I have a model in pytorch that I can export using onnx. The issue is I have a scaler, currently in sklearn, that I would also like to export with onnx to use in a different language.
Is there a better way than exporting the onnx model while also having two separate sklearn2onnx files for the input variables transform and the target inverse transform?
pipeline search results for pytorch pull up parallel computation which is irrelevant to this question.
I may have the pipeline with just this route because of expected float computations from https://onnx.ai/sklearn-onnx/auto_examples/plot_cast_transformer.html but I'm still just looking for the best route.