Changes to make this install and run cleanly on Windows.

[Meatfucker]

This includes changes to fix the precision issues in the training scripts, as well as a modification to the conda creation command so it pulls in a complete requirement list, including torch with CUDA.

[OliverKulinski]

This worked for me. The old error no longer shows but now I get CUDA out of memory RuntimeError: CUDA out of memory. Tried to allocate 642.00 MiB (GPU 0; 8.00 GiB total capacity; 7.21 GiB already allocated; 0 bytes free; 7.26 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

[Meatfucker]

The program requires extremely high VRAM. I oom'd about two thirds of the way training on a 24gb card. They recommend a 36gb GPU and they seem to mean it. If someone can get it to work on lower precision, itll use less ram.