update test slice

src/diffusers/pipelines/animatediff/pipeline_animatediff.py

@@ -16,8 +16,10 @@ import inspect
 from typing import Any, Callable, Dict, List, Optional, Union
 
 import torch
+from packaging import version
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
 
+from ...configuration_utils import FrozenDict
 from ...image_processor import PipelineImageInput
 from ...loaders import IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
 from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel, UNetMotionModel
@@ -124,7 +126,56 @@ class AnimateDiffPipeline(
         image_encoder: CLIPVisionModelWithProjection = None,
     ):
         super().__init__()
+
+        if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 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+                f"The configuration file of this scheduler: {scheduler} is outdated.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 "
+                "to update the config accordingly as leaving `steps_offset` might led 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 the `scheduler/scheduler_config.json`"
+                " file"
+            )
+            deprecate("steps_offset!=1", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["steps_offset"] = 1
+            scheduler._internal_dict = FrozenDict(new_config)
+
+        if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample 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"
+                " config accordingly as not setting `clip_sample` 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 the `scheduler/scheduler_config.json` file"
+            )
+            deprecate("clip_sample not set", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["clip_sample"] = False
+            scheduler._internal_dict = FrozenDict(new_config)
+
         if isinstance(unet, UNet2DConditionModel):
+            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
+            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"
+                    " 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"
+                    " the `unet/config.json` file"
+                )
+                deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False)
+                new_config = dict(unet.config)
+                new_config["sample_size"] = 64
+                unet._internal_dict = FrozenDict(new_config)
+
             unet = UNetMotionModel.from_unet2d(unet, motion_adapter)
 
         self.register_modules(

src/diffusers/pipelines/animatediff/pipeline_animatediff_video2video.py

@@ -16,8 +16,10 @@ import inspect
 from typing import Any, Callable, Dict, List, Optional, Union
 
 import torch
+from packaging import version
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
 
+from ...configuration_utils import FrozenDict
 from ...image_processor import PipelineImageInput
 from ...loaders import IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
 from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel, UNetMotionModel
@@ -31,7 +33,7 @@ from ...schedulers import (
     LMSDiscreteScheduler,
     PNDMScheduler,
 )
-from ...utils import USE_PEFT_BACKEND, logging, scale_lora_layers, unscale_lora_layers
+from ...utils import USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
 from ...utils.torch_utils import randn_tensor
 from ...video_processor import VideoProcessor
 from ..free_init_utils import FreeInitMixin
@@ -228,7 +230,55 @@ class AnimateDiffVideoToVideoPipeline(
         image_encoder: CLIPVisionModelWithProjection = None,
     ):
         super().__init__()
+
+        if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 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 "
+                "to update the config accordingly as leaving `steps_offset` might led 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 the `scheduler/scheduler_config.json`"
+                " file"
+            )
+            deprecate("steps_offset!=1", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["steps_offset"] = 1
+            scheduler._internal_dict = FrozenDict(new_config)
+
+        if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample 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"
+                " config accordingly as not setting `clip_sample` 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 the `scheduler/scheduler_config.json` file"
+            )
+            deprecate("clip_sample not set", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["clip_sample"] = False
+            scheduler._internal_dict = FrozenDict(new_config)
+
         if isinstance(unet, UNet2DConditionModel):
+            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
+            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"
+                    " 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"
+                    " the `unet/config.json` file"
+                )
+                deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False)
+                new_config = dict(unet.config)
+                new_config["sample_size"] = 64
+                unet._internal_dict = FrozenDict(new_config)
             unet = UNetMotionModel.from_unet2d(unet, motion_adapter)
 
         self.register_modules(

src/diffusers/pipelines/pia/pipeline_pia.py

@@ -19,8 +19,10 @@ from typing import Any, Callable, Dict, List, Optional, Union
 import numpy as np
 import PIL
 import torch
+from packaging import version
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
 
+from ...configuration_utils import FrozenDict
 from ...image_processor import PipelineImageInput
 from ...loaders import FromSingleFileMixin, IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
 from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel, UNetMotionModel
@@ -37,6 +39,7 @@ from ...schedulers import (
 from ...utils import (
     USE_PEFT_BACKEND,
     BaseOutput,
+    deprecate,
     logging,
     replace_example_docstring,
     scale_lora_layers,
@@ -183,7 +186,56 @@ class PIAPipeline(
         image_encoder: CLIPVisionModelWithProjection = None,
     ):
         super().__init__()
+
+        if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 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 "
+                "to update the config accordingly as leaving `steps_offset` might led 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 the `scheduler/scheduler_config.json`"
+                " file"
+            )
+            deprecate("steps_offset!=1", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["steps_offset"] = 1
+            scheduler._internal_dict = FrozenDict(new_config)
+
+        if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample 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"
+                " config accordingly as not setting `clip_sample` 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 the `scheduler/scheduler_config.json` file"
+            )
+            deprecate("clip_sample not set", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["clip_sample"] = False
+            scheduler._internal_dict = FrozenDict(new_config)
+
         if isinstance(unet, UNet2DConditionModel):
+            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
+            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"
+                    " 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"
+                    " the `unet/config.json` file"
+                )
+                deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False)
+                new_config = dict(unet.config)
+                new_config["sample_size"] = 64
+                unet._internal_dict = FrozenDict(new_config)
+
             unet = UNetMotionModel.from_unet2d(unet, motion_adapter)
 
         self.register_modules(

src/diffusers/pipelines/stable_diffusion_attend_and_excite/pipeline_stable_diffusion_attend_and_excite.py

@@ -18,9 +18,11 @@ from typing import Any, Callable, Dict, List, Optional, Tuple, Union
 
 import numpy as np
 import torch
+from packaging import version
 from torch.nn import functional as F
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
 
+from ...configuration_utils import FrozenDict
 from ...image_processor import VaeImageProcessor
 from ...loaders import LoraLoaderMixin, TextualInversionLoaderMixin
 from ...models import AutoencoderKL, UNet2DConditionModel
@@ -217,6 +219,33 @@ class StableDiffusionAttendAndExcitePipeline(DiffusionPipeline, StableDiffusionM
     ):
         super().__init__()
 
+        if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 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 "
+                "to update the config accordingly as leaving `steps_offset` might led 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 the `scheduler/scheduler_config.json`"
+                " file"
+            )
+            deprecate("steps_offset!=1", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["steps_offset"] = 1
+            scheduler._internal_dict = FrozenDict(new_config)
+
+        if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample 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"
+                " config accordingly as not setting `clip_sample` 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 the `scheduler/scheduler_config.json` file"
+            )
+            deprecate("clip_sample not set", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["clip_sample"] = False
+            scheduler._internal_dict = FrozenDict(new_config)
+
         if safety_checker is None and requires_safety_checker:
             logger.warning(
                 f"You have disabled the safety checker for {self.__class__} by passing `safety_checker=None`. Ensure"
@@ -233,6 +262,27 @@ class StableDiffusionAttendAndExcitePipeline(DiffusionPipeline, StableDiffusionM
                 " 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
+        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"
+                " 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"
+                " the `unet/config.json` file"
+            )
+            deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(unet.config)
+            new_config["sample_size"] = 64
+            unet._internal_dict = FrozenDict(new_config)
+
         self.register_modules(
             vae=vae,
             text_encoder=text_encoder,

src/diffusers/pipelines/stable_diffusion_panorama/pipeline_stable_diffusion_panorama.py

@@ -16,8 +16,10 @@ import inspect
 from typing import Any, Callable, Dict, List, Optional, Tuple, Union
 
 import torch
+from packaging import version
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
 
+from ...configuration_utils import FrozenDict
 from ...image_processor import PipelineImageInput, VaeImageProcessor
 from ...loaders import IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
 from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel
@@ -188,6 +190,33 @@ class StableDiffusionPanoramaPipeline(
     ):
         super().__init__()
 
+        if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 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 "
+                "to update the config accordingly as leaving `steps_offset` might led 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 the `scheduler/scheduler_config.json`"
+                " file"
+            )
+            deprecate("steps_offset!=1", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["steps_offset"] = 1
+            scheduler._internal_dict = FrozenDict(new_config)
+
+        if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample 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"
+                " config accordingly as not setting `clip_sample` 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 the `scheduler/scheduler_config.json` file"
+            )
+            deprecate("clip_sample not set", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["clip_sample"] = False
+            scheduler._internal_dict = FrozenDict(new_config)
+
         if safety_checker is None and requires_safety_checker:
             logger.warning(
                 f"You have disabled the safety checker for {self.__class__} by passing `safety_checker=None`. Ensure"
@@ -204,6 +233,27 @@ class StableDiffusionPanoramaPipeline(
                 " 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
+        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"
+                " 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"
+                " the `unet/config.json` file"
+            )
+            deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(unet.config)
+            new_config["sample_size"] = 64
+            unet._internal_dict = FrozenDict(new_config)
+
         self.register_modules(
             vae=vae,
             text_encoder=text_encoder,

src/diffusers/pipelines/stable_diffusion_sag/pipeline_stable_diffusion_sag.py

@@ -17,8 +17,10 @@ from typing import Any, Callable, Dict, List, Optional, Union
 
 import torch
 import torch.nn.functional as F
+from packaging import version
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
 
+from ...configuration_utils import FrozenDict
 from ...image_processor import PipelineImageInput, VaeImageProcessor
 from ...loaders import IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
 from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel
@@ -147,6 +149,70 @@ class StableDiffusionSAGPipeline(DiffusionPipeline, StableDiffusionMixin, Textua
     ):
         super().__init__()
 
+        if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 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 "
+                "to update the config accordingly as leaving `steps_offset` might led 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 the `scheduler/scheduler_config.json`"
+                " file"
+            )
+            deprecate("steps_offset!=1", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["steps_offset"] = 1
+            scheduler._internal_dict = FrozenDict(new_config)
+
+        if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample 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"
+                " config accordingly as not setting `clip_sample` 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 the `scheduler/scheduler_config.json` file"
+            )
+            deprecate("clip_sample not set", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["clip_sample"] = False
+            scheduler._internal_dict = FrozenDict(new_config)
+
+        if safety_checker is None and requires_safety_checker:
+            logger.warning(
+                f"You have disabled the safety checker for {self.__class__} by passing `safety_checker=None`. Ensure"
+                " that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered"
+                " results in services or applications open to the public. Both the diffusers team and Hugging Face"
+                " strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling"
+                " it only for use-cases that involve analyzing network behavior or auditing its results. For more"
+                " information, please have a look at https://github.com/huggingface/diffusers/pull/254 ."
+            )
+
+        if safety_checker is not None and feature_extractor is None:
+            raise ValueError(
+                "Make sure to define a feature extractor when loading {self.__class__} if you want to use the safety"
+                " 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
+        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"
+                " 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"
+                " the `unet/config.json` file"
+            )
+            deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(unet.config)
+            new_config["sample_size"] = 64
+            unet._internal_dict = FrozenDict(new_config)
+
         self.register_modules(
             vae=vae,
             text_encoder=text_encoder,

src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_adapter.py

@@ -19,8 +19,10 @@ from typing import Any, Callable, Dict, List, Optional, Union
 import numpy as np
 import PIL.Image
 import torch
+from packaging import version
 from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
 
+from ...configuration_utils import FrozenDict
 from ...image_processor import VaeImageProcessor
 from ...loaders import LoraLoaderMixin, TextualInversionLoaderMixin
 from ...models import AutoencoderKL, MultiAdapter, T2IAdapter, UNet2DConditionModel
@@ -230,6 +232,33 @@ class StableDiffusionAdapterPipeline(DiffusionPipeline, StableDiffusionMixin):
     ):
         super().__init__()
 
+        if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 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 "
+                "to update the config accordingly as leaving `steps_offset` might led 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 the `scheduler/scheduler_config.json`"
+                " file"
+            )
+            deprecate("steps_offset!=1", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["steps_offset"] = 1
+            scheduler._internal_dict = FrozenDict(new_config)
+
+        if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample 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"
+                " config accordingly as not setting `clip_sample` 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 the `scheduler/scheduler_config.json` file"
+            )
+            deprecate("clip_sample not set", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(scheduler.config)
+            new_config["clip_sample"] = False
+            scheduler._internal_dict = FrozenDict(new_config)
+
         if safety_checker is None and requires_safety_checker:
             logger.warning(
                 f"You have disabled the safety checker for {self.__class__} by passing `safety_checker=None`. Ensure"
@@ -246,6 +275,27 @@ class StableDiffusionAdapterPipeline(DiffusionPipeline, StableDiffusionMixin):
                 " 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
+        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"
+                " 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"
+                " the `unet/config.json` file"
+            )
+            deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False)
+            new_config = dict(unet.config)
+            new_config["sample_size"] = 64
+            unet._internal_dict = FrozenDict(new_config)
+
         if isinstance(adapter, (list, tuple)):
             adapter = MultiAdapter(adapter)
 

tests/pipelines/animatediff/test_animatediff.py

@@ -141,33 +141,33 @@ class AnimateDiffPipelineFastTests(
         if torch_device == "cpu":
             expected_pipe_slice = np.array(
                 [
-                    0.5541,
+                    0.55489814,
-                    0.5802,
+                    0.57733,
-                    0.5074,
+                    0.5055906,
-                    0.4583,
+                    0.45875436,
-                    0.4729,
+                    0.47063118,
-                    0.5374,
+                    0.5364321,
-                    0.4051,
+                    0.40426704,
-                    0.4495,
+                    0.44867495,
-                    0.4480,
+                    0.44591445,
-                    0.5292,
+                    0.5281508,
-                    0.6322,
+                    0.6315466,
-                    0.6265,
+                    0.62435544,
-                    0.5455,
+                    0.54455715,
-                    0.4771,
+                    0.477045,
-                    0.5795,
+                    0.5789583,
-                    0.5845,
+                    0.58372897,
-                    0.4172,
+                    0.41623285,
-                    0.6066,
+                    0.60586095,
-                    0.6535,
+                    0.65277576,
-                    0.4113,
+                    0.41136372,
-                    0.6833,
+                    0.6803191,
-                    0.5736,
+                    0.5731553,
-                    0.3589,
+                    0.3580435,
-                    0.5730,
+                    0.5695902,
-                    0.4205,
+                    0.42043707,
-                    0.3786,
+                    0.3784843,
-                    0.5323,
+                    0.53030723,
                 ]
             )
         return super().test_ip_adapter_single(expected_pipe_slice=expected_pipe_slice)
@@ -175,7 +175,19 @@ class AnimateDiffPipelineFastTests(
     def test_dict_tuple_outputs_equivalent(self):
         expected_slice = None
         if torch_device == "cpu":
-            expected_slice = np.array([0.4051, 0.4495, 0.4480, 0.5845, 0.4172, 0.6066, 0.4205, 0.3786, 0.5323])
+            expected_slice = np.array(
+                [
+                    0.40426707,
+                    0.4486752,
+                    0.44591454,
+                    0.583729,
+                    0.41623282,
+                    0.60586107,
+                    0.42043722,
+                    0.37848428,
+                    0.5303071,
+                ]
+            )
         return super().test_dict_tuple_outputs_equivalent(expected_slice=expected_slice)
 
     def test_inference_batch_single_identical(

tests/pipelines/animatediff/test_animatediff_video2video.py

@@ -143,24 +143,24 @@ class AnimateDiffVideoToVideoPipelineFastTests(
         if torch_device == "cpu":
             expected_pipe_slice = np.array(
                 [
-                    0.4947,
+                    0.4848796,
-                    0.4780,
+                    0.48586836,
-                    0.4340,
+                    0.443857,
-                    0.4666,
+                    0.4659326,
-                    0.4028,
+                    0.40280268,
-                    0.4645,
+                    0.46792588,
-                    0.4915,
+                    0.48351467,
-                    0.4101,
+                    0.4108287,
-                    0.4308,
+                    0.43699834,
-                    0.4581,
+                    0.4561751,
-                    0.3582,
+                    0.35249728,
-                    0.4953,
+                    0.49432203,
-                    0.4466,
+                    0.44738623,
-                    0.5348,
+                    0.5392915,
-                    0.5863,
+                    0.5923746,
-                    0.5299,
+                    0.53159416,
-                    0.5213,
+                    0.52147824,
-                    0.5017,
+                    0.5025812,
                 ]
             )
         return super().test_ip_adapter_single(expected_pipe_slice=expected_pipe_slice)

tests/pipelines/pia/test_pia.py

@@ -144,33 +144,33 @@ class PIAPipelineFastTests(IPAdapterTesterMixin, PipelineTesterMixin, PipelineFr
         if torch_device == "cpu":
             expected_pipe_slice = np.array(
                 [
-                    0.5609,
+                    0.56209797,
-                    0.5756,
+                    0.5758661,
-                    0.4830,
+                    0.48313263,
-                    0.4420,
+                    0.4432918,
-                    0.4547,
+                    0.4548767,
-                    0.5129,
+                    0.51363844,
-                    0.3779,
+                    0.37824085,
-                    0.4042,
+                    0.404838,
-                    0.3772,
+                    0.37794176,
-                    0.4450,
+                    0.44477728,
-                    0.5710,
+                    0.5720737,
-                    0.5536,
+                    0.55278736,
-                    0.4835,
+                    0.48393112,
-                    0.4308,
+                    0.43125585,
-                    0.5578,
+                    0.5575704,
-                    0.5578,
+                    0.55773294,
-                    0.4395,
+                    0.44008094,
-                    0.5440,
+                    0.5437152,
-                    0.6051,
+                    0.6051959,
-                    0.4651,
+                    0.4654531,
-                    0.6258,
+                    0.62552464,
-                    0.5662,
+                    0.56681484,
-                    0.3988,
+                    0.39887077,
-                    0.5108,
+                    0.5120458,
-                    0.4153,
+                    0.41586423,
-                    0.3993,
+                    0.39992803,
-                    0.4803,
+                    0.48050576,
                 ]
             )
         return super().test_ip_adapter_single(expected_pipe_slice=expected_pipe_slice)
@@ -178,7 +178,19 @@ class PIAPipelineFastTests(IPAdapterTesterMixin, PipelineTesterMixin, PipelineFr
     def test_dict_tuple_outputs_equivalent(self):
         expected_slice = None
         if torch_device == "cpu":
-            expected_slice = np.array([0.3740, 0.4284, 0.4038, 0.5417, 0.4405, 0.5521, 0.4273, 0.4124, 0.4997])
+            expected_slice = np.array(
+                [
+                    0.37420893,
+                    0.4281358,
+                    0.40268794,
+                    0.54140866,
+                    0.43910965,
+                    0.55197185,
+                    0.427696,
+                    0.41110283,
+                    0.49882218,
+                ]
+            )
         return super().test_dict_tuple_outputs_equivalent(expected_slice=expected_slice)
 
     @unittest.skip("Attention slicing is not enabled in this pipeline")

tests/pipelines/stable_diffusion_gligen/test_stable_diffusion_gligen.py

@@ -34,7 +34,6 @@ from ..pipeline_params import (
     TEXT_TO_IMAGE_PARAMS,
 )
 from ..test_pipelines_common import (
-    PipelineFromPipeTesterMixin,
     PipelineKarrasSchedulerTesterMixin,
     PipelineLatentTesterMixin,
     PipelineTesterMixin,
@@ -48,7 +47,6 @@ class GligenPipelineFastTests(
     PipelineLatentTesterMixin,
     PipelineKarrasSchedulerTesterMixin,
     PipelineTesterMixin,
-    PipelineFromPipeTesterMixin,
     unittest.TestCase,
 ):
     pipeline_class = StableDiffusionGLIGENPipeline

tests/pipelines/stable_diffusion_gligen_text_image/test_stable_diffusion_gligen_text_image.py

@@ -43,7 +43,6 @@ from ..pipeline_params import (
     TEXT_TO_IMAGE_PARAMS,
 )
 from ..test_pipelines_common import (
-    PipelineFromPipeTesterMixin,
     PipelineKarrasSchedulerTesterMixin,
     PipelineLatentTesterMixin,
     PipelineTesterMixin,
@@ -57,7 +56,6 @@ class GligenTextImagePipelineFastTests(
     PipelineLatentTesterMixin,
     PipelineKarrasSchedulerTesterMixin,
     PipelineTesterMixin,
-    PipelineFromPipeTesterMixin,
     unittest.TestCase,
 ):
     pipeline_class = StableDiffusionGLIGENTextImagePipeline

tests/pipelines/stable_diffusion_panorama/test_stable_diffusion_panorama.py

@@ -132,7 +132,9 @@ class StableDiffusionPanoramaPipelineFastTests(
         image_slice = image[0, -3:, -3:, -1]
         assert image.shape == (1, 64, 64, 3)
 
-        expected_slice = np.array([0.6186, 0.5374, 0.4915, 0.4135, 0.4114, 0.4563, 0.5128, 0.4977, 0.4757])
+        expected_slice = np.array(
+            [0.63905126, 0.62803805, 0.48622578, 0.5133644, 0.5549811, 0.4580565, 0.503026, 0.5019732, 0.45383182]
+        )
 
         assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
 
@@ -148,8 +150,9 @@ class StableDiffusionPanoramaPipelineFastTests(
         image_slice = image[0, -3:, -3:, -1]
         assert image.shape == (1, 64, 64, 3)
 
-        expected_slice = np.array([0.6127, 0.6299, 0.4595, 0.4051, 0.4543, 0.3925, 0.5510, 0.5693, 0.5031])
+        expected_slice = np.array(
-
+            [0.60976464, 0.6527951, 0.45149416, 0.49849328, 0.53450716, 0.39567426, 0.55500215, 0.5935768, 0.52607167]
+        )
         assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
 
     # override to speed the overall test timing up.
@@ -178,7 +181,9 @@ class StableDiffusionPanoramaPipelineFastTests(
 
         assert image.shape == (1, 64, 64, 3)
 
-        expected_slice = np.array([0.6187, 0.5375, 0.4915, 0.4136, 0.4114, 0.4563, 0.5128, 0.4976, 0.4757])
+        expected_slice = np.array(
+            [0.6391751, 0.6281538, 0.48634097, 0.51326203, 0.5548724, 0.45805028, 0.5029195, 0.5018764, 0.45382288]
+        )
 
         assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
 
@@ -196,7 +201,9 @@ class StableDiffusionPanoramaPipelineFastTests(
 
         assert image.shape == (1, 64, 64, 3)
 
-        expected_slice = np.array([0.6187, 0.5375, 0.4915, 0.4136, 0.4114, 0.4563, 0.5128, 0.4976, 0.4757])
+        expected_slice = np.array(
+            [0.63905126, 0.62803805, 0.48622578, 0.5133644, 0.5549811, 0.4580565, 0.503026, 0.5019732, 0.45383182]
+        )
 
         assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
 
@@ -214,7 +221,9 @@ class StableDiffusionPanoramaPipelineFastTests(
 
         assert image.shape == (1, 64, 64, 3)
 
-        expected_slice = np.array([0.6127, 0.6299, 0.4595, 0.4051, 0.4543, 0.3925, 0.5510, 0.5693, 0.5031])
+        expected_slice = np.array(
+            [0.6097647, 0.6527951, 0.4514941, 0.49849334, 0.53450716, 0.39567426, 0.5550021, 0.5935768, 0.5260716]
+        )
 
         assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
 

tests/pipelines/test_pipelines_common.py

@@ -301,7 +301,7 @@ class IPAdapterTesterMixin:
         self.assertLess(
             max_diff_without_adapter_scale,
             expected_max_diff,
-            "Output without ip-adapter must be same as normal inference",
+            f"Output without ip-adapter must be same as normal inference, exppected {output_without_adapter}, got {output_without_adapter_scale}",
         )
         self.assertGreater(
             max_diff_with_adapter_scale, 1e-2, "Output with ip-adapter must be different from normal inference"
@@ -658,6 +658,14 @@ class PipelineFromPipeTesterMixin:
 
         # create original_pipeline_class(sd/sdxl)
         pipe_original = self.original_pipeline_class.from_pretrained(original_repo, **original_kwargs)
+        model_config_original = {}
+        for name, component in pipe_original.components.items():
+            if not hasattr(component, "config"):
+                continue
+            if hasattr(component.config, "to_dict"):
+                model_config_original[name] = component.config.to_dict()
+            else:
+                model_config_original[name] = dict(component.config)
 
         # original_pipeline_class(sd/sdxl) -> pipeline_class
         pipe_components = self.get_dummy_components()
@@ -675,23 +683,43 @@ class PipelineFromPipeTesterMixin:
                 original_pipe_additional_components[name] = component
 
         pipe_original_2 = self.original_pipeline_class.from_pipe(pipe, **original_pipe_additional_components)
+        model_config_original_2 = {}
+        for name, component in pipe_original_2.components.items():
+            if not hasattr(component, "config"):
+                continue
+            if hasattr(component.config, "to_dict"):
+                model_config_original_2[name] = component.config.to_dict()
+            else:
+                model_config_original_2[name] = dict(component.config)
 
         # compare the config
         original_config = {k: v for k, v in pipe_original.config.items() if not k.startswith("_")}
         original_config_2 = {k: v for k, v in pipe_original_2.config.items() if not k.startswith("_")}
         assert original_config_2 == original_config
 
+        assert set(model_config_original_2) == set(model_config_original)
+        for name, config_dict in model_config_original_2.items():
+            assert config_dict == model_config_original[name]
+
     def test_from_pipe_consistent_forward_pass(self, expected_max_diff=1e-3):
         components = self.get_dummy_components()
-        original_expected_modules, _ = self.original_pipeline_class._get_signature_keys(self.original_pipeline_class)
 
+        pipe = self.pipeline_class(**components)
+        for component in pipe.components.values():
+            if hasattr(component, "set_default_attn_processor"):
+                component.set_default_attn_processor()
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+        inputs = self.get_dummy_inputs_pipe(torch_device)
+        output = pipe(**inputs)[0]
+
+        original_expected_modules, _ = self.original_pipeline_class._get_signature_keys(self.original_pipeline_class)
         # pipeline components that are also expected to be in the original pipeline
         original_pipe_components = {}
         # additional components that are not in the pipeline, but expected in the original pipeline
         original_pipe_additional_components = {}
         # additional components that are in the pipeline, but not expected in the original pipeline
         current_pipe_additional_components = {}
-
         for name, component in components.items():
             if name in original_expected_modules:
                 original_pipe_components[name] = component
@@ -713,15 +741,6 @@ class PipelineFromPipeTesterMixin:
         inputs = self.get_dummy_inputs_for_pipe_original(torch_device)
         output_original = pipe_original(**inputs)[0]
 
-        pipe = self.pipeline_class(**components)
-        for component in pipe.components.values():
-            if hasattr(component, "set_default_attn_processor"):
-                component.set_default_attn_processor()
-        pipe.to(torch_device)
-        pipe.set_progress_bar_config(disable=None)
-        inputs = self.get_dummy_inputs_pipe(torch_device)
-        output = pipe(**inputs)[0]
-
         pipe_from_original = self.pipeline_class.from_pipe(pipe_original, **current_pipe_additional_components)
         pipe_from_original.to(torch_device)
         pipe_from_original.set_progress_bar_config(disable=None)