multilingual/c4-und.00000-00001-00002-00003-00004-00005-00006-00007.json.gz

i find mc4 undefined language has some good quality

here is sampe 20 language from mc4 undefined classify by fasttext
https://huggingface.co/datasets/acul3/mc4-und-fasttext/tree/main

nguyenhuuthuat09
 UONLP org

Hi @acul3 , thank you for your questions and information.

We didn't include mC4 undefined data in CulturaX yet. Its quality looks good and it is in our list to expand CulturaX for the next versions.

Thank you!

nguyenhuuthuat09 changed discussion status to closed

Sign up or log in to comment