Rate limit on Nvidia images

I would like to run one of the MLPERF workloads which uses nvidia docker. According to the readme, the command is

IMAGE=`sudo docker build . | tail -n 1 | awk '{print $3}'

and the dockerfile contains FROM nvidia/cuda:9.0-cudnn7-runtime-ubuntu16.04. But I get the rate limit error as below:

toomanyrequests: You have reached your pull rate limit. You may increase the limit by authenticating and upgrading: https://www.docker.com/increase-rate-limit

I also checked the Docker documents about that. I would like to know what rate limit is imposed by Nvidia and how can I fixed that?