LoraSave - ComfyUI Built-in Node Documentation

Complete documentation for the LoraSave node in ComfyUI. Learn its inputs, outputs, parameters and usage.

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

The LoraSave node extracts and saves LoRA (Low-Rank Adaptation) files from model differences. It can process diffusion model differences, text encoder differences, or both, converting them into LoRA format with specified rank and type. The resulting LoRA file is saved to the output directory for later use.

Inputs

Parameter	Data Type	Required	Range	Description
filename_prefix	STRING	Yes	-	The prefix for the output filename (default: “loras/ComfyUI_extracted_lora”)
rank	INT	Yes	1-4096	The rank value for the LoRA, controlling the size and complexity (default: 8)
lora_type	COMBO	Yes	Multiple options available	The type of LoRA to create, with various available options
bias_diff	BOOLEAN	Yes	-	Whether to include bias differences in the LoRA calculation (default: True)
model_diff	MODEL	No	-	The ModelSubtract output to be converted to a lora
text_encoder_diff	CLIP	No	-	The CLIPSubtract output to be converted to a lora

Note: At least one of model_diff or text_encoder_diff must be provided for the node to function. If both are omitted, the node will produce no output.

Outputs

Output Name	Data Type	Description
-	-	This node saves a LoRA file to the output directory but does not return any data through the workflow