missing GPU after reboot

I using Standard NC24rs_v3 (24 <g class="gr_ gr_45 gr-alert gr_spell gr_inline_cards gr_disable_anim_appear ContextualSpelling" data-gr-id="45" id="45">vcpus</g>, 448 GB memory) size VMs with the ``data science virtual machine'' image in Azure. It works fine <g class="gr_ gr_196 gr-alert gr_gramm gr_inline_cards gr_disable_anim_appear Grammar multiReplace" data-gr-id="196" id="196">at</g> the first times with 4 GPUs. But if I rebooted the VM, <g class="gr_ gr_313 gr-alert gr_spell gr_inline_cards gr_disable_anim_appear ContextualSpelling ins-del multiReplace" data-gr-id="313" id="313">nvidia</g>-<g class="gr_ gr_314 gr-alert gr_spell gr_inline_cards gr_disable_anim_appear ContextualSpelling ins-del multiReplace" data-gr-id="314" id="314">smi</g> can only find 1 GPU. I had to recreate a new VM again and again. I think this's a bug about the <g class="gr_ gr_548 gr-alert gr_spell gr_inline_cards gr_disable_anim_appear ContextualSpelling ins-del multiReplace" data-gr-id="548" id="548">cuda</g> driver.

I used the following commands to install CUDA 10.

sudo apt-get -y remove --purge nvidia*
sudo apt-get -y autoremove
mkdir ~/tmp
cd ~/tmp
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/cuda-repo-ubuntu1604_10.0.130-1_amd64.deb
sudo dpkg -i cuda-repo-ubuntu1604_10.0.130-1_amd64.deb
sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub
sudo apt-get update
sudo apt-get -y install cuda