added instance method isfinite to pytorch tensor (pytorch frontend)

[VeraChristina]

Hi, I have added the instance method isfinite.
Close #21943
Would you please have a look? Thanks!

[github-actions]

Thanks for contributing to Ivy! 😊👏
Here are some of the important points from our Contributing Guidelines 📝:
1. Feel free to ignore the run_tests (1), run_tests (2), … jobs, and only look at the display_test_results job. 👀 It contains the following two sections:
- Combined Test Results: This shows the results of all the ivy tests that ran on the PR. ✔️
- New Failures Introduced: This lists the tests that are passing on main, but fail on the PR Fork. Please try to make sure that there are no such tests. 💪
2. The lint / Check formatting / check-formatting tests check for the formatting of your code. 📜 If it fails, please check the exact error message in the logs and fix the same. ⚠️🔧
3. Finally, the test-docstrings / run-docstring-tests check for the changes made in docstrings of the functions. This may be skipped, as well. 📚
Happy coding! 🎉👨‍💻

[VeraChristina]

I am not able to link the corresponding issue under development. It is: isfinite #21943
UPDATE: solved!

[ivy-leaves]

If you are working on an open task, please edit the PR description to link to the issue you've created.

For more information, please check ToDo List Issues Guide.

Thank you 🤗

[umairjavaid]

The tests are not passing. Please fix this and request a review

E   ivy.utils.exceptions.IvyBackendException: numpy: nested_map: numpy: nested_map: can't convert np.ndarray of type bfloat16. The only supported types are: float64, float32, float16, complex64, complex128, int64, int32, int16, int8, uint8, and bool.
E   Falsifying example: test_torch_tensor_isfinite(
E       frontend='torch',
E       on_device='cpu',
E       backend_fw='torch',
E       dtype_and_x=(['bfloat16'], [array(-1, dtype=bfloat16)]),
E       init_flags=FrontendMethodTestFlags(
E           num_positional_args=0,
E           as_variable=[False],
E           native_arrays=[False],
E       ),
E       method_flags=FrontendMethodTestFlags(
E           num_positional_args=0,
E           as_variable=[False],
E           native_arrays=[False],
E       ),
E       frontend_method_data=FrontendMethodData(ivy_init_module='ivy.functional.frontends.torch', framework_init_module='torch', init_name='tensor', method_name='isfinite'),
E   )
E   
E   You can reproduce this example by temporarily adding @reproduce_failure('6.82.4', b'AXicY2YAAkYGCIDRDAAAcgAG') as a decorator on your test case

[VeraChristina]

The test test_torch_tensor_isfinite now passes (when run with pytest in my environment), there remain unsuccessful checks, but the error messages are not very telling. Where should I look instead?

[VeraChristina]

@umairjavaid Can't request a review in the sidebar -- it just says 'No reviews'. Can you please help with the above question?

[umairjavaid]

I just checked isfinite supports bfloat16

[umairjavaid]

bool is also supported

[umairjavaid]

uint8 is also supported

[umairjavaid]

int8 is also supported

[umairjavaid]

int16 is also supported

[VeraChristina]

But all those tests fail at a deeper level than this instance method. What's the reason for that?

[VeraChristina]

It seems that they are not supported in certain backends? How do I change that?

[umairjavaid]

complex64 is also supported

[umairjavaid]

complex128 is also supported

[VeraChristina]

But all those tests fail at a deeper level than this instance method (if I exclude the dtype ommissions). What's the reason for that?
It seems that they are not supported in certain backends? How do I change that?
Compare for instance the test result you copied in your first comment above for bfloat16.

[umairjavaid]

Please remove the supported dtype from the @with_unsupported_dtypes

[VeraChristina]

I can do that, but then the tests will fail. (See my comment above this one.)

[VeraChristina]

I have removed the supported dtypes as requested. Now the tests fail.
I have also run the test test_torch_isfinite for the respective function isfinite in the torch frontend (i.e. the one that my instance method calls) and that test fails (with the same error messages as my test does), so apparently this function does not support these dtypes. Or am I getting this totally wrong? I'd be grateful for feedback.

[VeraChristina]

Just for comparison, I ran the tests for the very similar instance method isinf -- same test failures. Please let me know what to do about this PR.

[umairjavaid]

You need add bool to the unsupported dtype decorator of the isfinite tensorflow backend function, so that bool is not passed to it during testing. Similarly, you also need to implement isfinite in paddle backend if it's missing along with its unsupported dtype decorator.

[VeraChristina]

Thank you! This was very helpful. I have added bool to unsupp dtypes in tensorflow backend. The function isfinite in the paddle backend is already implemented along with the unsupported dtype decorator (which looks correct to me) and I don't know why it's not found during test. Similarly, I don't know why the test for the torch backend fails; as far as I can see bfloat16 should be supported.