Release WebThinker artifacts (datasets, models) on Hugging Face

[NielsRogge]

Hi @dongguanting 🤗

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/2504.21776.
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 great to see that you've already made the WebThinker model checkpoints available on the 🤗 Hub! We would love to help you link these existing models to your paper page on Hugging Face to improve their discoverability and visibility. This would allow users to find your models directly from your paper page.

Additionally, I noticed you've made pre-processed data for benchmarks like GPQA, GAIA, WebWalkerQA, HLE, and Glaive available in your GitHub repository's ./data/ directory. Would you be interested in also hosting these processed datasets on https://huggingface.co/datasets? Hosting them on the Hub would enable:

from datasets import load_dataset

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

It would also provide a dataset viewer for easy exploration in the browser.

Linking models to the paper page

It's great that your models are already on the Hub! We can help you link these existing model repositories to your paper page (read here) so people can easily discover them directly from your paper.

Uploading 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.

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

Cheers,

Niels
ML Engineer @ HF 🤗