BlueAmulet

The `_broken_` and `_missing_` names are just names I came up with to highlight the issue. They were originally `metallic` and `roughness`, and the issue manifested itself by making it...

I'm using Windows 10 2004 in English, US-QWERTY keyboard, after testing a bit I can see that it actually says "Processing inode" and gets to around ... 60? before it...

The problem was because of some non UTF-16 filenames in a Windows.old\\$RECYCLE.BIN, after removing them conversion went smoothly

Unfortunately I don't have them anymore. To be honest I think that drive might have been corrupted at some point. $Recycle.Bin is not supposed to be all uppercase and it's...

Apologies for the close and open, accidentally clicked the button. Here is the binary [x86binary.zip](https://github.com/NationalSecurityAgency/ghidra/files/6718777/x86binary.zip)

Why not just run stable diffusion under Windows without WSL? You don't need WSL to be able to use stable diffusion or pytorch or CUDA on Windows.

This occurs in both the official and unofficial versions

If you're using anaconda, it has it's own virtual environment system separate from virtualenv/venv: **To set up the environment:** Create a new environment for so-vits-svc-fork, also adding in pip ```...

The way of fixing/mitigating this is to use a custom unpickler with torch.load that only allows certain safe classes to be used. An example of that can be found here...

> As of pytorch-1.13 there is `weights_only` option designed to mitigate the problem. Good to know that's now a thing, the unpickler idea I referenced at the beginning of this...