Add support for quantization int4 for faster inference. - View it on GitHub