How to upload dataset on datasets instead of under a user repo?

Hi, trying to contribute a new dataset/benchmark.

We would like to have it available on datasets directly instead of under a user/organization profile.

For eg

from datasets import load_dataset

dataset = load_dataste("<benchmark>", split="train")


from datasets import load_dataset

dataset = load_dataste("<user_profile>/<benchmark>", split="train")

How can we upload the dataset this way, presumable write access to datasets is restricted?

Hi @skgouda

cc’ing @lhoestq @albertvillanova @polinaeterna @severo on this

hi @skgouda !
It’s not possible anymore to host datasets without a user/organization namespace. We encourage users to put datasets under organization namespaces when it’s possible - for better discoverability.

What dataset do you want to upload? If it’s not associated with any specific organization, you can create an organization named as a benchmark (for example, as it is done here, check datasets section).

1 Like