Release MTAttack code and artifacts on Hugging Face #1
Description
Hi @mala-lab 🤗
Niels here from the open-source team at Hugging Face. I discovered your work on Arxiv and its corresponding Hugging Face paper page: https://huggingface.co/papers/2511.10098. The paper page lets people discuss your paper and find related artifacts (models, datasets, demos). You can also claim the paper as yours, which will show up on your public profile at HF, and add GitHub and project page URLs.
I saw on your GitHub repository (https://github.com/mala-lab/MTAttack) that "Code is Coming Soon!". That's exciting!
Once your code for the MTAttack framework and any associated artifacts (like example backdoored LVLM checkpoints or newly generated datasets of poisoned samples) are ready for release, would you be interested in hosting them on the 🤗 Hub? This would significantly improve their discoverability and visibility within the AI community. We can add metadata tags to their cards, link them directly to your paper page, and more.
If you're down, here are some guides:
- For models (e.g., backdoored LVLMs): https://huggingface.co/docs/hub/models-uploading. You could leverage the PyTorchModelHubMixin for easy integration. We encourage researchers to push each model checkpoint to a separate model repository for better tracking.
- For datasets (e.g., poisoned samples): https://huggingface.co/docs/datasets/loading. This would allow users to easily load them with
from datasets import load_dataset. - For a demo: You can also build a demo for your framework on Spaces, and we can provide you a ZeroGPU grant, which gives you A100 GPUs for free.
Let me know if you're interested or need any guidance as you prepare for the release!
Cheers,
Niels
ML Engineer @ HF 🤗