forked from divamgupta/stable-diffusion-tensorflow
-
Notifications
You must be signed in to change notification settings - Fork 5
Weights
soten355 edited this page Aug 9, 2023
·
1 revision
MetalDiffusion is an empty model until weights are applied. The weights are a result of training using the model. MetalDiffusion does not come with any pre-trained weights for Stable Diffusion. It is up to the user to load weights into the program.
The most important weights are referred to as the "model" or "checkpoint", the weights for the entire model. Without these weights, MetalDiffusion can not create any images.
Text Embeddings are the result of Textural Inversion training. They are commonly referred to as Textural Inversion.
LoRA stands for Low Rank Adaption and are weights from LoRA training.
ControlNet weights are the result of training via the ControlNet method.