Skip to content

vllm_omni.diffusion.lora.loader

logger module-attribute 

logger = init_logger(__name__)

lora_convert_mapping module-attribute 

lora_convert_mapping: dict[str, Callable] = {
    "QwenImagePipeline": _convert_non_diffusers_qwen_lora_to_diffusers,
    "QwenImageEditPipeline": _convert_non_diffusers_qwen_lora_to_diffusers,
    "QwenImageEditPlusPipeline": _convert_non_diffusers_qwen_lora_to_diffusers,
    "Wan22Pipeline": _convert_non_diffusers_wan_lora_to_diffusers,
    "Wan22I2VPipeline": _convert_non_diffusers_wan_lora_to_diffusers,
}

LoraLoaderMixin

lora_is_fused property writable 

lora_is_fused: bool

True when LoRA weights are fused into base weights.

Mixin loads are in-place fusions, so non-empty _lora_loaded implies fused.

lora_loaded property writable 

lora_loaded

transformer instance-attribute 

transformer: Module

transformer_name class-attribute instance-attribute 

transformer_name = 'transformer'

load_lora_into_module classmethod 

load_lora_into_module(
    state_dict,
    module,
    prefix: str = "transformer",
    lora_a_suffix: str = "lora_A.weight",
    lora_b_suffix: str = "lora_B.weight",
    lora_bias_suffix: str = "bias",
)

unload_module_lora classmethod 

unload_module_lora(
    state_dict,
    module,
    prefix: str = "transformer",
    lora_a_suffix: str = "lora_A.weight",
    lora_b_suffix: str = "lora_B.weight",
    lora_bias_suffix: str = "bias",
)

QwenImageLoraLoaderMixin

Bases: LoraLoaderMixin

load_lora_weights 

load_lora_weights(
    pretrained_model_name_or_path_or_dict: str
    | dict[str, Tensor],
    adapter_name: str | None = None,
)

unload_lora_weights 

unload_lora_weights(adapter_name: str)

WanLoraLoaderMixin

Bases: LoraLoaderMixin

has_transformer_2 instance-attribute 

has_transformer_2: bool

transformer_2_name class-attribute instance-attribute 

transformer_2_name = 'transformer_2'

load_lora_weights 

load_lora_weights(
    pretrained_model_name_or_path: str | list[str],
    adapter_name: str | None = None,
)

Load LoRA weights into the pipeline's transformer(s).

Parameters:

Name Type Description Default
pretrained_model_name_or_path str | list[str]

Path(s) to concrete .safetensors LoRA files. - str: a single .safetensors file. Only valid for Wan2.1 (single-transformer pipelines). - list[str]: one file per transformer, matched by position to [transformer, transformer_2]. Required for Wan2.2 MoE.

 required
adapter_name str | None

Name to register the adapter under.

 None

unload_lora_weights 

unload_lora_weights(adapter_name: str)

get_converter_by_pipeline 

get_converter_by_pipeline(pipeline)