diff --git a/docs/gpu-k8s-role.md b/docs/gpu-k8s-role.md
index 14e35db..03cf1dd 100644
--- a/docs/gpu-k8s-role.md
+++ b/docs/gpu-k8s-role.md
@@ -30,12 +30,20 @@ sealos run \
   --env '{}' \
   --cmd "kubeadm init --skip-phases=addon/kube-proxy"
 ```
+
 By default the role runs the command as root. Set `root_mode` to `false` to
 deploy in rootless mode, which adds `--user` and `--pk` options pointing to the
 SSH key for `ssh_user`. In rootless mode the host running Sealos must have
 `newuidmap` and `newgidmap` installed (typically provided by the `uidmap`
 package) along with the `fuse-overlayfs` binary to enable user namespaces.
 
+If deploying with a non-root user the command also requires `--user` and
+`--pk` options pointing to the user's SSH key. The host running Sealos must have
+`newuidmap` and `newgidmap` installed (typically provided by the `uidmap` package) 
+along with the `fuse-overlayfs` binary to enable user namespaces.
+
+
+
 After the cluster is running the role installs the NVIDIA device plugin and runs a test pod to ensure `nvidia-smi` works inside the cluster.
 
 ## Usage