refiners/tests/foundationals/latent_diffusion/test_freeu.py

70 lines
2.3 KiB
Python
Raw Permalink Normal View History

2023-11-18 14:24:11 +00:00
from typing import Iterator
import pytest
import torch
2023-11-18 14:24:11 +00:00
from refiners.fluxion import manual_seed
from refiners.fluxion.layers import Chain
from refiners.fluxion.utils import no_grad
from refiners.foundationals.latent_diffusion import SD1UNet, SDXLUNet
from refiners.foundationals.latent_diffusion.freeu import FreeUResidualConcatenator, SDFreeUAdapter
2023-11-18 14:24:11 +00:00
@pytest.fixture(scope="module", params=[True, False])
def unet(request: pytest.FixtureRequest) -> Iterator[SD1UNet | SDXLUNet]:
xl: bool = request.param
unet = SDXLUNet(in_channels=4) if xl else SD1UNet(in_channels=4)
yield unet
def test_freeu_adapter(unet: SD1UNet | SDXLUNet) -> None:
freeu = SDFreeUAdapter(unet, backbone_scales=[1.2, 1.2], skip_scales=[0.9, 0.9])
assert len(list(unet.walk(FreeUResidualConcatenator))) == 0
with pytest.raises(AssertionError) as exc:
freeu.eject()
assert "could not find" in str(exc.value)
freeu.inject()
assert len(list(unet.walk(FreeUResidualConcatenator))) == 2
freeu.eject()
assert len(list(unet.walk(FreeUResidualConcatenator))) == 0
def test_freeu_adapter_too_many_scales(unet: SD1UNet | SDXLUNet) -> None:
num_blocks = len(unet.layer("UpBlocks", Chain))
2023-11-18 14:24:11 +00:00
with pytest.raises(AssertionError):
SDFreeUAdapter(unet, backbone_scales=[1.2] * (num_blocks + 1), skip_scales=[0.9] * (num_blocks + 1))
def test_freeu_adapter_inconsistent_scales(unet: SD1UNet | SDXLUNet) -> None:
with pytest.raises(AssertionError):
SDFreeUAdapter(unet, backbone_scales=[1.2, 1.2], skip_scales=[0.9, 0.9, 0.9])
def test_freeu_identity_scales() -> None:
manual_seed(0)
text_embedding = torch.randn(1, 77, 768)
timestep = torch.randint(0, 999, size=(1, 1))
x = torch.randn(1, 4, 32, 32)
unet = SD1UNet(in_channels=4)
unet.set_clip_text_embedding(clip_text_embedding=text_embedding) # not flushed between forward-s
with no_grad():
unet.set_timestep(timestep=timestep)
y_1 = unet(x.clone())
freeu = SDFreeUAdapter(unet, backbone_scales=[1.0, 1.0], skip_scales=[1.0, 1.0])
freeu.inject()
with no_grad():
unet.set_timestep(timestep=timestep)
y_2 = unet(x.clone())
# The FFT -> inverse FFT sequence (skip features) introduces small numerical differences
assert torch.allclose(y_1, y_2, atol=1e-5)