feat(frontends): Implement rfft in PaddlePaddle frontend and fix fft for Tensorflow backend

[AwkNinja]

Close #19383

[AwkNinja]

Still working on fixing the failing tests for tensorflow and torch.

[AwkNinja]

@hmahmood24 Can you please have a look at the implementation here. I can't get the tensorflow and torch tests to pass.

[AwkNinja]

The torch tests seem to be passing now but still having issues with tensorflow. The main error is that the elements list is empty so it throws the error cannot draw from an empty list. At first I thought it was because of the norm list that I am generating but it doesn't seem like its related to it.

[hmahmood24]

Why can't we use something like this?

dtype_input_axis=helpers.dtype_values_axis(
        available_dtypes=helpers.get_dtypes("valid"),
        min_num_dims=1,
        min_dim_size=2,
        shape=helpers.get_shape(...),
        force_int_axis=True,
        valid_axis=True,
        allow_neg_axis=True,
    )

[hmahmood24]

We also need to test for the default n=None case

[AwkNinja]

Hi @hmahmood24 sorry for the late update but please take a look.

[hmahmood24]

@asad0991 Perhaps you forgot to install/run pre-commit to fix your lint/formatting errors?

[AwkNinja]

@hmahmood24 Fixed the lint issues I changed devices forgot to install pre-commit on the new device.

[hmahmood24]

@AwkNinja I can see a couple of failing tests still associated with this PR

[AwkNinja]

@hmahmood24 The tests for tensorflow gives me keyError: "cup" its like the error you get when the version is wrong in the decorator but I used the correct one. For the paddle error it just started to occur just now. Its basically in ivy.dft function I'll take a look at it.

[hmahmood24]

@hmahmood24 The tests for tensorflow gives me keyError: "cup" its like the error you get when the version is wrong in the decorator but I used the correct one. For the paddle error it just started to occur just now. Its basically in ivy.dft function I'll take a look at it.

Pretty sure you can find a solution to that key error problem with a simple search in discord since that's been brought up numerous times already

[AwkNinja]

@hmahmood24 The tests for tensorflow gives me keyError: "cup" its like the error you get when the version is wrong in the decorator but I used the correct one. For the paddle error it just started to occur just now. Its basically in ivy.dft function I'll take a look at it.

Pretty sure you can find a solution to that key error problem with a simple search in discord since that's been brought up numerous times already

I did read on discord changing version to 2.5.1 and below should have fixed it. I'll have a look again might be another solution someone brought up.

[github-actions]

PR Compliance Checks Passed!

[AwkNinja]

Hi @hmahmood24, the tensorflow tests are passing now with the change I made but It should no longer be needed once this feature is implemented. Also for the paddle tests I found the issue here. Basically query_ is a slice object and at some point value for query_ is slice(0, ivy.Array(2), None) hence giving the error TypeError: slice indices must be integers or None or have an __index__ method. Maybe someone with more experience in the handling of paddle framework can look into this.

[hmahmood24]

Hi @hmahmood24, the tensorflow tests are passing now with the change I made but It should no longer be needed once this feature is implemented. Also for the paddle tests I found the issue here. Basically query_ is a slice object and at some point value for query_ is slice(0, ivy.Array(2), None) hence giving the error TypeError: slice indices must be integers or None or have an __index__ method. Maybe someone with more experience in the handling of paddle framework can look into this.

Hey @AwkNinja, thanks a lot for the effort. I'll ask someone from the team to take a look at this issue you brought up with _parse_query. In the meanwhile I think it should be okay to merge the PR because this issue stems from set_item itself and not with your PR. What do you think @AnnaTz ?

[AnnaTz]

Hi @hmahmood24, the tensorflow tests are passing now with the change I made but It should no longer be needed once this feature is implemented. Also for the paddle tests I found the issue here. Basically query_ is a slice object and at some point value for query_ is slice(0, ivy.Array(2), None) hence giving the error TypeError: slice indices must be integers or None or have an __index__ method. Maybe someone with more experience in the handling of paddle framework can look into this.

Hey @AwkNinja, thanks a lot for the effort. I'll ask someone from the team to take a look at this issue you brought up with _parse_query. In the meanwhile I think it should be okay to merge the PR because this issue stems from set_item itself and not with your PR. What do you think @AnnaTz ?

This wasn't a set_item issue. I have just resolved it with a quick fix in ivy.dft.

[AwkNinja]

Hi @hmahmood24 all the test cases are passing now with the fix that @AnnaTz did. Please take a look.