Gradient-based Automatic Per-Weight Mixed Precision Quantization for Neural Networks On-Chip

Open in new window