Added tf.diag frontend + test

[mobley-trent]

Closes #13868

[ivy-seed]

This PR has been labelled as stale because it has been inactive for more than 7 days. If you would like to continue working on this PR, then please add another comment or this PR will be closed in 7 days.

[mobley-trent]

Hi @zaeemansari70, I am requesting a code review 🙂.
I've implemented the test but I am getting this error:

FAILED ivy_tests/test_ivy/test_frontends/test_tensorflow/test_linalg.py::test_tensorflow_diag[cpu-ivy.functional.backends.numpy-False-False] - AttributeError: 'NoneType' object has no attribute 'backend'
FAILED ivy_tests/test_ivy/test_frontends/test_tensorflow/test_linalg.py::test_tensorflow_diag[cpu-ivy.functional.backends.torch-False-False] - AttributeError: 'NoneType' object has no attribute 'backend'

Does this mean the backends are not yet implemented ?

[mobley-trent]

Hey @hirwa-nshuti I implemented the function with the appropriate parameters but I am getting the same error that I got beforehand

[AnnaTz]

Hi @mobley-trent, I tried running your code locally and I don't get the AttributeError: 'NoneType' object has no attribute 'backend' that you mentioned. Are you getting the same when you try running other tests as well? If so, there is something wrong with your ide setup. Otherwise, I would suggest syncing with the latest master and trying again then.

[fnhirwa]

Hey @mobley-trent

Good Job so far,
But there is a mismatch between the shapes of returned array and the ground truth one🙂
Would you look into that?

[mobley-trent]

@hirwa-nshuti I actually had a question about the testing. The test case which is failing produces this array:

x = array([[0, 0],[0, 0]], dtype=int64)
k = 0

But when I test this input on the diag function I have implemented and tf.linalg.diag, they both produce the same array with shape (2, 2, 2) yet the error trace for the test failure indicates that my function returns an array of shape (2, 3, 3). How is this so ? I tested the function in a Jupyter notebook to make sure that both the ground truth function and my implementation produce the same outputs.

[fnhirwa]

Hey @mobley-trent

Now looks great but the tests for paddle backend are failing, fix this then we can merge this PR.

Great job done😊

[fnhirwa]

Hey @mobley-trent

Thanks for contributing😊
Going to merge