Skip to main content
This node creates a hook that applies a LoRA (Low-Rank Adaptation) model to modify only the model component of a neural network. It loads a checkpoint file and applies it with a specified strength to the model while leaving the CLIP component unchanged. This is an experimental node that extends the functionality of the base CreateHookModelAsLora class.

Inputs

ParameterDescriptionData TypeRequiredRange
ckpt_nameThe checkpoint file to load as a LoRA model. Available options depend on the checkpoints folder contents.STRINGYesMultiple options available
strength_modelThe strength multiplier for applying the LoRA to the model component (default: 1.0)FLOATYes-20.0 to 20.0
prev_hooksOptional previous hooks to chain with this hookHOOKSNo-

Outputs

Output NameDescriptionData Type
hooksThe created hook group containing the LoRA model modificationHOOKS
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): adbeaede65aa89d48c59225ca1c8edc4c9394a364f93a00dae4a83a2270f093b