adapt test

src/diffusers/pipelines/latent_diffusion_uncond/pipeline_latent_diffusion_uncond.py

@@ -36,7 +36,8 @@ class LatentDiffusionUncondPipeline(DiffusionPipeline):
         self.scheduler.set_timesteps(num_inference_steps)
 
         for t in tqdm.tqdm(self.scheduler.timesteps):
-            model_output = self.unet(image, t)
+            with torch.no_grad():
+                model_output = self.unet(image, t)
 
             if isinstance(model_output, dict):
                 model_output = model_output["sample"]
@@ -46,5 +47,6 @@ class LatentDiffusionUncondPipeline(DiffusionPipeline):
             image = self.scheduler.step(model_output, t, image, eta)["prev_sample"]
 
         # decode image with vae
-        image = self.vqvae.decode(image)
+        with torch.no_grad():
+            image = self.vqvae.decode(image)
         return {"sample": image}

tests/test_modeling_utils.py

@@ -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"]