We will add quantized CodeGen for fast inference on CPUs up on cformers (https://github.com/NolanoOrg/cformers/) by later today.