diffusers 0.33.1__py3-none-any.whl → 0.35.0__py3-none-any.whl

diffusers/pipelines/chroma/pipeline_output.py

@@ -0,0 +1,21 @@
+from dataclasses import dataclass
+from typing import List, Union
+
+import numpy as np
+import PIL.Image
+
+from ...utils import BaseOutput
+
+
+@dataclass
+class ChromaPipelineOutput(BaseOutput):
+    """
+    Output class for Stable Diffusion pipelines.
+
+    Args:
+        images (`List[PIL.Image.Image]` or `np.ndarray`)
+            List of denoised PIL images of length `batch_size` or numpy array of shape `(batch_size, height, width,
+            num_channels)`. PIL images or numpy array present the denoised images of the diffusion pipeline.
+    """
+
+    images: Union[List[PIL.Image.Image], np.ndarray]

diffusers/pipelines/cogvideo/pipeline_cogvideox.py

@@ -1,4 +1,4 @@
-# Copyright 2024 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
+# Copyright 2025 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
 # All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -359,7 +359,7 @@ class CogVideoXPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
     def prepare_extra_step_kwargs(self, generator, eta):
         # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
         # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
-        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
+        # eta corresponds to η in DDIM paper: https://huggingface.co/papers/2010.02502
         # and should be between [0, 1]
 
         accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
@@ -558,11 +558,11 @@ class CogVideoXPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
                 in their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is
                 passed will be used. Must be in descending order.
             guidance_scale (`float`, *optional*, defaults to 7.0):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `guidance_scale` is defined as `w` of equation 2. of [Imagen
-                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
-                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
-                usually at the expense of lower image quality.
+                Guidance scale as defined in [Classifier-Free Diffusion
+                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
+                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
+                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
+                the text `prompt`, usually at the expense of lower image quality.
             num_videos_per_prompt (`int`, *optional*, defaults to 1):
                 The number of videos to generate per prompt.
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
@@ -645,7 +645,7 @@ class CogVideoXPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
         device = self._execution_device
 
         # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-        # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+        # of the Imagen paper: https://huggingface.co/papers/2205.11487 . `guidance_scale = 1`
         # corresponds to doing no classifier free guidance.
         do_classifier_free_guidance = guidance_scale > 1.0
 
@@ -718,14 +718,15 @@ class CogVideoXPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
                 timestep = t.expand(latent_model_input.shape[0])
 
                 # predict noise model_output
-                noise_pred = self.transformer(
-                    hidden_states=latent_model_input,
-                    encoder_hidden_states=prompt_embeds,
-                    timestep=timestep,
-                    image_rotary_emb=image_rotary_emb,
-                    attention_kwargs=attention_kwargs,
-                    return_dict=False,
-                )[0]
+                with self.transformer.cache_context("cond_uncond"):
+                    noise_pred = self.transformer(
+                        hidden_states=latent_model_input,
+                        encoder_hidden_states=prompt_embeds,
+                        timestep=timestep,
+                        image_rotary_emb=image_rotary_emb,
+                        attention_kwargs=attention_kwargs,
+                        return_dict=False,
+                    )[0]
                 noise_pred = noise_pred.float()
 
                 # perform guidance

diffusers/pipelines/cogvideo/pipeline_cogvideox_fun_control.py

@@ -1,4 +1,4 @@
-# Copyright 2024 The CogVideoX team, Tsinghua University & ZhipuAI, Alibaba-PAI and The HuggingFace Team.
+# Copyright 2025 The CogVideoX team, Tsinghua University & ZhipuAI, Alibaba-PAI and The HuggingFace Team.
 # All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -398,7 +398,7 @@ class CogVideoXFunControlPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
     def prepare_extra_step_kwargs(self, generator, eta):
         # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
         # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
-        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
+        # eta corresponds to η in DDIM paper: https://huggingface.co/papers/2010.02502
         # and should be between [0, 1]
 
         accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
@@ -603,11 +603,11 @@ class CogVideoXFunControlPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
                 in their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is
                 passed will be used. Must be in descending order.
             guidance_scale (`float`, *optional*, defaults to 6.0):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `guidance_scale` is defined as `w` of equation 2. of [Imagen
-                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
-                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
-                usually at the expense of lower image quality.
+                Guidance scale as defined in [Classifier-Free Diffusion
+                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
+                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
+                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
+                the text `prompt`, usually at the expense of lower image quality.
             num_videos_per_prompt (`int`, *optional*, defaults to 1):
                 The number of videos to generate per prompt.
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
@@ -698,7 +698,7 @@ class CogVideoXFunControlPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
         device = self._execution_device
 
         # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-        # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+        # of the Imagen paper: https://huggingface.co/papers/2205.11487 . `guidance_scale = 1`
         # corresponds to doing no classifier free guidance.
         do_classifier_free_guidance = guidance_scale > 1.0
 
@@ -784,14 +784,15 @@ class CogVideoXFunControlPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
                 timestep = t.expand(latent_model_input.shape[0])
 
                 # predict noise model_output
-                noise_pred = self.transformer(
-                    hidden_states=latent_model_input,
-                    encoder_hidden_states=prompt_embeds,
-                    timestep=timestep,
-                    image_rotary_emb=image_rotary_emb,
-                    attention_kwargs=attention_kwargs,
-                    return_dict=False,
-                )[0]
+                with self.transformer.cache_context("cond_uncond"):
+                    noise_pred = self.transformer(
+                        hidden_states=latent_model_input,
+                        encoder_hidden_states=prompt_embeds,
+                        timestep=timestep,
+                        image_rotary_emb=image_rotary_emb,
+                        attention_kwargs=attention_kwargs,
+                        return_dict=False,
+                    )[0]
                 noise_pred = noise_pred.float()
 
                 # perform guidance

diffusers/pipelines/cogvideo/pipeline_cogvideox_image2video.py

@@ -1,4 +1,4 @@
-# Copyright 2024 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
+# Copyright 2025 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
 # All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -442,7 +442,7 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
     def prepare_extra_step_kwargs(self, generator, eta):
         # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
         # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
-        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
+        # eta corresponds to η in DDIM paper: https://huggingface.co/papers/2010.02502
         # and should be between [0, 1]
 
         accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
@@ -658,11 +658,11 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
                 in their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is
                 passed will be used. Must be in descending order.
             guidance_scale (`float`, *optional*, defaults to 7.0):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `guidance_scale` is defined as `w` of equation 2. of [Imagen
-                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
-                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
-                usually at the expense of lower image quality.
+                Guidance scale as defined in [Classifier-Free Diffusion
+                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
+                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
+                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
+                the text `prompt`, usually at the expense of lower image quality.
             num_videos_per_prompt (`int`, *optional*, defaults to 1):
                 The number of videos to generate per prompt.
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
@@ -747,7 +747,7 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
         device = self._execution_device
 
         # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-        # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+        # of the Imagen paper: https://huggingface.co/papers/2205.11487 . `guidance_scale = 1`
         # corresponds to doing no classifier free guidance.
         do_classifier_free_guidance = guidance_scale > 1.0
 
@@ -831,15 +831,16 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
                 timestep = t.expand(latent_model_input.shape[0])
 
                 # predict noise model_output
-                noise_pred = self.transformer(
-                    hidden_states=latent_model_input,
-                    encoder_hidden_states=prompt_embeds,
-                    timestep=timestep,
-                    ofs=ofs_emb,
-                    image_rotary_emb=image_rotary_emb,
-                    attention_kwargs=attention_kwargs,
-                    return_dict=False,
-                )[0]
+                with self.transformer.cache_context("cond_uncond"):
+                    noise_pred = self.transformer(
+                        hidden_states=latent_model_input,
+                        encoder_hidden_states=prompt_embeds,
+                        timestep=timestep,
+                        ofs=ofs_emb,
+                        image_rotary_emb=image_rotary_emb,
+                        attention_kwargs=attention_kwargs,
+                        return_dict=False,
+                    )[0]
                 noise_pred = noise_pred.float()
 
                 # perform guidance

diffusers/pipelines/cogvideo/pipeline_cogvideox_video2video.py

@@ -1,4 +1,4 @@
-# Copyright 2024 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
+# Copyright 2025 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
 # All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -418,7 +418,7 @@ class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
     def prepare_extra_step_kwargs(self, generator, eta):
         # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
         # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
-        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
+        # eta corresponds to η in DDIM paper: https://huggingface.co/papers/2010.02502
         # and should be between [0, 1]
 
         accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
@@ -628,11 +628,11 @@ class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
             strength (`float`, *optional*, defaults to 0.8):
                 Higher strength leads to more differences between original video and generated video.
             guidance_scale (`float`, *optional*, defaults to 7.0):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `guidance_scale` is defined as `w` of equation 2. of [Imagen
-                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
-                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
-                usually at the expense of lower image quality.
+                Guidance scale as defined in [Classifier-Free Diffusion
+                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
+                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
+                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
+                the text `prompt`, usually at the expense of lower image quality.
             num_videos_per_prompt (`int`, *optional*, defaults to 1):
                 The number of videos to generate per prompt.
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
@@ -718,7 +718,7 @@ class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
         device = self._execution_device
 
         # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-        # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+        # of the Imagen paper: https://huggingface.co/papers/2205.11487 . `guidance_scale = 1`
         # corresponds to doing no classifier free guidance.
         do_classifier_free_guidance = guidance_scale > 1.0
 
@@ -799,14 +799,15 @@ class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
                 timestep = t.expand(latent_model_input.shape[0])
 
                 # predict noise model_output
-                noise_pred = self.transformer(
-                    hidden_states=latent_model_input,
-                    encoder_hidden_states=prompt_embeds,
-                    timestep=timestep,
-                    image_rotary_emb=image_rotary_emb,
-                    attention_kwargs=attention_kwargs,
-                    return_dict=False,
-                )[0]
+                with self.transformer.cache_context("cond_uncond"):
+                    noise_pred = self.transformer(
+                        hidden_states=latent_model_input,
+                        encoder_hidden_states=prompt_embeds,
+                        timestep=timestep,
+                        image_rotary_emb=image_rotary_emb,
+                        attention_kwargs=attention_kwargs,
+                        return_dict=False,
+                    )[0]
                 noise_pred = noise_pred.float()
 
                 # perform guidance

diffusers/pipelines/cogview3/pipeline_cogview3plus.py

@@ -1,4 +1,4 @@
-# Copyright 2024 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
+# Copyright 2025 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
 # All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -319,7 +319,7 @@ class CogView3PlusPipeline(DiffusionPipeline):
     def prepare_extra_step_kwargs(self, generator, eta):
         # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
         # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
-        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
+        # eta corresponds to η in DDIM paper: https://huggingface.co/papers/2010.02502
         # and should be between [0, 1]
 
         accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
@@ -390,7 +390,7 @@ class CogView3PlusPipeline(DiffusionPipeline):
         return self._guidance_scale
 
     # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-    # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+    # of the Imagen paper: https://huggingface.co/papers/2205.11487 . `guidance_scale = 1`
     # corresponds to doing no classifier free guidance.
     @property
     def do_classifier_free_guidance(self):
@@ -453,11 +453,11 @@ class CogView3PlusPipeline(DiffusionPipeline):
                 in their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is
                 passed will be used. Must be in descending order.
             guidance_scale (`float`, *optional*, defaults to `5.0`):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `guidance_scale` is defined as `w` of equation 2. of [Imagen
-                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
-                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
-                usually at the expense of lower image quality.
+                Guidance scale as defined in [Classifier-Free Diffusion
+                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
+                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
+                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
+                the text `prompt`, usually at the expense of lower image quality.
             num_images_per_prompt (`int`, *optional*, defaults to `1`):
                 The number of images to generate per prompt.
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
@@ -547,7 +547,7 @@ class CogView3PlusPipeline(DiffusionPipeline):
         device = self._execution_device
 
         # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-        # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+        # of the Imagen paper: https://huggingface.co/papers/2205.11487 . `guidance_scale = 1`
         # corresponds to doing no classifier free guidance.
         do_classifier_free_guidance = guidance_scale > 1.0
 

diffusers/pipelines/cogview4/pipeline_cogview4.py

@@ -1,4 +1,4 @@
-# Copyright 2024 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
+# Copyright 2025 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
 # All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -377,7 +377,7 @@ class CogView4Pipeline(DiffusionPipeline, CogView4LoraLoaderMixin):
         return self._guidance_scale
 
     # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-    # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+    # of the Imagen paper: https://huggingface.co/papers/2205.11487 . `guidance_scale = 1`
     # corresponds to doing no classifier free guidance.
     @property
     def do_classifier_free_guidance(self):
@@ -453,11 +453,11 @@ class CogView4Pipeline(DiffusionPipeline, CogView4LoraLoaderMixin):
                 their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is passed
                 will be used.
             guidance_scale (`float`, *optional*, defaults to `5.0`):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `guidance_scale` is defined as `w` of equation 2. of [Imagen
-                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
-                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
-                usually at the expense of lower image quality.
+                Guidance scale as defined in [Classifier-Free Diffusion
+                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
+                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
+                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
+                the text `prompt`, usually at the expense of lower image quality.
             num_images_per_prompt (`int`, *optional*, defaults to `1`):
                 The number of images to generate per prompt.
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
@@ -619,22 +619,10 @@ class CogView4Pipeline(DiffusionPipeline, CogView4LoraLoaderMixin):
                 # broadcast to batch dimension in a way that's compatible with ONNX/Core ML
                 timestep = t.expand(latents.shape[0])
 
-                noise_pred_cond = self.transformer(
-                    hidden_states=latent_model_input,
-                    encoder_hidden_states=prompt_embeds,
-                    timestep=timestep,
-                    original_size=original_size,
-                    target_size=target_size,
-                    crop_coords=crops_coords_top_left,
-                    attention_kwargs=attention_kwargs,
-                    return_dict=False,
-                )[0]
-
-                # perform guidance
-                if self.do_classifier_free_guidance:
-                    noise_pred_uncond = self.transformer(
+                with self.transformer.cache_context("cond"):
+                    noise_pred_cond = self.transformer(
                         hidden_states=latent_model_input,
-                        encoder_hidden_states=negative_prompt_embeds,
+                        encoder_hidden_states=prompt_embeds,
                         timestep=timestep,
                         original_size=original_size,
                         target_size=target_size,
@@ -643,6 +631,19 @@ class CogView4Pipeline(DiffusionPipeline, CogView4LoraLoaderMixin):
                         return_dict=False,
                     )[0]
 
+                # perform guidance
+                if self.do_classifier_free_guidance:
+                    with self.transformer.cache_context("uncond"):
+                        noise_pred_uncond = self.transformer(
+                            hidden_states=latent_model_input,
+                            encoder_hidden_states=negative_prompt_embeds,
+                            timestep=timestep,
+                            original_size=original_size,
+                            target_size=target_size,
+                            crop_coords=crops_coords_top_left,
+                            attention_kwargs=attention_kwargs,
+                            return_dict=False,
+                        )[0]
                     noise_pred = noise_pred_uncond + self.guidance_scale * (noise_pred_cond - noise_pred_uncond)
                 else:
                     noise_pred = noise_pred_cond

diffusers/pipelines/cogview4/pipeline_cogview4_control.py

@@ -1,4 +1,4 @@
-# Copyright 2024 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
+# Copyright 2025 The CogVideoX team, Tsinghua University & ZhipuAI and The HuggingFace Team.
 # All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -409,7 +409,7 @@ class CogView4ControlPipeline(DiffusionPipeline):
         return self._guidance_scale
 
     # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-    # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+    # of the Imagen paper: https://huggingface.co/papers/2205.11487 . `guidance_scale = 1`
     # corresponds to doing no classifier free guidance.
     @property
     def do_classifier_free_guidance(self):
@@ -486,11 +486,11 @@ class CogView4ControlPipeline(DiffusionPipeline):
                 their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is passed
                 will be used.
             guidance_scale (`float`, *optional*, defaults to `5.0`):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `guidance_scale` is defined as `w` of equation 2. of [Imagen
-                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
-                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
-                usually at the expense of lower image quality.
+                Guidance scale as defined in [Classifier-Free Diffusion
+                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
+                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
+                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
+                the text `prompt`, usually at the expense of lower image quality.
             num_images_per_prompt (`int`, *optional*, defaults to `1`):
                 The number of images to generate per prompt.
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):

diffusers/pipelines/consisid/consisid_utils.py

@@ -166,7 +166,7 @@ def process_face_embeddings(
         raise RuntimeError("facexlib align face fail")
     align_face = face_helper_1.cropped_faces[0]  # (512, 512, 3)  # RGB
 
-    # incase insightface didn't detect face
+    # in case insightface didn't detect face
     if id_ante_embedding is None:
         logger.warning("Failed to detect face using insightface. Extracting embedding with align face")
         id_ante_embedding = face_helper_2.get_feat(align_face)
@@ -294,7 +294,7 @@ def prepare_face_models(model_path, device, dtype):
 
     Parameters:
     - model_path: Path to the directory containing model files.
-    - device: The device (e.g., 'cuda', 'cpu') where models will be loaded.
+    - device: The device (e.g., 'cuda', 'xpu', 'cpu') where models will be loaded.
     - dtype: Data type (e.g., torch.float32) for model inference.
 
     Returns:

diffusers/pipelines/consisid/pipeline_consisid.py

@@ -1,4 +1,4 @@
-# Copyright 2024 ConsisID Authors and The HuggingFace Team. All rights reserved.
+# Copyright 2025 ConsisID Authors and The HuggingFace Team. All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -540,7 +540,7 @@ class ConsisIDPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
     def prepare_extra_step_kwargs(self, generator, eta):
         # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
         # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
-        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
+        # eta corresponds to η in DDIM paper: https://huggingface.co/papers/2010.02502
         # and should be between [0, 1]
 
         accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
@@ -715,11 +715,11 @@ class ConsisIDPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
                 The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                 expense of slower inference.
             guidance_scale (`float`, *optional*, defaults to 6):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `guidance_scale` is defined as `w` of equation 2. of [Imagen
-                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
-                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
-                usually at the expense of lower image quality.
+                Guidance scale as defined in [Classifier-Free Diffusion
+                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
+                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
+                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
+                the text `prompt`, usually at the expense of lower image quality.
             use_dynamic_cfg (`bool`, *optional*, defaults to `False`):
                 If True, dynamically adjusts the guidance scale during inference. This allows the model to use a
                 progressive guidance scale, improving the balance between text-guided generation and image quality over
@@ -821,7 +821,7 @@ class ConsisIDPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
         device = self._execution_device
 
         # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-        # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+        # of the Imagen paper: https://huggingface.co/papers/2205.11487 . `guidance_scale = 1`
         # corresponds to doing no classifier free guidance.
         do_classifier_free_guidance = guidance_scale > 1.0
 

diffusers/pipelines/consistency_models/pipeline_consistency_models.py

@@ -1,4 +1,4 @@
-# Copyright 2024 The HuggingFace Team. All rights reserved.
+# Copyright 2025 The HuggingFace Team. All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diffusers/pipelines/controlnet/pipeline_controlnet.py

@@ -1,4 +1,4 @@
-# Copyright 2024 The HuggingFace Team. All rights reserved.
+# Copyright 2025 The HuggingFace Team. All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -37,7 +37,7 @@ from ...utils import (
     scale_lora_layers,
     unscale_lora_layers,
 )
-from ...utils.torch_utils import is_compiled_module, is_torch_version, randn_tensor
+from ...utils.torch_utils import empty_device_cache, is_compiled_module, is_torch_version, randn_tensor
 from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
 from ..stable_diffusion.pipeline_output import StableDiffusionPipelineOutput
 from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker
@@ -579,7 +579,7 @@ class StableDiffusionControlNetPipeline(
     def prepare_extra_step_kwargs(self, generator, eta):
         # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
         # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
-        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
+        # eta corresponds to η in DDIM paper: https://huggingface.co/papers/2010.02502
         # and should be between [0, 1]
 
         accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
@@ -886,7 +886,7 @@ class StableDiffusionControlNetPipeline(
         return self._clip_skip
 
     # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-    # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+    # of the Imagen paper: https://huggingface.co/papers/2205.11487 . `guidance_scale = 1`
     # corresponds to doing no classifier free guidance.
     @property
     def do_classifier_free_guidance(self):
@@ -979,8 +979,8 @@ class StableDiffusionControlNetPipeline(
             num_images_per_prompt (`int`, *optional*, defaults to 1):
                 The number of images to generate per prompt.
             eta (`float`, *optional*, defaults to 0.0):
-                Corresponds to parameter eta (η) from the [DDIM](https://arxiv.org/abs/2010.02502) paper. Only applies
-                to the [`~schedulers.DDIMScheduler`], and is ignored in other schedulers.
+                Corresponds to parameter eta (η) from the [DDIM](https://huggingface.co/papers/2010.02502) paper. Only
+                applies to the [`~schedulers.DDIMScheduler`], and is ignored in other schedulers.
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
                 generation deterministic.
@@ -1339,7 +1339,7 @@ class StableDiffusionControlNetPipeline(
         if hasattr(self, "final_offload_hook") and self.final_offload_hook is not None:
             self.unet.to("cpu")
             self.controlnet.to("cpu")
-            torch.cuda.empty_cache()
+            empty_device_cache()
 
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False, generator=generator)[

diffusers/pipelines/controlnet/pipeline_controlnet_blip_diffusion.py

@@ -1,5 +1,5 @@
-# Copyright 2024 Salesforce.com, inc.
-# Copyright 2024 The HuggingFace Team. All rights reserved.
+# Copyright 2025 Salesforce.com, inc.
+# Copyright 2025 The HuggingFace Team. All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -29,7 +29,7 @@ from ...utils.torch_utils import randn_tensor
 from ..blip_diffusion.blip_image_processing import BlipImageProcessor
 from ..blip_diffusion.modeling_blip2 import Blip2QFormerModel
 from ..blip_diffusion.modeling_ctx_clip import ContextCLIPTextModel
-from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, ImagePipelineOutput
 
 
 if is_torch_xla_available():
@@ -88,7 +88,7 @@ EXAMPLE_DOC_STRING = """
 """
 
 
-class BlipDiffusionControlNetPipeline(DiffusionPipeline):
+class BlipDiffusionControlNetPipeline(DeprecatedPipelineMixin, DiffusionPipeline):
     """
     Pipeline for Canny Edge based Controlled subject-driven generation using Blip Diffusion.
 
@@ -116,6 +116,7 @@ class BlipDiffusionControlNetPipeline(DiffusionPipeline):
             Position of the context token in the text encoder.
     """
 
+    _last_supported_version = "0.33.1"
     model_cpu_offload_seq = "qformer->text_encoder->unet->vae"
 
     def __init__(
@@ -149,7 +150,7 @@ class BlipDiffusionControlNetPipeline(DiffusionPipeline):
     def get_query_embeddings(self, input_image, src_subject):
         return self.qformer(image_input=input_image, text_input=src_subject, return_dict=False)
 
-    # from the original Blip Diffusion code, speciefies the target subject and augments the prompt by repeating it
+    # from the original Blip Diffusion code, specifies the target subject and augments the prompt by repeating it
     def _build_prompt(self, prompts, tgt_subjects, prompt_strength=1.0, prompt_reps=20):
         rv = []
         for prompt, tgt_subject in zip(prompts, tgt_subjects):
@@ -280,11 +281,11 @@ class BlipDiffusionControlNetPipeline(DiffusionPipeline):
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor will ge generated by random sampling.
             guidance_scale (`float`, *optional*, defaults to 7.5):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `guidance_scale` is defined as `w` of equation 2. of [Imagen
-                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
-                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
-                usually at the expense of lower image quality.
+                Guidance scale as defined in [Classifier-Free Diffusion
+                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
+                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
+                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
+                the text `prompt`, usually at the expense of lower image quality.
             height (`int`, *optional*, defaults to 512):
                 The height of the generated image.
             width (`int`, *optional*, defaults to 512):