Using TensorFlow Lite we see a considerable speed increase when compared with the original results from our previous benchmarks using full TensorFlow . We see an approximately ×2 increase in inferencing speed between the original TensorFlow figures and the new results using TensorFlow Lite.