added log_poisson_loss

[rullo16]

Hello, can you please check the code and comment on it? Thank you

Close #16743

[rullo16]

hi @MuhammadSaeedBatikh Can you please give me some feedback? thank you

[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 :hugs:

[MuhammadSaeedBatikh]

Hi @rullo16, thanks a lot for your PR. Apologies for not getting to you earlier. Everything looks good. Just some minor modifications. Since TensorFlow supports float16 for log_poission_loss, there seems to be a problem with this particular dtype conversion for paddle backend.

E           AssertionError:  the results from backend paddle and ground truth framework tensorflow do not match
E            [2.7182817 2.7182817 1.3686724]!=[2.7182817 2.7182817 1.3706255] 
E           
E           
E           Falsifying example: test_log_poisson_loss(
E               dtype_and_targets=(['float16'], [array([0., 0., 2.], dtype=float16)]),
E               dtype_and_log_input=(['float32'], [array([1., 1., 1.], dtype=float32)]),
E               compute_full_loss=True,
E               fn_name='log_poisson_loss',
E               test_flags=FunctionTestFlags(
E                   num_positional_args=0,
E                   with_out=False,
E                   instance_method=False,
E                   test_gradients=None,
E                   test_compile=None,
E                   as_variable=[False],
E                   native_arrays=[False],
E                   container=[False],
E               ),
E               ground_truth_backend='tensorflow',
E               backend_fw=<module 'ivy.functional.backends.paddle' from '/opt/project/ivy/functional/backends/paddle/__init__.py'>,
E               on_device='cpu',
E           )

It seems like ivy.exp is returning slightly different results when running paddle backend for float16. (probably because of type promotion?) Since the difference is "1.370-1.368= 0.002", we can just decrease the absolute tolerance value in tests. Please add atol_=1e-2 to helpers.test_function after compute_full_loss=compute_full_loss

Also, for the sake of completeness, could you please add test function in test_tensorflow/test_nn.py to be consistent.

P.S. Make sure that your first comment starts with Close #issue _num. Thanks.

[rullo16]

Hi @MuhammadSaeedBatikh, thanks for the feedback, everything should be fine now 👍

[MuhammadSaeedBatikh]

Hi @rullo16, lgtm.