Post-training Quantization with Multiple Points: Mixed Precision without Mixed Precision | Read Paper on Bytez