I'm not able to install the Nvidia and CUDA drivers into my Ubuntu Azure VM.

I want to deploy the Llama2 model using a docker container into my azure VM, to achieve this I need to first create a GPU based VM and install the concerned Nvidia and CUDA drivers into the VM, here is where my problem is, I followed 3 methods to achieve this, {( the VM image I used is Ubuntu 22.04, and the VM family is Standard_NV36adms_A10_v5 )},

1. I used the Nvidia GPU driver extension to install the drivers which was the recommended approach in the Microsoft website, https://learn.microsoft.com/en-us/azure/virtual-machines/linux/n-series-driver-setup, but it keeps giving provisioning error.
2. I tried the manual method, the steps for this are also provided in the same link I provided above, the problem I faced here was that the VM wasn't rebooting after I entered "sudo reboot", after this the VM status is always set to "not ready". There was no way I could reconnect to the VM.

3)Then I even tried using the Image provided by Nvidia in the Microsoft Azure Marketplace, to be precise I used this image "NVIDIA GPU-Optimized VMI with vGPU driver - v22.08.0" the issue I faced here is this image has the CUDA driver version 11.2 I need at least 11.8 to run Llama2, I couldn't update the CUDA driver, it wasn't letting me, I also tried other images, "NVIDIA GPU-Optimized VMI - v23.09.1", "NVIDIA GPU-Optimized VMI - v23.03.0" and 2 similar one's the problem with these is when i enter "nvidia-smi", it says no device nvidia found, it should actually provide the specs of the Nvida driver.