Quantization and Huffman coding

how to implement or use existing Quantization and Huffman coding in Pytorch?