initial implementation of unique

[Ferox98]

Modified tests pass for numpy and jax but fail for pytorch and tensorflow. PyTorch's and Tensorflow's implementations of unique are somewhat different from numpy's and jax's. I used a few tricks to make them similar on PyTorch but get an error regarding the dtype of the torch tensor, which for some reason is generating a KeyError. Your help would be highly appreciated.

[djl11]

Thanks a lot for the PR!

I'll need to rebase this to the latest master branch before continuing. Could you please give me write access to the PR? 🙂

[djl11]

Thanks a lot for your contribution. I'll check it out in more detail soon. In the meantime, could you make sure the methods all follow the formatting explained in the contributor guide in the docs? I also go through some examples in the Array API tutorial series on YouTube.

Essentially:

• All methods should be defined with def func_name explicitly, no lambdas or direct binding.
• All Ivy methods should have type-hints and doc strings
• All backend methods should have type-hints, but no doc strings.
• All Array API methods should be moved to the correct sub-modules and files.

Please check the contributor guide in the docs for more details. Thanks!

Additionally, it may be necessary to resolve new conflicts with master since the PR was made. As a guide for resolving conflicts with upstream, I have created this short YouTube series.

Your PR may already be perfect, and if this is the case then no need to do anything!

Some tests will be failing, and thus the X shown, this is fine, as long as the tests for your own function are passing.

I'll take a look at the tests soon and see if it's ready to be merged :)

[djl11]

Please remove all changes to .idea.

Please also reformat the code as describe in the contributor guide in the docs, and also as explained in this YouTube series.

[1Doomdie1]

Closing PR due to inactivity.