I tried to look for it but didn't find how much better your compiled model is when compared to tensorflow/pytorch natively run on the device. Do you have this somewhere?
We'll be expanding this list and adding more comparisons to other platforms in the near future.