diffusers 0.32.2__py3-none-any.whl → 0.33.0__py3-none-any.whl

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_depth2img.py

@@ -28,11 +28,26 @@ from ...loaders import StableDiffusionLoraLoaderMixin, TextualInversionLoaderMix
 from ...models import AutoencoderKL, UNet2DConditionModel
 from ...models.lora import adjust_lora_scale_text_encoder
 from ...schedulers import KarrasDiffusionSchedulers
-from ...utils import PIL_INTERPOLATION, USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
+from ...utils import (
+    PIL_INTERPOLATION,
+    USE_PEFT_BACKEND,
+    deprecate,
+    is_torch_xla_available,
+    logging,
+    scale_lora_layers,
+    unscale_lora_layers,
+)
 from ...utils.torch_utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
+if is_torch_xla_available():
+    import torch_xla.core.xla_model as xm
+
+    XLA_AVAILABLE = True
+else:
+    XLA_AVAILABLE = False
+
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 
@@ -115,17 +130,21 @@ class StableDiffusionDepth2ImgPipeline(DiffusionPipeline, TextualInversionLoader
     ):
         super().__init__()
 
-        is_unet_version_less_0_9_0 = hasattr(unet.config, "_diffusers_version") and version.parse(
-            version.parse(unet.config._diffusers_version).base_version
-        ) < version.parse("0.9.0.dev0")
-        is_unet_sample_size_less_64 = hasattr(unet.config, "sample_size") and unet.config.sample_size < 64
+        is_unet_version_less_0_9_0 = (
+            unet is not None
+            and hasattr(unet.config, "_diffusers_version")
+            and version.parse(version.parse(unet.config._diffusers_version).base_version) < version.parse("0.9.0.dev0")
+        )
+        is_unet_sample_size_less_64 = (
+            unet is not None and hasattr(unet.config, "sample_size") and unet.config.sample_size < 64
+        )
         if is_unet_version_less_0_9_0 and is_unet_sample_size_less_64:
             deprecation_message = (
                 "The configuration file of the unet has set the default `sample_size` to smaller than"
                 " 64 which seems highly unlikely .If you're checkpoint is a fine-tuned version of any of the"
                 " following: \n- CompVis/stable-diffusion-v1-4 \n- CompVis/stable-diffusion-v1-3 \n-"
-                " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- runwayml/stable-diffusion-v1-5"
-                " \n- runwayml/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the"
+                " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- stable-diffusion-v1-5/stable-diffusion-v1-5"
+                " \n- stable-diffusion-v1-5/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the"
                 " configuration file. Please make sure to update the config accordingly as leaving `sample_size=32`"
                 " in the config might lead to incorrect results in future versions. If you have downloaded this"
                 " checkpoint from the Hugging Face Hub, it would be very nice if you could open a Pull request for"
@@ -145,7 +164,7 @@ class StableDiffusionDepth2ImgPipeline(DiffusionPipeline, TextualInversionLoader
             depth_estimator=depth_estimator,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline._encode_prompt
@@ -861,6 +880,9 @@ class StableDiffusionDepth2ImgPipeline(DiffusionPipeline, TextualInversionLoader
                         step_idx = i // getattr(self.scheduler, "order", 1)
                         callback(step_idx, t, latents)
 
+                if XLA_AVAILABLE:
+                    xm.mark_step()
+
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]
         else:

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_image_variation.py

@@ -24,13 +24,20 @@ from ...configuration_utils import FrozenDict
 from ...image_processor import VaeImageProcessor
 from ...models import AutoencoderKL, UNet2DConditionModel
 from ...schedulers import KarrasDiffusionSchedulers
-from ...utils import deprecate, logging
+from ...utils import deprecate, is_torch_xla_available, logging
 from ...utils.torch_utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
 from . import StableDiffusionPipelineOutput
 from .safety_checker import StableDiffusionSafetyChecker
 
 
+if is_torch_xla_available():
+    import torch_xla.core.xla_model as xm
+
+    XLA_AVAILABLE = True
+else:
+    XLA_AVAILABLE = False
+
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 
@@ -57,8 +64,8 @@ class StableDiffusionImageVariationPipeline(DiffusionPipeline, StableDiffusionMi
             [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
         safety_checker ([`StableDiffusionSafetyChecker`]):
             Classification module that estimates whether generated images could be considered offensive or harmful.
-            Please refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for more details
-            about a model's potential harms.
+            Please refer to the [model card](https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5) for
+            more details about a model's potential harms.
         feature_extractor ([`~transformers.CLIPImageProcessor`]):
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
@@ -97,17 +104,21 @@ class StableDiffusionImageVariationPipeline(DiffusionPipeline, StableDiffusionMi
                 " checker. If you do not want to use the safety checker, you can pass `'safety_checker=None'` instead."
             )
 
-        is_unet_version_less_0_9_0 = hasattr(unet.config, "_diffusers_version") and version.parse(
-            version.parse(unet.config._diffusers_version).base_version
-        ) < version.parse("0.9.0.dev0")
-        is_unet_sample_size_less_64 = hasattr(unet.config, "sample_size") and unet.config.sample_size < 64
+        is_unet_version_less_0_9_0 = (
+            unet is not None
+            and hasattr(unet.config, "_diffusers_version")
+            and version.parse(version.parse(unet.config._diffusers_version).base_version) < version.parse("0.9.0.dev0")
+        )
+        is_unet_sample_size_less_64 = (
+            unet is not None and hasattr(unet.config, "sample_size") and unet.config.sample_size < 64
+        )
         if is_unet_version_less_0_9_0 and is_unet_sample_size_less_64:
             deprecation_message = (
                 "The configuration file of the unet has set the default `sample_size` to smaller than"
                 " 64 which seems highly unlikely .If you're checkpoint is a fine-tuned version of any of the"
                 " following: \n- CompVis/stable-diffusion-v1-4 \n- CompVis/stable-diffusion-v1-3 \n-"
-                " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- runwayml/stable-diffusion-v1-5"
-                " \n- runwayml/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the"
+                " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- stable-diffusion-v1-5/stable-diffusion-v1-5"
+                " \n- stable-diffusion-v1-5/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the"
                 " configuration file. Please make sure to update the config accordingly as leaving `sample_size=32`"
                 " in the config might lead to incorrect results in future versions. If you have downloaded this"
                 " checkpoint from the Hugging Face Hub, it would be very nice if you could open a Pull request for"
@@ -126,7 +137,7 @@ class StableDiffusionImageVariationPipeline(DiffusionPipeline, StableDiffusionMi
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 
@@ -401,6 +412,9 @@ class StableDiffusionImageVariationPipeline(DiffusionPipeline, StableDiffusionMi
                         step_idx = i // getattr(self.scheduler, "order", 1)
                         callback(step_idx, t, latents)
 
+                if XLA_AVAILABLE:
+                    xm.mark_step()
+
         self.maybe_free_model_hooks()
 
         if not output_type == "latent":

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py

@@ -32,6 +32,7 @@ from ...utils import (
     PIL_INTERPOLATION,
     USE_PEFT_BACKEND,
     deprecate,
+    is_torch_xla_available,
     logging,
     replace_example_docstring,
     scale_lora_layers,
@@ -43,8 +44,16 @@ from . import StableDiffusionPipelineOutput
 from .safety_checker import StableDiffusionSafetyChecker
 
 
+if is_torch_xla_available():
+    import torch_xla.core.xla_model as xm
+
+    XLA_AVAILABLE = True
+else:
+    XLA_AVAILABLE = False
+
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
+
 EXAMPLE_DOC_STRING = """
     Examples:
         ```py
@@ -56,7 +65,7 @@ EXAMPLE_DOC_STRING = """
         >>> from diffusers import StableDiffusionImg2ImgPipeline
 
         >>> device = "cuda"
-        >>> model_id_or_path = "runwayml/stable-diffusion-v1-5"
+        >>> model_id_or_path = "stable-diffusion-v1-5/stable-diffusion-v1-5"
         >>> pipe = StableDiffusionImg2ImgPipeline.from_pretrained(model_id_or_path, torch_dtype=torch.float16)
         >>> pipe = pipe.to(device)
 
@@ -205,8 +214,8 @@ class StableDiffusionImg2ImgPipeline(
             [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
         safety_checker ([`StableDiffusionSafetyChecker`]):
             Classification module that estimates whether generated images could be considered offensive or harmful.
-            Please refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for more details
-            about a model's potential harms.
+            Please refer to the [model card](https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5) for
+            more details about a model's potential harms.
         feature_extractor ([`~transformers.CLIPImageProcessor`]):
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
@@ -230,7 +239,7 @@ class StableDiffusionImg2ImgPipeline(
     ):
         super().__init__()
 
-        if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 1:
+        if scheduler is not None and getattr(scheduler.config, "steps_offset", 1) != 1:
             deprecation_message = (
                 f"The configuration file of this scheduler: {scheduler} is outdated. `steps_offset`"
                 f" should be set to 1 instead of {scheduler.config.steps_offset}. Please make sure "
@@ -244,7 +253,7 @@ class StableDiffusionImg2ImgPipeline(
             new_config["steps_offset"] = 1
             scheduler._internal_dict = FrozenDict(new_config)
 
-        if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample is True:
+        if scheduler is not None and getattr(scheduler.config, "clip_sample", False) is True:
             deprecation_message = (
                 f"The configuration file of this scheduler: {scheduler} has not set the configuration `clip_sample`."
                 " `clip_sample` should be set to False in the configuration file. Please make sure to update the"
@@ -273,17 +282,21 @@ class StableDiffusionImg2ImgPipeline(
                 " checker. If you do not want to use the safety checker, you can pass `'safety_checker=None'` instead."
             )
 
-        is_unet_version_less_0_9_0 = hasattr(unet.config, "_diffusers_version") and version.parse(
-            version.parse(unet.config._diffusers_version).base_version
-        ) < version.parse("0.9.0.dev0")
-        is_unet_sample_size_less_64 = hasattr(unet.config, "sample_size") and unet.config.sample_size < 64
+        is_unet_version_less_0_9_0 = (
+            unet is not None
+            and hasattr(unet.config, "_diffusers_version")
+            and version.parse(version.parse(unet.config._diffusers_version).base_version) < version.parse("0.9.0.dev0")
+        )
+        is_unet_sample_size_less_64 = (
+            unet is not None and hasattr(unet.config, "sample_size") and unet.config.sample_size < 64
+        )
         if is_unet_version_less_0_9_0 and is_unet_sample_size_less_64:
             deprecation_message = (
                 "The configuration file of the unet has set the default `sample_size` to smaller than"
                 " 64 which seems highly unlikely. If your checkpoint is a fine-tuned version of any of the"
                 " following: \n- CompVis/stable-diffusion-v1-4 \n- CompVis/stable-diffusion-v1-3 \n-"
-                " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- runwayml/stable-diffusion-v1-5"
-                " \n- runwayml/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the"
+                " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- stable-diffusion-v1-5/stable-diffusion-v1-5"
+                " \n- stable-diffusion-v1-5/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the"
                 " configuration file. Please make sure to update the config accordingly as leaving `sample_size=32`"
                 " in the config might lead to incorrect results in future versions. If you have downloaded this"
                 " checkpoint from the Hugging Face Hub, it would be very nice if you could open a Pull request for"
@@ -304,7 +317,7 @@ class StableDiffusionImg2ImgPipeline(
             feature_extractor=feature_extractor,
             image_encoder=image_encoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 
@@ -1120,6 +1133,9 @@ class StableDiffusionImg2ImgPipeline(
                         step_idx = i // getattr(self.scheduler, "order", 1)
                         callback(step_idx, t, latents)
 
+                if XLA_AVAILABLE:
+                    xm.mark_step()
+
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False, generator=generator)[
                 0

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py

@@ -27,13 +27,27 @@ from ...loaders import FromSingleFileMixin, IPAdapterMixin, StableDiffusionLoraL
 from ...models import AsymmetricAutoencoderKL, AutoencoderKL, ImageProjection, UNet2DConditionModel
 from ...models.lora import adjust_lora_scale_text_encoder
 from ...schedulers import KarrasDiffusionSchedulers
-from ...utils import USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
+from ...utils import (
+    USE_PEFT_BACKEND,
+    deprecate,
+    is_torch_xla_available,
+    logging,
+    scale_lora_layers,
+    unscale_lora_layers,
+)
 from ...utils.torch_utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
 from . import StableDiffusionPipelineOutput
 from .safety_checker import StableDiffusionSafetyChecker
 
 
+if is_torch_xla_available():
+    import torch_xla.core.xla_model as xm
+
+    XLA_AVAILABLE = True
+else:
+    XLA_AVAILABLE = False
+
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 
@@ -146,8 +160,8 @@ class StableDiffusionInpaintPipeline(
             [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
         safety_checker ([`StableDiffusionSafetyChecker`]):
             Classification module that estimates whether generated images could be considered offensive or harmful.
-            Please refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for more details
-            about a model's potential harms.
+            Please refer to the [model card](https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5) for
+            more details about a model's potential harms.
         feature_extractor ([`~transformers.CLIPImageProcessor`]):
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
@@ -171,7 +185,7 @@ class StableDiffusionInpaintPipeline(
     ):
         super().__init__()
 
-        if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 1:
+        if scheduler is not None and getattr(scheduler.config, "steps_offset", 1) != 1:
             deprecation_message = (
                 f"The configuration file of this scheduler: {scheduler} is outdated. `steps_offset`"
                 f" should be set to 1 instead of {scheduler.config.steps_offset}. Please make sure "
@@ -185,7 +199,7 @@ class StableDiffusionInpaintPipeline(
             new_config["steps_offset"] = 1
             scheduler._internal_dict = FrozenDict(new_config)
 
-        if hasattr(scheduler.config, "skip_prk_steps") and scheduler.config.skip_prk_steps is False:
+        if scheduler is not None and getattr(scheduler.config, "skip_prk_steps", True) is False:
             deprecation_message = (
                 f"The configuration file of this scheduler: {scheduler} has not set the configuration"
                 " `skip_prk_steps`. `skip_prk_steps` should be set to True in the configuration file. Please make"
@@ -215,17 +229,21 @@ class StableDiffusionInpaintPipeline(
                 " checker. If you do not want to use the safety checker, you can pass `'safety_checker=None'` instead."
             )
 
-        is_unet_version_less_0_9_0 = hasattr(unet.config, "_diffusers_version") and version.parse(
-            version.parse(unet.config._diffusers_version).base_version
-        ) < version.parse("0.9.0.dev0")
-        is_unet_sample_size_less_64 = hasattr(unet.config, "sample_size") and unet.config.sample_size < 64
+        is_unet_version_less_0_9_0 = (
+            unet is not None
+            and hasattr(unet.config, "_diffusers_version")
+            and version.parse(version.parse(unet.config._diffusers_version).base_version) < version.parse("0.9.0.dev0")
+        )
+        is_unet_sample_size_less_64 = (
+            unet is not None and hasattr(unet.config, "sample_size") and unet.config.sample_size < 64
+        )
         if is_unet_version_less_0_9_0 and is_unet_sample_size_less_64:
             deprecation_message = (
                 "The configuration file of the unet has set the default `sample_size` to smaller than"
                 " 64 which seems highly unlikely .If you're checkpoint is a fine-tuned version of any of the"
                 " following: \n- CompVis/stable-diffusion-v1-4 \n- CompVis/stable-diffusion-v1-3 \n-"
-                " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- runwayml/stable-diffusion-v1-5"
-                " \n- runwayml/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the"
+                " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- stable-diffusion-v1-5/stable-diffusion-v1-5"
+                " \n- stable-diffusion-v1-5/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the"
                 " configuration file. Please make sure to update the config accordingly as leaving `sample_size=32`"
                 " in the config might lead to incorrect results in future versions. If you have downloaded this"
                 " checkpoint from the Hugging Face Hub, it would be very nice if you could open a Pull request for"
@@ -237,7 +255,7 @@ class StableDiffusionInpaintPipeline(
             unet._internal_dict = FrozenDict(new_config)
 
         # Check shapes, assume num_channels_latents == 4, num_channels_mask == 1, num_channels_masked == 4
-        if unet.config.in_channels != 9:
+        if unet is not None and unet.config.in_channels != 9:
             logger.info(f"You have loaded a UNet with {unet.config.in_channels} input channels which.")
 
         self.register_modules(
@@ -250,7 +268,7 @@ class StableDiffusionInpaintPipeline(
             feature_extractor=feature_extractor,
             image_encoder=image_encoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.mask_processor = VaeImageProcessor(
             vae_scale_factor=self.vae_scale_factor, do_normalize=False, do_binarize=True, do_convert_grayscale=True
@@ -642,7 +660,7 @@ class StableDiffusionInpaintPipeline(
         if padding_mask_crop is not None:
             if not isinstance(image, PIL.Image.Image):
                 raise ValueError(
-                    f"The image should be a PIL image when inpainting mask crop, but is of type" f" {type(image)}."
+                    f"The image should be a PIL image when inpainting mask crop, but is of type {type(image)}."
                 )
             if not isinstance(mask_image, PIL.Image.Image):
                 raise ValueError(
@@ -650,7 +668,7 @@ class StableDiffusionInpaintPipeline(
                     f" {type(mask_image)}."
                 )
             if output_type != "pil":
-                raise ValueError(f"The output type should be PIL when inpainting mask crop, but is" f" {output_type}.")
+                raise ValueError(f"The output type should be PIL when inpainting mask crop, but is {output_type}.")
 
         if ip_adapter_image is not None and ip_adapter_image_embeds is not None:
             raise ValueError(
@@ -1014,7 +1032,7 @@ class StableDiffusionInpaintPipeline(
         >>> mask_image = download_image(mask_url).resize((512, 512))
 
         >>> pipe = StableDiffusionInpaintPipeline.from_pretrained(
-        ...     "runwayml/stable-diffusion-inpainting", torch_dtype=torch.float16
+        ...     "stable-diffusion-v1-5/stable-diffusion-inpainting", torch_dtype=torch.float16
         ... )
         >>> pipe = pipe.to("cuda")
 
@@ -1200,7 +1218,7 @@ class StableDiffusionInpaintPipeline(
 
         # 8. Check that sizes of mask, masked image and latents match
         if num_channels_unet == 9:
-            # default case for runwayml/stable-diffusion-inpainting
+            # default case for stable-diffusion-v1-5/stable-diffusion-inpainting
             num_channels_mask = mask.shape[1]
             num_channels_masked_image = masked_image_latents.shape[1]
             if num_channels_latents + num_channels_mask + num_channels_masked_image != self.unet.config.in_channels:
@@ -1208,7 +1226,7 @@ class StableDiffusionInpaintPipeline(
                     f"Incorrect configuration settings! The config of `pipeline.unet`: {self.unet.config} expects"
                     f" {self.unet.config.in_channels} but received `num_channels_latents`: {num_channels_latents} +"
                     f" `num_channels_mask`: {num_channels_mask} + `num_channels_masked_image`: {num_channels_masked_image}"
-                    f" = {num_channels_latents+num_channels_masked_image+num_channels_mask}. Please verify the config of"
+                    f" = {num_channels_latents + num_channels_masked_image + num_channels_mask}. Please verify the config of"
                     " `pipeline.unet` or your `mask_image` or `image` input."
                 )
         elif num_channels_unet != 4:
@@ -1303,6 +1321,9 @@ class StableDiffusionInpaintPipeline(
                         step_idx = i // getattr(self.scheduler, "order", 1)
                         callback(step_idx, t, latents)
 
+                if XLA_AVAILABLE:
+                    xm.mark_step()
+
         if not output_type == "latent":
             condition_kwargs = {}
             if isinstance(self.vae, AsymmetricAutoencoderKL):

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_instruct_pix2pix.py

@@ -22,16 +22,23 @@ from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPV
 
 from ...callbacks import MultiPipelineCallbacks, PipelineCallback
 from ...image_processor import PipelineImageInput, VaeImageProcessor
-from ...loaders import IPAdapterMixin, StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
+from ...loaders import FromSingleFileMixin, IPAdapterMixin, StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
 from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel
 from ...schedulers import KarrasDiffusionSchedulers
-from ...utils import PIL_INTERPOLATION, deprecate, logging
+from ...utils import PIL_INTERPOLATION, deprecate, is_torch_xla_available, logging
 from ...utils.torch_utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
 from . import StableDiffusionPipelineOutput
 from .safety_checker import StableDiffusionSafetyChecker
 
 
+if is_torch_xla_available():
+    import torch_xla.core.xla_model as xm
+
+    XLA_AVAILABLE = True
+else:
+    XLA_AVAILABLE = False
+
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 
@@ -79,6 +86,7 @@ class StableDiffusionInstructPix2PixPipeline(
     TextualInversionLoaderMixin,
     StableDiffusionLoraLoaderMixin,
     IPAdapterMixin,
+    FromSingleFileMixin,
 ):
     r"""
     Pipeline for pixel-level image editing by following text instructions (based on Stable Diffusion).
@@ -106,8 +114,8 @@ class StableDiffusionInstructPix2PixPipeline(
             [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
         safety_checker ([`StableDiffusionSafetyChecker`]):
             Classification module that estimates whether generated images could be considered offensive or harmful.
-            Please refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for more details
-            about a model's potential harms.
+            Please refer to the [model card](https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5) for
+            more details about a model's potential harms.
         feature_extractor ([`~transformers.CLIPImageProcessor`]):
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
@@ -157,7 +165,7 @@ class StableDiffusionInstructPix2PixPipeline(
             feature_extractor=feature_extractor,
             image_encoder=image_encoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 
@@ -393,7 +401,7 @@ class StableDiffusionInstructPix2PixPipeline(
                 f"Incorrect configuration settings! The config of `pipeline.unet`: {self.unet.config} expects"
                 f" {self.unet.config.in_channels} but received `num_channels_latents`: {num_channels_latents} +"
                 f" `num_channels_image`: {num_channels_image} "
-                f" = {num_channels_latents+num_channels_image}. Please verify the config of"
+                f" = {num_channels_latents + num_channels_image}. Please verify the config of"
                 " `pipeline.unet` or your `image` input."
             )
 
@@ -457,6 +465,9 @@ class StableDiffusionInstructPix2PixPipeline(
                         step_idx = i // getattr(self.scheduler, "order", 1)
                         callback(step_idx, t, latents)
 
+                if XLA_AVAILABLE:
+                    xm.mark_step()
+
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]
             image, has_nsfw_concept = self.run_safety_checker(image, device, prompt_embeds.dtype)

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_latent_upscale.py

@@ -25,11 +25,18 @@ from ...image_processor import PipelineImageInput, VaeImageProcessor
 from ...loaders import FromSingleFileMixin
 from ...models import AutoencoderKL, UNet2DConditionModel
 from ...schedulers import EulerDiscreteScheduler
-from ...utils import deprecate, logging
+from ...utils import deprecate, is_torch_xla_available, logging
 from ...utils.torch_utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, StableDiffusionMixin
 
 
+if is_torch_xla_available():
+    import torch_xla.core.xla_model as xm
+
+    XLA_AVAILABLE = True
+else:
+    XLA_AVAILABLE = False
+
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 
@@ -116,7 +123,7 @@ class StableDiffusionLatentUpscalePipeline(DiffusionPipeline, StableDiffusionMix
             unet=unet,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor, resample="bicubic")
 
     def _encode_prompt(
@@ -593,7 +600,7 @@ class StableDiffusionLatentUpscalePipeline(DiffusionPipeline, StableDiffusionMix
                 f"Incorrect configuration settings! The config of `pipeline.unet`: {self.unet.config} expects"
                 f" {self.unet.config.in_channels} but received `num_channels_latents`: {num_channels_latents} +"
                 f" `num_channels_image`: {num_channels_image} "
-                f" = {num_channels_latents+num_channels_image}. Please verify the config of"
+                f" = {num_channels_latents + num_channels_image}. Please verify the config of"
                 " `pipeline.unet` or your `image` input."
             )
 
@@ -640,6 +647,9 @@ class StableDiffusionLatentUpscalePipeline(DiffusionPipeline, StableDiffusionMix
                         step_idx = i // getattr(self.scheduler, "order", 1)
                         callback(step_idx, t, latents)
 
+                if XLA_AVAILABLE:
+                    xm.mark_step()
+
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]
         else:

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_upscale.py

@@ -30,12 +30,26 @@ from ...models.attention_processor import (
 )
 from ...models.lora import adjust_lora_scale_text_encoder
 from ...schedulers import DDPMScheduler, KarrasDiffusionSchedulers
-from ...utils import USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
+from ...utils import (
+    USE_PEFT_BACKEND,
+    deprecate,
+    is_torch_xla_available,
+    logging,
+    scale_lora_layers,
+    unscale_lora_layers,
+)
 from ...utils.torch_utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
 from . import StableDiffusionPipelineOutput
 
 
+if is_torch_xla_available():
+    import torch_xla.core.xla_model as xm
+
+    XLA_AVAILABLE = True
+else:
+    XLA_AVAILABLE = False
+
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 
@@ -149,7 +163,7 @@ class StableDiffusionUpscalePipeline(
             watermarker=watermarker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor, resample="bicubic")
         self.register_to_config(max_noise_level=max_noise_level)
 
@@ -726,7 +740,7 @@ class StableDiffusionUpscalePipeline(
                 f"Incorrect configuration settings! The config of `pipeline.unet`: {self.unet.config} expects"
                 f" {self.unet.config.in_channels} but received `num_channels_latents`: {num_channels_latents} +"
                 f" `num_channels_image`: {num_channels_image} "
-                f" = {num_channels_latents+num_channels_image}. Please verify the config of"
+                f" = {num_channels_latents + num_channels_image}. Please verify the config of"
                 " `pipeline.unet` or your `image` input."
             )
 
@@ -769,6 +783,9 @@ class StableDiffusionUpscalePipeline(
                         step_idx = i // getattr(self.scheduler, "order", 1)
                         callback(step_idx, t, latents)
 
+                if XLA_AVAILABLE:
+                    xm.mark_step()
+
         if not output_type == "latent":
             # make sure the VAE is in float32 mode, as it overflows in float16
             needs_upcasting = self.vae.dtype == torch.float16 and self.vae.config.force_upcast

diffusers/pipelines/stable_diffusion/pipeline_stable_unclip.py

@@ -28,6 +28,7 @@ from ...schedulers import KarrasDiffusionSchedulers
 from ...utils import (
     USE_PEFT_BACKEND,
     deprecate,
+    is_torch_xla_available,
     logging,
     replace_example_docstring,
     scale_lora_layers,
@@ -38,8 +39,16 @@ from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, StableDiffu
 from .stable_unclip_image_normalizer import StableUnCLIPImageNormalizer
 
 
+if is_torch_xla_available():
+    import torch_xla.core.xla_model as xm
+
+    XLA_AVAILABLE = True
+else:
+    XLA_AVAILABLE = False
+
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
+
 EXAMPLE_DOC_STRING = """
     Examples:
         ```py
@@ -132,7 +141,7 @@ class StableUnCLIPPipeline(
         image_noising_scheduler: KarrasDiffusionSchedulers,
         # regular denoising components
         tokenizer: CLIPTokenizer,
-        text_encoder: CLIPTextModelWithProjection,
+        text_encoder: CLIPTextModel,
         unet: UNet2DConditionModel,
         scheduler: KarrasDiffusionSchedulers,
         # vae
@@ -154,7 +163,7 @@ class StableUnCLIPPipeline(
             vae=vae,
         )
 
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
     # Copied from diffusers.pipelines.unclip.pipeline_unclip.UnCLIPPipeline._encode_prompt with _encode_prompt->_encode_prior_prompt, tokenizer->prior_tokenizer, text_encoder->prior_text_encoder
@@ -924,6 +933,9 @@ class StableUnCLIPPipeline(
                 step_idx = i // getattr(self.scheduler, "order", 1)
                 callback(step_idx, t, latents)
 
+            if XLA_AVAILABLE:
+                xm.mark_step()
+
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]
         else: