Protobuf Parsing Failed when loading Keras model with custom layer

[gjiremadzeBECKMAN]

Hey all I'm having an error when trying to load a keras model with a custom layer. When trying to call Transforms.ApplyOnnxModel(modelFile) I'm receiving an exception:

Microsoft.ML.OnnxRuntime.OnnxRuntimeException: '[ErrorCode:InvalidProtobuf] Protobuf parsing failed.'

Has anyone tried loading models with custom layers with onnxruntime and have you had any success?

Does anyone know which ONNX opsets are supported in the current ML.NET? Or where I can find this information?

[michaelgsharp]

Latest released version of ML.NET is using OnnxRuntime 1.6. The main branch has been updated to use 1.10. If you need a newer version than 1.6 you can directly reference a newer version in your project and it will overwrite the older version.

If its a custom layer created with Keras and only supported by Keras than this might be an issue. Please try the suggestion and let us know if you still have issues.

[gjiremadzeBECKMAN]

@michaelgsharp I'm able to take the same onnx model in python and load/perform inference on it without a problem. Is the onnxruntime library 1-1 from python to C#?

I'm currently using the NuGet package for ML.NET how would I get the 1.10 version of the DLL?

[gjiremadzeBECKMAN]

Latest released version of ML.NET is using OnnxRuntime 1.6. The main branch has been updated to use 1.10. If you need a newer version than 1.6 you can directly reference a newer version in your project and it will overwrite the older version.

If its a custom layer created with Keras and only supported by Keras than this might be an issue. Please try the suggestion and let us know if you still have issues.

@michaelgsharp
I went ahead and added the Microsoft.ML.Onnxruntime NuGet to my solution and made sure that the version was up to date 1.10 but am still receiving the same error.

I also took the 1.10 DLL and copied it over to the original solution to replace the Onnxruntime.DLL that's included in Microsoft.ML.Transforms.Onnx but still getting the same Protobuf Parsing error.

Would there be a way to get a more descriptive error message to maybe understand what's going on better?

[gjiremadzeBECKMAN]

@michaelgsharp is there any way to get a more descriptive error from the exception?

[michaelgsharp]

Do you know how the custom layer is saved? Is it actualy saved in the onnx model or is it saved as python? Would you be able to upload the onnx model here so we can take a look?

This issue has been automatically marked no-recent-activity because it has not had any activity for 14 days. It will be closed if no further activity occurs within 14 more days. Any new comment (by anyone, not necessarily the author) will remove no-recent-activity.

This issue will now be closed since it had been marked no-recent-activity but received no further activity in the past 14 days. It is still possible to reopen or comment on the issue, but please note that the issue will be locked if it remains inactive for another 30 days.