Renato Caldas

I'm also experiencing a this but in MicroOS (#12777). In the case of MicroOS there's no way of installing a working version of `ifup` or `ifdown`, so there's no workaround...

I believe both this and https://github.com/hashicorp/vagrant/issues/12762 are caused by a more general problem. Found this 2018 report which is also related to the same thing: https://github.com/cilium/cilium/issues/1918. It seems like Vagrant...

> That won't work or matter. Cura 5.0.0 uses qt6, not qt5. Is qt6 being statically linked? I didn't have qt6 installed, and installing all of it changed nothing.

BTW, a simple workaround is opening the file from the command line: ``` $ cura yourfile.stl ```

I'm also using a system-wide pytorch installation, and had to set `LIBTORCH=/usr`. Perhaps just checking for `libtorch.so` in `/usr/lib/` would be enough.

I've submitted a PR (#344) which should allow using a system-wide installation. I've only added support for Linux, as I don't have access to the other common OS's. @nagisa I'm...

@nagisa that's a good point. It's fairly easy to add additional paths to the search, but perhaps ldconfig is a better alternative.

In `data_utils.py` ~remove/comment~ **change** the `Normalize()` lines to: ```python transform = transforms.Compose([ transforms.ToTensor(), transforms.Normalize((0.5), (0.5)) ]) ```

Turns out you should still normalize the data, just edited my answer above.

I just found this PR, after implementing most of 49fc191 myself xD Could this be added behind an "unstable" feature flag? It should be safe for anyone willing to use...