Release dataset on HF (with gated access)

[NielsRogge]

Hi @yixchen,

Niels here from the open-source team at Hugging Face. I discovered your work through ECCV, it was featured on daily papers here: https://huggingface.co/papers/2401.09340 (feel free to claim it with your HF account). I work together with AK on improving the visibility of researchers' work on the hub.

It'd be great to make the dataset available on the 🤗 hub, we can add tags so that people find them when filtering https://huggingface.co/datasets, so that people can do:

from datasets import load_dataset

dataset = load_dataset("your-hf-username-or-organization/sceneverse")

See here for a guide: https://huggingface.co/docs/datasets/loading.

There's then also the dataset viewer which allows people to see the first few rows in the browser: https://huggingface.co/docs/hub/en/datasets-viewer.

This would make the dataset easier accessible, and also discoverable. We can then also link the dataset to the paper page.

Gating mechanism

We support gating (similar to how this works for llama/mistral models), so that you can manually review who can access the dataset: https://huggingface.co/docs/hub/en/datasets-gated. See FineVideo (a video dataset created by HF) as an example: https://huggingface.co/datasets/HuggingFaceFV/finevideo.

Let us know if you need any help.

Cheers,

Niels ML Engineer @ HF 🤗

[Buzz-Beater]

Sure! Thanks for reaching out, we will look into it :-)

[Buzz-Beater]

Hi, just curious, if we want to still hold all the data access license on HF Datasets (which is now provided in the google form), how should we properly do it when uploading to HF?

[NielsRogge]

Hi, this is explained here: https://huggingface.co/docs/hub/en/datasets-gated.

Usually people start by uploading the data as a private dataset repository, then make it public when gated access is enabled