Dataset card template overhaul

[mariosasko]

It seems to me that new model releases are more likely to have a filled-out (or fewer blank and "out-of-template" sections/fields) repo card than new dataset releases.
So, this PR introduces the following changes to the dataset card template to make working with it less cumbersome and improve the situation (hopefully):

• Aligns the structure and section names with the model card template. Besides consistency, this should also mean less time is needed to write the repo cards for joint model-dataset releases
• Adds (markdown) comments to the sections to make it clear what information a particular section expects
• Removes unnecessary/outdated sections (e.g., the viewer displays data samples, value statistics/ranges, split sizes, and more, so we can simplify the Dataset Structure section) and fields (e.g., homepage, leaderboard, etc.)

[HuggingFaceDocBuilderDev]

The documentation is not available anymore as the PR was closed or merged.

[codecov]

Codecov Report

All modified lines are covered by tests ✅

Comparison is base (90c1182) 82.30% compared to head (af900cc) 82.42%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1708      +/-   ##
==========================================
+ Coverage   82.30%   82.42%   +0.12%     
==========================================
  Files          62       62              
  Lines        7226     7226              
==========================================
+ Hits         5947     5956       +9     
+ Misses       1279     1270       -9     

see 3 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[julien-c]

👍

[julien-c]

looks good to me from a super quick glance

I assume you have checked it still works in DatasetCard.from_template?

cc @EziOzoani @meg-huggingface too on dataset documentation. cc @yjernite too

[Wauplin]

Thanks for opening the PR @mariosasko! Looks good to me as well but from a non-expert point of view.

I posted a message in slack (private) as well. To ease the review, here are the actual files (I found it easier to review than the diff):

• Current dataset card template
• Current model card template
• Suggested dataset card template

[mariosasko]

@julien-c

I assume you have checked it still works in DatasetCard.from_template?

We have tests in tests/test_repocard.py that check this :)

[meg-huggingface]

Looks great, thank you for this!!
Adding some minor changes in wording/spelling.

[mariosasko]

Ready for the final review :)!

(CI failures are unrelated to the changes)

[meg-huggingface]

So happy for this, thanks for the work!

[Wauplin]

Great, I'm merging it and will release it this week :)