Improving Quantization-aware Training of Low-Precision Network via Block Replacement on Full-Precision Counterpart

Open in new window