diffengine.models.archs

Submodules

• diffengine.models.archs.ip_adapter
• diffengine.models.archs.peft

Package Contents

Functions

load_ip_adapter(unet, image_projection, ...)	Load IP-Adapter pretrained weights.
process_ip_adapter_state_dict(unet, image_projection)	Process IP-Adapter state dict.
set_unet_ip_adapter(unet)	Set IP-Adapter for Unet.
create_peft_config(config)	Create a PEFT config from a DiffEngine config.

diffengine.models.archs.load_ip_adapter(unet, image_projection, pretrained_adapter, subfolder, weights_name)

Load IP-Adapter pretrained weights.

Reference to diffusers/loaders/ip_adapter.py. and diffusers/loaders/unet.py.

Parameters:

• unet (torch.nn.Module) –

• image_projection (torch.nn.Module) –

• pretrained_adapter (str) –

• subfolder (str) –

• weights_name (str) –

Return type:

None

diffengine.models.archs.process_ip_adapter_state_dict(unet, image_projection)

Process IP-Adapter state dict.

Parameters:

• unet (torch.nn.Module) –

• image_projection (torch.nn.Module) –

Return type:

dict

diffengine.models.archs.set_unet_ip_adapter(unet)

Set IP-Adapter for Unet.

Args:

unet (nn.Module): The unet to set IP-Adapter.

Parameters:

unet (torch.nn.Module) –

Return type:

None

diffengine.models.archs.create_peft_config(config)

Create a PEFT config from a DiffEngine config.

Args:

config: DiffEngine config.

Parameters:

config (dict) –

Return type:

peft.PeftConfig