3 ms·8-Bit Quantization and TFLite: Faster mobile inference with low precision5 points by austin_kodra 8y ago