Register Model as LoRA Hook (MO) 🎭🅐🅓

Documentation

• Class name: ADE_RegisterModelAsLoraHookModelOnly
• Category: Animate Diff 🎭🅐🅓/conditioning/register lora hooks
• Output node: False

This node specializes in registering a model as a LoRA hook with a focus on model-only modifications. It enables the integration of LoRA (Low-Rank Adaptation) techniques into a specific model, enhancing its adaptability and performance for specific tasks without affecting other components.

Input types

Required

• model
  • The model to be adapted using LoRA techniques. It serves as the primary target for the application of low-rank adaptations, aiming to enhance its performance or adaptability for specific tasks.
  • Comfy dtype: MODEL
  • Python dtype: ModelPatcher
• ckpt_name
  • The name of the checkpoint to apply LoRA configurations from. This parameter specifies which set of LoRA adaptations to use, guiding the customization of the model's behavior.
  • Comfy dtype: COMBO[STRING]
  • Python dtype: str
• strength_model
  • A floating-point value that determines the intensity of the LoRA adaptation on the model. It modulates how significantly the LoRA parameters influence the model's behavior.
  • Comfy dtype: FLOAT
  • Python dtype: float

Output types

• model
  • Comfy dtype: MODEL
  • The model after applying the specified LoRA adaptations. This output reflects the enhanced or customized version of the model, tailored through the LoRA technique.
  • Python dtype: ModelPatcher
• lora_hook
  • Comfy dtype: LORA_HOOK
  • A reference to the LoRA hook that has been integrated into the model. This output provides access to the LoRA adaptations applied, facilitating further manipulations or analyses.
  • Python dtype: LoraHook

Usage tips

• Infra type: CPU
• Common nodes: unknown

Source code

class MaskableSDModelLoaderModelOnly(MaskableSDModelLoader):
    @classmethod
    def INPUT_TYPES(s):
        return {
            "required": {
                "model": ("MODEL",),
                "ckpt_name": (folder_paths.get_filename_list("checkpoints"), ),
                "strength_model": ("FLOAT", {"default": 1.0, "min": -20.0, "max": 20.0, "step": 0.01}),
            }
        }

    RETURN_TYPES = ("MODEL", "LORA_HOOK")
    CATEGORY = "Animate Diff 🎭🅐🅓/conditioning/register lora hooks"
    FUNCTION = "load_model_as_lora_model_only"

    def load_model_as_lora_model_only(self, model: ModelPatcher, ckpt_name: str, strength_model: float):
        model_lora, clip_lora, lora_hook = self.load_model_as_lora(model=model, clip=None, ckpt_name=ckpt_name,
                                                                   strength_model=strength_model, strength_clip=0)
        return (model_lora, lora_hook)