Datasets:

Size:
n>1T
ArXiv:
License:

How to download only one language?

#1
by musabg - opened

Hi πŸ‘‹
I hope you're all doing well.

It seems that the current download process for the dataset doesn't allow for the selection of a single language, and as a result, the entire dataset is downloaded. Unfortunately, my computer doesn't have enough space to handle such a large dataset. 😞

Do you have any suggestion?

OSCAR org

Hello @musabg ,

Hope you're doing well as well.

We haven't finished the python script to load the data with the datasets library yet. For the moment, the solution would be to download it manually either with git or with the http client: https://huggingface.co/docs/huggingface_hub/guides/download

You can potentially create a for loop with the hf_hub_download() method over all the files of a single language folder.

OSCAR org

Hello @musabg ,

The script is finally ready and you should be able to download a single language using the language=foo parameter in the load_dataset function.

uj changed discussion status to closed

Sign up or log in to comment