AI GitHub
kohya_ss for configuring diffusion model training
Configure image-model adaptation jobs through a graphical training interface.
Key Takeaways
- The interface prepares training options for underlying scripts, including data, captions and adapter settings.
- Prepare captioned training data, a base checkpoint and enough memory for the selected recipe.
- Train a small adapter and render fixed validation prompts at different checkpoints.
- Training settings and base model compatibility affect the result.
How it works
The interface prepares training options for underlying scripts, including data, captions and adapter settings. Saving the configuration helps explain what produced a given result.
What you need
Prepare captioned training data, a base checkpoint and enough memory for the selected recipe. Choose fixed validation prompts before adjusting adapter and optimizer settings.
What to try first
Train a small adapter and render fixed validation prompts at different checkpoints. Keep an untrained base output for comparison.
Hold the subject prompt fixed and vary the background at evaluation time. It gives you a direct check for training images that taught the adapter too much about the setting.
Where to look for failures
Training settings and base model compatibility affect the result. Evaluate whether the adapter generalizes beyond the training images and confirm asset permissions.
Related Franklin pages
Separate adapter configuration from evaluation. Use held-out prompts and keep a base-model output so changes in memorization and generalization are visible.
- Run documented adapter and model-training recipes using explicit job configuration. AI Toolkit for configurable diffusion training jobs
- Adapt a supported model by training a subset of parameters and saving compact artifacts. PEFT for training and loading model adapters
Franklin’s image tests give you examples of outputs to compare across visual models. Franklin’s image tests
Project documentation
Try it for your workflow
Explore the project’s documentation for setup and usage, or save this page to a collection for later.
Read the project documentation ↗