Release RBTransformer preprocessed datasets on Hugging Face

[NielsRogge]

Hi @nnilayy 🤗

Niels here from the open-source team at Hugging Face. I discovered your work through Hugging Face's daily papers as yours got featured: https://huggingface.co/papers/2511.13954.
The paper page lets people discuss about your paper and lets them find artifacts about it (your models, datasets or demo for instance), you can also claim
the paper as yours which will show up on your public profile at HF, add Github and project page URLs.

It's fantastic to see that you've already made the RBTransformer model checkpoints available in collections on the 🤗 Hub! This significantly improves their discoverability and visibility.

I also noticed that your work introduces 13 preprocessed datasets (e.g., seed_multi_emotion_dataset.pkl) that are generated by your preprocessing scripts. We would be very excited if you would consider making these preprocessed EEG emotion recognition datasets available on the Hub as well.

Hosting these datasets on Hugging Face will further improve their discoverability and visibility. It would also allow people to easily load them using the datasets library:

from datasets import load_dataset

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

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

Besides that, there's the dataset viewer which allows people to quickly explore the first few rows of the data in the browser. We can also help you link both your model collections and these new datasets to your paper page on Hugging Face.

Let me know if you're interested/need any help regarding this!

Cheers,

Niels
ML Engineer @ HF 🤗

[nnilayy]

Hi @NielsRogge,

Thank you so much for reaching out 🤗, and a big thanks to Hugging Face’s Daily Papers for featuring our paper. It truly means a lot to our team, and we appreciate the visibility it has already brought to RBTransformer.

Regarding the preprocessed datasets:
We actually generated 13 benchmark-ready EEG datasets (for example, the seed_multi_emotion_dataset.pkl variants) from SEED, DEAP, and DREAMER using our preprocessing pipeline. However, all three original datasets fall under restrictive EULAs, and prohibit redistribution of the raw or derivative datasets unless explicitly approved by the dataset authors:

• SEED: https://bcmi.sjtu.edu.cn/home/seed/downloads.html
• DEAP (EULA): https://www.eecs.qmul.ac.uk/mmv/datasets/deap/doc/eula.pdf
• DREAMER: https://zenodo.org/records/546113

Because of this, we have been careful about publishing anything derived from these datasets, even if the processed versions differ significantly from the raw data, as it is not fully clear whether our preprocessing qualifies as a "transformative" dataset that can be redistributed under the original EULA.

That said, we would love to make these datasets available on the Hub if it is legally safe to do so. If Hugging Face recommends a standard practice for hosting EULA-bound preprocessed datasets, we are absolutely open to following that.

Happy to collaborate on this, and thank you again for the support!

Kind regards,
Nilay

[NielsRogge]

Thanks for your reply! In that case, I would recommend gated datasets: https://huggingface.co/docs/hub/en/datasets-gated. It allows you to require that anyone who wants to access the dataset needs to fill in a form (created by you) so that you can either manually/automatically approve anyone who fills in the form.