VLCQ: Post-training quantization for deep neural networks using variable length coding | ArxivCSExplorer