The LD_LIBRARY_PATH needs to include: /usr/local/cuda/lib64/.

Currently, all YAML files exist within specific application directory. Sharing them will make it easier to add supplemental YAML files for new models as well as making common updates.

Update to the latest source if specified. If not specified, do not affect the current codebase.

Create a script based on the instructions on the project website.

After the virtual environment is setup, don't use the version number as part of the Python command.

Instead of using a command like python3.10 -m pip install --upgrade pip
use python -m pip install --upgrade pip.

This will make upgrading python versions easier in the future.

Create a script based on the instructions on the project website.

Create a script based on the instructions on the project website.

Create a script based on the instructions on the project website.

For clarity and readability, use long options instead of short. Readability is more important than keystrokes for scripts.

The directory structure for training embeddings, LoRA, and hypernetworks should be standardized for simplicity. Also, for convenience, so that key directories are not missed.

CUDA 12.3 doesn't work by default. Version 11.8 is expected.

Create a script based on the instructions on the project website.

Create a script based on the instructions on the project website.

When starting to train with Kohya's GUI, a module error is displayed indicating the need for the "scipy" package.