Why do we need onnx2torch runtime dependency after the onnx model is converted to Torch

[satishlokkoju]

I was able to successfully convert a ONNX model to Pytorch.

I observed that to run the pytorch model we need onnx2torch depndency. Why is that the case ?

Once the is converted and saved as torch model, why do we need the dependency ?

[goberreu]

Because the package uses it for testing.