Function scale_lora_layers

src/diffusers/utils/peft_utils.py:106–123

Adjust the weightage given to the LoRA layers of the model. Args: model (`torch.nn.Module`): The model to scale. weight (`float`): The weight to be given to the LoRA layers.

(model, weight)

def scale_lora_layers(model, weight):
    """
    Adjust the weightage given to the LoRA layers of the model.

    Args:
        model (`torch.nn.Module`):
            The model to scale.
        weight (`float`):
            The weight to be given to the LoRA layers.
    """
    from peft.tuners.tuners_utils import BaseTunerLayer

    if weight == 1.0:
        return

    for module in model.modules():
        if isinstance(module, BaseTunerLayer):
            module.scale_layer(weight)


def unscale_lora_layers(model, weight: float | None = None):