update expected results of slow tests (huggingface#268)

* update expected results of slow tests

* relax sum and mean tests

* Print shapes when reporting exception

* formatting

* fix sentence

* relax test_stable_diffusion_fast_ddim for gpu fp16

* relax flakey tests on GPU

* added comment on large tolerences

* black

* format

* set scheduler seed

* added generator

* use np.isclose

* set num_inference_steps to 50

* fix dep. warning

* update expected_slice

* preprocess if image

* updated expected results

* updated expected from CI

* pass generator to VAE

* undo change back to orig

* use orignal

* revert back the expected on cpu

* revert back values for CPU

* more undo

* update result after using gen

* update mean

* set generator for mps

* update expected on CI server

* undo

* use new seed every time

* cpu manual seed

* reduce num_inference_steps

* style

* use generator for randn

Co-authored-by: Patrick von Platen <[email protected]>

Author: kashif

models/vae.py

@@ -557,7 +557,11 @@ def decode(self, z: torch.FloatTensor, return_dict: bool = True) -> Union[Decode
         return DecoderOutput(sample=dec)
 
     def forward(
-        self, sample: torch.FloatTensor, sample_posterior: bool = False, return_dict: bool = True
+        self,
+        sample: torch.FloatTensor,
+        sample_posterior: bool = False,
+        return_dict: bool = True,
+        generator: Optional[torch.Generator] = None,
     ) -> Union[DecoderOutput, torch.FloatTensor]:
         r"""
         Args:
@@ -570,7 +574,7 @@ def forward(
         x = sample
         posterior = self.encode(x).latent_dist
         if sample_posterior:
-            z = posterior.sample()
+            z = posterior.sample(generator=generator)
         else:
             z = posterior.mode()
         dec = self.decode(z).sample

pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py

@@ -178,7 +178,7 @@ def __call__(
 
         self.scheduler.set_timesteps(num_inference_steps, **extra_set_kwargs)
 
-        if not isinstance(init_image, torch.FloatTensor):
+        if isinstance(init_image, PIL.Image.Image):
             init_image = preprocess(init_image)
 
         # encode the init image into latents and scale the latents