Original comment is referring to Nvidia triton inference server