product quantization networks (pqns)

These networks use product quantization to compress the weights, reducing memory requirements and speeding up computation.