The sdxl example is not compatible with RTX5070TI-----is there root password could provided for sdxl example?

[whyfp537]

I setup a AI PC with ARM soc with RTX5070TI, after nvidia-ai-workbench installed, and sdxl example clone, the jupyter example of sdxl example cannot work normal.

1. notify peft version should >0.15.2, I try to upgrade peft to 0.17 version, then function OK, but fail at next step.
2. I check torch.cuda.available(), result is false, I try to re-install and upgrade torch_cuda, but sudo need root password.

is there any chance to provide password for the sdxl example container image ?

[ekzotech]

You can get the root access via powershell console with command wsl -u root -d NVIDIA-Workbench. I was able to install and use python3-pip package to play with versions.