Rebase LoRA From Finetune

RebaseLoRAFromFinetune

Applies a source LoRA to the base MODEL/CLIP, subtracts the finetuned MODEL/CLIP, then writes a new LoRA/patch file relative to the finetuned weights.

Pack: ComfyUI-Rebase-LoRA

custom_nodes.ComfyUI-Rebase-LoRA

Inputs (21)

Name	Type	Required
base_model	MODEL	required
finetuned_model	MODEL	required
base_lora_name	COMBO	required
save_relpath	STRING	required
rank	INT	required
layer_scope	COMBO	required
rank_mode	COMBO	required
output_mode	COMBO	required
strength_model	FLOAT	required
strength_clip	FLOAT	required
save_dtype	COMBO	required
bias_diff	BOOLEAN	required
clamp_quantile	FLOAT	required
include_clip	BOOLEAN	required
compute_device	COMBO	required
svd_mode	COMBO	required
cuda_svd_driver	COMBO	required
lowrank_oversample	INT	required
lowrank_niter	INT	required
base_clip	CLIP	optional
finetuned_clip	CLIP	optional

Outputs (2)

Name	Type
saved_lora	STRING
report	STRING