Omar Sanseviero

As discussed in DMs, feel free to open PRs directly on this feedback. For BibTeX, you can use https://github.com/huggingface/diffusers#citation as a reference. If someone disagrees with any feedback/change, it will...

Hey there! We're adding advanced modelcard functionalities that might make this easier. CC @nateraw

Actually for csv pandas has `usecols` which allows loading a subset of columns in a more efficient way afaik, but yes, you're right this might be more complex than I...

There are multiple BigCode models (1.1B, 3B, 7B, and 15B), so there are smaller version indeed. As for which model, it depends on what exact behaviour one would want. https://twitter.com/lvwerra/status/1691127139314159628...

The ideal setup would be for a member of Orion AI Lab to create an account and an organization for the lab, where they can add other team members. They...

cc @lhoestq and @severo on it, but the Dataset Preview uses the `datasets` library to load the dataset. From doing ```python from datasets import load_dataset dataset = load_dataset("paren8esis/S4A") ``` I...

Note that this PR is already incorporated in https://github.com/huggingface/huggingface_hub/pull/940/files#diff-5b64a68c6ba7c6aeb5c5ebfb29207ae15efa93fe7001a3a149bae41af1d1f29f, in particular https://github.com/huggingface/huggingface_hub/blob/5db0c978bf15c7a9c492c804fc842507fcd055f8/src/huggingface_hub/templates/modelcard_template.md

Yes, feel free to open a PR against that PR as adding more features directly in the original PR would make the PR too large.

What would be the expected behaviour for you?

For all tasks we have a formal spec that is used in downstream applications (e.g. inference API, API clients, etc). For example, here is the spec for `feature-extraction` https://github.com/huggingface/huggingface.js/tree/main/packages/tasks/src/tasks/feature-extraction/spec which...