In this page,TRANSFERING A MODEL FROM PYTORCH TO CAFFE2 AND MOBILE USING ONNX, the output is different.
# Verify the numerical correctness upto 3 decimal places
np.testing.assert_almost_equal(torch_out.data.cpu().numpy(), c2_out, decimal=3)
print("Exported model has been executed on Caffe2 backend, and the result looks good!")
I just copy the code, and the output is :
(mismatch 99.88728564342404%)
x: array([ 0.24 , 0.332, 0.502, ..., 0.025, -0.106, -0.055], dtype=float32)
y: array([ 0.24 , 0.665, 1.075, ..., 0.087, 0.369, -0.055], dtype=float32)