3 ms·8-bit quantization in TensorFlow Lite for faster mobile inference3 points by austin_kodra 7y ago