Mirrors
/
diffusers
mirror of https://github.com/ShivamShrirao/diffusers.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

adapt test

This commit is contained in:
Patrick von Platen 2022-07-15 18:37:15 +00:00
parent 1c14ce9509
commit 43bbc78123
2 changed files with 6 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/diffusers/pipelines/latent_diffusion_uncond/pipeline_latent_diffusion_uncond.py
View File

@ -36,6 +36,7 @@ class LatentDiffusionUncondPipeline(DiffusionPipeline):
self.scheduler.set_timesteps(num_inference_steps)
for t in tqdm.tqdm(self.scheduler.timesteps):
with torch.no_grad():
model_output = self.unet(image, t)
if isinstance(model_output, dict):
@ -46,5 +47,6 @@ class LatentDiffusionUncondPipeline(DiffusionPipeline):
image = self.scheduler.step(model_output, t, image, eta)["prev_sample"]
# decode image with vae
with torch.no_grad():
image = self.vqvae.decode(image)
return {"sample": image}

3
tests/test_modeling_utils.py
View File

@ -1070,7 +1070,8 @@ class PipelineTesterMixin(unittest.TestCase):
@slow
def test_ldm_uncond(self):
ldm = LatentDiffusionUncondPipeline.from_pretrained("fusing/latent-diffusion-celeba-256", ldm=True)
# ldm = LatentDiffusionUncondPipeline.from_pretrained("fusing/latent-diffusion-celeba-256", ldm=True)
ldm = LatentDiffusionUncondPipeline.from_pretrained("CompVis/latent-diffusion-celeba-256")
generator = torch.manual_seed(0)
image = ldm(generator=generator, num_inference_steps=5)["sample"]