Hi @BounharAbdelaziz 🤗
Niels here from the open-source team at Hugging Face. I discovered your work on Arxiv and was wondering whether you would like to submit it to hf.co/papers to improve its discoverability. If you are one of the authors, you can submit it at https://huggingface.co/papers/submit.
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, and add GitHub and project page URLs.
It'd be great to make the learned sparse steering vectors and the cultural alignment benchmark you've curated available on the 🤗 hub, to improve their discoverability and visibility. We can add tags so that people find them when filtering https://huggingface.co/models and https://huggingface.co/datasets.
Uploading steering vectors (models)
Even though these are activation steering vectors rather than full model weights, they are extremely valuable for the mechanistic interpretability and alignment communities. You can host them in model repositories on Hugging Face. See here for a guide: https://huggingface.co/docs/hub/models-uploading.
One can leverage the hf_hub_download one-liner to download specific vectors from the hub.
Uploading dataset
Would be awesome to make the cultural alignment benchmark available on 🤗, so that people can do:
from datasets import load_dataset
dataset = load_dataset("your-hf-org-or-username/cultural-alignment")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.
I see you have a GitHub repository set up at https://github.com/MBZUAI-Paris/YaPO. Let me know if you're interested or need any help regarding hosting these artifacts on the Hub!
Cheers,
Niels
ML Engineer @ HF 🤗
Hi @BounharAbdelaziz 🤗
Niels here from the open-source team at Hugging Face. I discovered your work on Arxiv and was wondering whether you would like to submit it to hf.co/papers to improve its discoverability. If you are one of the authors, you can submit it at https://huggingface.co/papers/submit.
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, and add GitHub and project page URLs.
It'd be great to make the learned sparse steering vectors and the cultural alignment benchmark you've curated available on the 🤗 hub, to improve their discoverability and visibility. We can add tags so that people find them when filtering https://huggingface.co/models and https://huggingface.co/datasets.
Uploading steering vectors (models)
Even though these are activation steering vectors rather than full model weights, they are extremely valuable for the mechanistic interpretability and alignment communities. You can host them in model repositories on Hugging Face. See here for a guide: https://huggingface.co/docs/hub/models-uploading.
One can leverage the hf_hub_download one-liner to download specific vectors from the hub.
Uploading dataset
Would be awesome to make the cultural alignment benchmark available on 🤗, so that people can do:
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.
I see you have a GitHub repository set up at https://github.com/MBZUAI-Paris/YaPO. Let me know if you're interested or need any help regarding hosting these artifacts on the Hub!
Cheers,
Niels
ML Engineer @ HF 🤗