How do you ensure your ML models maintain accuracy while being ported to target hardware for Syntiant edge inference?
