mirror of
https://github.com/finegrain-ai/refiners.git
synced 2024-11-23 14:48:45 +00:00
58 lines
1.7 KiB
Python
58 lines
1.7 KiB
Python
from typing import overload
|
|
|
|
import pytest
|
|
import torch
|
|
|
|
from refiners.fluxion.utils import no_grad
|
|
from refiners.foundationals.latent_diffusion import SD1T2IAdapter, SD1UNet, SDXLT2IAdapter, SDXLUNet
|
|
from refiners.foundationals.latent_diffusion.t2i_adapter import T2IFeatures
|
|
|
|
|
|
@overload
|
|
def new_adapter(target: SD1UNet, name: str) -> SD1T2IAdapter:
|
|
...
|
|
|
|
|
|
@overload
|
|
def new_adapter(target: SDXLUNet, name: str) -> SDXLT2IAdapter:
|
|
...
|
|
|
|
|
|
def new_adapter(target: SD1UNet | SDXLUNet, name: str) -> SD1T2IAdapter | SDXLT2IAdapter:
|
|
if isinstance(target, SD1UNet):
|
|
return SD1T2IAdapter(target=target, name=name)
|
|
else:
|
|
return SDXLT2IAdapter(target=target, name=name)
|
|
|
|
|
|
@no_grad()
|
|
@pytest.mark.parametrize("k_unet", [SD1UNet, SDXLUNet])
|
|
def test_inject_eject(k_unet: type[SD1UNet] | type[SDXLUNet], test_device: torch.device):
|
|
unet = k_unet(in_channels=4, device=test_device, dtype=torch.float16)
|
|
initial_repr = repr(unet)
|
|
adapter_1 = new_adapter(unet, "adapter 1")
|
|
assert repr(unet) == initial_repr
|
|
adapter_1.inject()
|
|
assert repr(unet) != initial_repr
|
|
|
|
with pytest.raises(AssertionError) as already_injected_error:
|
|
new_adapter(unet, "adapter 1").inject()
|
|
|
|
assert str(already_injected_error.value) == "T2I-Adapter named adapter 1 is already injected"
|
|
|
|
adapter_2 = new_adapter(unet, "adapter 2").inject()
|
|
|
|
adapter_1.eject()
|
|
|
|
new_adapter_1 = new_adapter(unet, "adapter 1").inject()
|
|
new_adapter_1.eject()
|
|
|
|
assert unet.parent == adapter_2
|
|
assert unet.find(T2IFeatures) is not None
|
|
|
|
adapter_2.eject()
|
|
|
|
assert unet.parent is None
|
|
assert unet.find(T2IFeatures) is None
|
|
assert repr(unet) == initial_repr
|