ENH Add Array API compatibility to cosine_similarity

[EdAbati]

Reference Issues/PRs

Towards #26024

What does this implement/fix? Explain your changes.

It makes thecosine_similarity implementation compatible and tested with the Array API.

Please let me know how I can improve it :)

cc @ogrisel @betatim

[github-actions]

✔️ Linting Passed

All linting checks passed. Your pull request is in excellent shape! ☀️

_{Generated for commit: eb8836c. Link to the linter CI: here}

[ogrisel]

Thanks for the PR @EdAbati. However it already needs an update with conflict resolution w.r.t. the main branch.

Do you have a chance to run the tests on a CUDA host or do you want somebody else to do it? (I can do it).

[EdAbati]

Thank you @ogrisel for the quick reply!

Do you have a chance to run the tests on a CUDA host or do you want somebody else to do it? (I can do it).

Yes, I tested on CUDA in Google Colab and on my Mac (using mps). All green on my side ✅

[ogrisel]

LGTM once the proposed argument renaming discussed above is implemented.

[EdAbati]

Thank you very much!:)

[OmarManzoor]

LGTM. Thanks @EdAbati

[OmarManzoor]

@EdAbati Just for some guidance, how did you run this code in google colab so that we can test with gpus?

[EdAbati]

Thanks again @OmarManzoor :)

I just published the Google Colab notebook in a gist: https://gist.github.com/EdAbati/ff3bdc06bafeb92452b3740686cc8d7c

[OmarManzoor]

Thanks again @OmarManzoor :)

I just published the Google Colab notebook in a gist: https://gist.github.com/EdAbati/ff3bdc06bafeb92452b3740686cc8d7c

Thank you for sharing!