Container won't start after executing command "lxc config set container_name nvidia.runtime=true" from host

LXD
1

I have host and container both using ubuntu 18.04

I want to pass nvidia-smi from host to container following this video LXD and NVIDIA GPUs - YouTube

From the video I executed this command:
lxc config device add container_name gpu gpu gputype=physical

Then executing this command “lxc config set container_name nvidia.runtime=true”

But after executing “lxc config set container_name nvidia.runtime=true”, the container can’t be started with the following error
start_container_error

And after executing “lxc info --show-log cuda1”, the result is like this:

error_lxc_start_cuda1
error_lxc_start_cuda11280×395 133 KB

But after I executing “lxc config set container_name nvidia.runtime=false”, container can be run again

what do I have to do?

thanks in advance