Leveraging Inter-Layer Dependency for Post-Training Quantization

Prior works on Post-training Quantization (PTQ) typically separate a neural network into sub-nets and quantize them sequentially.