Skip to main content
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
The SaveLoRA node saves a LoRA (Low-Rank Adaptation) model to a file. It takes a LoRA model as input and writes it to a .safetensors file in the output directory. You can specify a filename prefix and an optional step count to be included in the final filename.

Inputs

ParameterData TypeRequiredRangeDescription
loraMODELYesN/AThe LoRA model to save. Do not use the model with LoRA layers.
prefixSTRINGYesN/AThe prefix to use for the saved LoRA file (default: “loras/ComfyUI_trained_lora”).
stepsINTNoN/AOptional: The number of steps the LoRA has been trained for, used to name the saved file.
Note: The lora input must be a pure LoRA model. Do not provide a base model that has LoRA layers applied to it.

Outputs

Output NameData TypeDescription
NoneN/AThis node does not output any data to the workflow. It is an output node that saves a file to disk.