CloudDock Header

What Is LoRA

LoRA (Low-Rank Adaptation) is a lightweight method that injects a small file into a base model to change styles, characters, or objects without retraining the whole model.

Key Points

1Small Size

Usually 10–200 MB, far smaller than the multi-gigabyte base model.

2Plug-and-Play

Drop the file into models/Lora, restart the WebUI, and select it in the LoRA tab or prompt with <lora:name:1>.

3Combine Multiple

You can use several LoRAs in one prompt to mix styles, characters, or effects.

4Get LoRAs

Download from the Applications store or import your own .safetensors files through the cloud drive.

LoRA gives you fast, flexible control without replacing the base model.