Rad2deg

[leoheim]

Details:

• *Added support and test case files for aten::rad2deg *

Tickets:

• thttps://github.com/[Good First Issue]: Support aten::rad2deg #28873

[leoheim]

Hi @rkazants ,

I hope you're doing well! Just wanted to follow up on my PR whenever you have a moment to review them. I’d appreciate any feedback or suggestions.

Thank you!

[mvafin]

Please fix code style

[leoheim]

Hi @mvafin,

I’ve made the requested changes. Let me know if there’s anything else needed!

[mvafin]

Fix code style

src/frontends/pytorch/src/op/rad2deg.cpp:5:+#include <cmath>
src/frontends/pytorch/src/op/rad2deg.cpp:6:+
src/frontends/pytorch/src/op/rad2deg.cpp:6:-#include "openvino/op/multiply.hpp"
src/frontends/pytorch/src/op/rad2deg.cpp:8:-#include <cmath>
src/frontends/pytorch/src/op/rad2deg.cpp:9:+#include "openvino/op/multiply.hpp"
src/frontends/pytorch/src/op/rad2deg.cpp:18:-    
src/frontends/pytorch/src/op/rad2deg.cpp:19:+
src/frontends/pytorch/src/op/rad2deg.cpp:26:-    
src/frontends/pytorch/src/op/rad2deg.cpp:27:+
src/frontends/pytorch/src/op/rad2deg.cpp:29:-    
src/frontends/pytorch/src/op/rad2deg.cpp:30:+
src/frontends/pytorch/src/op_table.cpp:354:-

[leoheim]

Hello @mvafin,

I’ve made the requested changes. Let me know if there’s anything else needed!

[mvafin]

Code style still fails:

src/frontends/pytorch/src/op/rad2deg.cpp:8:-#include "openvino/op/multiply.hpp"
src/frontends/pytorch/src/op/rad2deg.cpp:9:+#include "openvino/op/multiply.hpp"
src/frontends/pytorch/src/op/rad2deg.cpp:19:-    
src/frontends/pytorch/src/op/rad2deg.cpp:19:+
src/frontends/pytorch/src/op/rad2deg.cpp:27:-    
src/frontends/pytorch/src/op/rad2deg.cpp:27:+
src/frontends/pytorch/src/op/rad2deg.cpp:30:-    
src/frontends/pytorch/src/op/rad2deg.cpp:30:+
src/frontends/pytorch/src/op_table.cpp:354:-

You can run clang-format to check for style issues

[leoheim]

Hi @mvafin, I ran clang-format, and I think is correct now. Thanks for your time!

[leoheim]

@mvafin Included now. Thanks!

[mvafin]

why this change?

[mvafin]

why this change?

[leoheim]

Hi @mvafin, I apologize for the confusion. It seems I accidentally mixed changes from another branch into this one. I've corrected the issue now, and the branch should be clean. Thank you for pointing this out!

[mvafin]

@leoheim Please add test. Seems like you removed it by accident

[leoheim]

@mvafin I have added the test now. Apologies for the confusion earlier!

[leoheim]

@mvafin
Thank you for the feedback! I have updated the test to use only one shape ((3, 3)) and added support for multiple data types (np.float32, np.float64, np.int32).

[leoheim]

Hi @mvafin ,

I have updated the code as requested.

[mvafin]

Please fix code style:

src/frontends/pytorch/src/op/rad2deg.cpp:9:-#include "openvino/op/multiply.hpp"
src/frontends/pytorch/src/op/rad2deg.cpp:10:+#include "openvino/op/multiply.hpp"
src/frontends/pytorch/src/op/rad2deg.cpp:31:-    auto conversion_factor_fp64 = context.mark_node(ov::op::v0::Constant::create(ov::element::f64, Shape{}, {180.0 / pi_val}));
src/frontends/pytorch/src/op/rad2deg.cpp:31:+    auto conversion_factor_fp64 =
src/frontends/pytorch/src/op/rad2deg.cpp:32:+        context.mark_node(ov::op::v0::Constant::create(ov::element::f64, Shape{}, {180.0 / pi_val}));
src/frontends/pytorch/src/op/rad2deg.cpp:34:-    auto conversion_factor = context.mark_node(std::make_shared<ov::op::v1::ConvertLike>(conversion_factor_fp64, input));
src/frontends/pytorch/src/op/rad2deg.cpp:35:+    auto conversion_factor =
src/frontends/pytorch/src/op/rad2deg.cpp:36:+        context.mark_node(std::make_shared<ov::op::v1::ConvertLike>(conversion_factor_fp64, input));

[mvafin]

Test for integer case fails due to type difference. Converting to input type is not correct for integer case. Use align_eltwise_input_types with input and conversion_factor_fp64, leave the constant fp64, but pass true as if the second input is python scalar, that way it will scale down to fp32 both input and scalar, not fp64.

[leoheim]

@mvafin I have updated the code as requested. The constant remains as fp64, and I used align_eltwise_input_types with the input and the constant, passing true to handle integer cases correctly. Let me know if there's anything else to adjust. Thanks!