Quantization is the process of mapping the high precision values (a large set of possible values) to low precision values(a smaller set of possible values). Quantization can be done on both weights and activations of a model.

The post What is Apple’s Quant for Neural Networks Quantization appeared first on Analytics India Magazine.