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

diffusers/pipelines/stable_audio/modeling_stable_audio.py

@@ -1,4 +1,4 @@
-# Copyright 2024 Stability AI and The HuggingFace Team. All rights reserved.
+# Copyright 2025 Stability AI 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.

diffusers/pipelines/stable_audio/pipeline_stable_audio.py

@@ -1,4 +1,4 @@
-# Copyright 2024 Stability AI and The HuggingFace Team. All rights reserved.
+# Copyright 2025 Stability AI 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.
@@ -306,7 +306,7 @@ class StableAudioPipeline(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())
@@ -526,8 +526,8 @@ class StableAudioPipeline(DiffusionPipeline):
             num_waveforms_per_prompt (`int`, *optional*, defaults to 1):
                 The number of waveforms 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.
@@ -616,7 +616,7 @@ class StableAudioPipeline(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/stable_cascade/pipeline_stable_cascade.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.
@@ -332,11 +332,11 @@ class StableCascadeDecoderPipeline(DiffusionPipeline):
                 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 0.0):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `decoder_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
-                `decoder_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). `decoder_guidance_scale` is defined as `w` of
+                equation 2. of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by
+                setting `decoder_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.
             negative_prompt (`str` or `List[str]`, *optional*):
                 The prompt or prompts not to guide the image generation. Ignored when not using guidance (i.e., ignored
                 if `decoder_guidance_scale` is less than `1`).
@@ -524,9 +524,9 @@ class StableCascadeDecoderPipeline(DiffusionPipeline):
             latents = self.vqgan.config.scale_factor * latents
             images = self.vqgan.decode(latents).sample.clamp(0, 1)
             if output_type == "np":
-                images = images.permute(0, 2, 3, 1).cpu().float().numpy()  # float() as bfloat16-> numpy doesnt work
+                images = images.permute(0, 2, 3, 1).cpu().float().numpy()  # float() as bfloat16-> numpy doesn't work
             elif output_type == "pil":
-                images = images.permute(0, 2, 3, 1).cpu().float().numpy()  # float() as bfloat16-> numpy doesnt work
+                images = images.permute(0, 2, 3, 1).cpu().float().numpy()  # float() as bfloat16-> numpy doesn't work
                 images = self.numpy_to_pil(images)
         else:
             images = latents

diffusers/pipelines/stable_cascade/pipeline_stable_cascade_combined.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.
@@ -125,7 +125,7 @@ class StableCascadeCombinedPipeline(DiffusionPipeline):
     def enable_xformers_memory_efficient_attention(self, attention_op: Optional[Callable] = None):
         self.decoder_pipe.enable_xformers_memory_efficient_attention(attention_op)
 
-    def enable_model_cpu_offload(self, gpu_id: Optional[int] = None, device: Union[torch.device, str] = "cuda"):
+    def enable_model_cpu_offload(self, gpu_id: Optional[int] = None, device: Union[torch.device, str] = None):
         r"""
         Offloads all models to CPU using accelerate, reducing memory usage with a low impact on performance. Compared
         to `enable_sequential_cpu_offload`, this method moves one whole model at a time to the GPU when its `forward`
@@ -135,7 +135,7 @@ class StableCascadeCombinedPipeline(DiffusionPipeline):
         self.prior_pipe.enable_model_cpu_offload(gpu_id=gpu_id, device=device)
         self.decoder_pipe.enable_model_cpu_offload(gpu_id=gpu_id, device=device)
 
-    def enable_sequential_cpu_offload(self, gpu_id: Optional[int] = None, device: Union[torch.device, str] = "cuda"):
+    def enable_sequential_cpu_offload(self, gpu_id: Optional[int] = None, device: Union[torch.device, str] = None):
         r"""
         Offloads all models (`unet`, `text_encoder`, `vae`, and `safety checker` state dicts) to CPU using 🤗
         Accelerate, significantly reducing memory usage. Models are moved to a `torch.device('meta')` and loaded on a
@@ -212,11 +212,11 @@ class StableCascadeCombinedPipeline(DiffusionPipeline):
             width (`int`, *optional*, defaults to 512):
                 The width in pixels of the generated image.
             prior_guidance_scale (`float`, *optional*, defaults to 4.0):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `prior_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
-                `prior_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). `prior_guidance_scale` is defined as `w` of
+                equation 2. of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by
+                setting `prior_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.
             prior_num_inference_steps (`Union[int, Dict[float, int]]`, *optional*, defaults to 60):
                 The number of prior denoising steps. More denoising steps usually lead to a higher quality image at the
                 expense of slower inference. For more specific timestep spacing, you can pass customized
@@ -226,11 +226,11 @@ class StableCascadeCombinedPipeline(DiffusionPipeline):
                 the expense of slower inference. For more specific timestep spacing, you can pass customized
                 `timesteps`
             decoder_guidance_scale (`float`, *optional*, defaults to 0.0):
-                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.
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                 to make generation deterministic.

diffusers/pipelines/stable_cascade/pipeline_stable_cascade_prior.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.
@@ -409,11 +409,11 @@ class StableCascadePriorPipeline(DiffusionPipeline):
                 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 8.0):
-                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
-                `decoder_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
-                `decoder_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). `decoder_guidance_scale` is defined as `w` of
+                equation 2. of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by
+                setting `decoder_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.
             negative_prompt (`str` or `List[str]`, *optional*):
                 The prompt or prompts not to guide the image generation. Ignored when not using guidance (i.e., ignored
                 if `decoder_guidance_scale` is less than `1`).
@@ -626,11 +626,11 @@ class StableCascadePriorPipeline(DiffusionPipeline):
         self.maybe_free_model_hooks()
 
         if output_type == "np":
-            latents = latents.cpu().float().numpy()  # float() as bfloat16-> numpy doesnt work
-            prompt_embeds = prompt_embeds.cpu().float().numpy()  # float() as bfloat16-> numpy doesnt work
+            latents = latents.cpu().float().numpy()  # float() as bfloat16-> numpy doesn't work
+            prompt_embeds = prompt_embeds.cpu().float().numpy()  # float() as bfloat16-> numpy doesn't work
             negative_prompt_embeds = (
                 negative_prompt_embeds.cpu().float().numpy() if negative_prompt_embeds is not None else None
-            )  # float() as bfloat16-> numpy doesnt work
+            )  # float() as bfloat16-> numpy doesn't work
 
         if not return_dict:
             return (

diffusers/pipelines/stable_diffusion/__init__.py

@@ -30,18 +30,11 @@ except OptionalDependencyNotAvailable:
     _dummy_objects.update(get_objects_from_module(dummy_torch_and_transformers_objects))
 else:
     _import_structure["clip_image_project_model"] = ["CLIPImageProjection"]
-    _import_structure["pipeline_cycle_diffusion"] = ["CycleDiffusionPipeline"]
     _import_structure["pipeline_stable_diffusion"] = ["StableDiffusionPipeline"]
-    _import_structure["pipeline_stable_diffusion_attend_and_excite"] = ["StableDiffusionAttendAndExcitePipeline"]
-    _import_structure["pipeline_stable_diffusion_gligen"] = ["StableDiffusionGLIGENPipeline"]
-    _import_structure["pipeline_stable_diffusion_gligen_text_image"] = ["StableDiffusionGLIGENTextImagePipeline"]
     _import_structure["pipeline_stable_diffusion_img2img"] = ["StableDiffusionImg2ImgPipeline"]
     _import_structure["pipeline_stable_diffusion_inpaint"] = ["StableDiffusionInpaintPipeline"]
-    _import_structure["pipeline_stable_diffusion_inpaint_legacy"] = ["StableDiffusionInpaintPipelineLegacy"]
     _import_structure["pipeline_stable_diffusion_instruct_pix2pix"] = ["StableDiffusionInstructPix2PixPipeline"]
     _import_structure["pipeline_stable_diffusion_latent_upscale"] = ["StableDiffusionLatentUpscalePipeline"]
-    _import_structure["pipeline_stable_diffusion_model_editing"] = ["StableDiffusionModelEditingPipeline"]
-    _import_structure["pipeline_stable_diffusion_paradigms"] = ["StableDiffusionParadigmsPipeline"]
     _import_structure["pipeline_stable_diffusion_upscale"] = ["StableDiffusionUpscalePipeline"]
     _import_structure["pipeline_stable_unclip"] = ["StableUnCLIPPipeline"]
     _import_structure["pipeline_stable_unclip_img2img"] = ["StableUnCLIPImg2ImgPipeline"]

diffusers/pipelines/stable_diffusion/clip_image_project_model.py

@@ -1,4 +1,4 @@
-# Copyright 2024 The GLIGEN Authors and HuggingFace Team. All rights reserved.
+# Copyright 2025 The GLIGEN Authors and 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/stable_diffusion/convert_from_ckpt.py

@@ -53,6 +53,7 @@ from ...schedulers import (
 )
 from ...utils import is_accelerate_available, logging
 from ...utils.constants import DIFFUSERS_REQUEST_TIMEOUT
+from ...utils.torch_utils import get_device
 from ..latent_diffusion.pipeline_latent_diffusion import LDMBertConfig, LDMBertModel
 from ..paint_by_example import PaintByExampleImageEncoder
 from ..pipeline_utils import DiffusionPipeline
@@ -350,8 +351,14 @@ def create_vae_diffusers_config(original_config, image_size: int):
     _ = original_config["model"]["params"]["first_stage_config"]["params"]["embed_dim"]
 
     block_out_channels = [vae_params["ch"] * mult for mult in vae_params["ch_mult"]]
-    down_block_types = ["DownEncoderBlock2D"] * len(block_out_channels)
-    up_block_types = ["UpDecoderBlock2D"] * len(block_out_channels)
+    down_block_types = [
+        "DownEncoderBlock2D" if image_size // 2**i not in vae_params["attn_resolutions"] else "AttnDownEncoderBlock2D"
+        for i, _ in enumerate(block_out_channels)
+    ]
+    up_block_types = [
+        "UpDecoderBlock2D" if image_size // 2**i not in vae_params["attn_resolutions"] else "AttnUpDecoderBlock2D"
+        for i, _ in enumerate(block_out_channels)
+    ][::-1]
 
     config = {
         "sample_size": image_size,
@@ -1266,7 +1273,7 @@ def download_from_original_stable_diffusion_ckpt(
             checkpoint = safe_load(checkpoint_path_or_dict, device="cpu")
         else:
             if device is None:
-                device = "cuda" if torch.cuda.is_available() else "cpu"
+                device = get_device()
                 checkpoint = torch.load(checkpoint_path_or_dict, map_location=device)
             else:
                 checkpoint = torch.load(checkpoint_path_or_dict, map_location=device)
@@ -1836,7 +1843,7 @@ def download_controlnet_from_original_ckpt(
                 checkpoint[key] = f.get_tensor(key)
     else:
         if device is None:
-            device = "cuda" if torch.cuda.is_available() else "cpu"
+            device = get_device()
             checkpoint = torch.load(checkpoint_path, map_location=device)
         else:
             checkpoint = torch.load(checkpoint_path, map_location=device)

diffusers/pipelines/stable_diffusion/pipeline_flax_stable_diffusion.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/stable_diffusion/pipeline_flax_stable_diffusion_img2img.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/stable_diffusion/pipeline_flax_stable_diffusion_inpaint.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/stable_diffusion/pipeline_onnx_stable_diffusion.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.
@@ -294,11 +294,11 @@ class OnnxStableDiffusionPipeline(DiffusionPipeline):
                 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 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.
             negative_prompt (`str` or `List[str]`, *optional*):
                 The prompt or prompts not to guide the image generation. If not defined, one has to pass
                 `negative_prompt_embeds`. instead. Ignored when not using guidance (i.e., ignored if `guidance_scale`
@@ -306,8 +306,8 @@ class OnnxStableDiffusionPipeline(DiffusionPipeline):
             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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
-                [`schedulers.DDIMScheduler`], will be ignored for others.
+                Corresponds to parameter eta (η) in the DDIM paper: https://huggingface.co/papers/2010.02502. Only
+                applies to [`schedulers.DDIMScheduler`], will be ignored for others.
             generator (`np.random.RandomState`, *optional*):
                 One or a list of [numpy generator(s)](TODO) to make generation deterministic.
             latents (`np.ndarray`, *optional*):
@@ -359,7 +359,7 @@ class OnnxStableDiffusionPipeline(DiffusionPipeline):
             generator = np.random
 
         # 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
 
@@ -383,11 +383,12 @@ class OnnxStableDiffusionPipeline(DiffusionPipeline):
         # set timesteps
         self.scheduler.set_timesteps(num_inference_steps)
 
-        latents = latents * np.float64(self.scheduler.init_noise_sigma)
+        # scale the initial noise by the standard deviation required by the scheduler
+        latents = latents * self.scheduler.init_noise_sigma
 
         # 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())
         extra_step_kwargs = {}

diffusers/pipelines/stable_diffusion/pipeline_onnx_stable_diffusion_img2img.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.
@@ -348,19 +348,19 @@ class OnnxStableDiffusionImg2ImgPipeline(DiffusionPipeline):
                 The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                 expense of slower inference. This parameter will be modulated by `strength`.
             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.
             negative_prompt (`str` or `List[str]`, *optional*):
                 The prompt or prompts not to guide the image generation. Ignored when not using guidance (i.e., ignored
                 if `guidance_scale` is less than `1`).
             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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
-                [`schedulers.DDIMScheduler`], will be ignored for others.
+                Corresponds to parameter eta (η) in the DDIM paper: https://huggingface.co/papers/2010.02502. Only
+                applies to [`schedulers.DDIMScheduler`], will be ignored for others.
             generator (`np.random.RandomState`, *optional*):
                 A np.random.RandomState to make generation deterministic.
             prompt_embeds (`np.ndarray`, *optional*):
@@ -414,7 +414,7 @@ class OnnxStableDiffusionImg2ImgPipeline(DiffusionPipeline):
         image = preprocess(image).cpu().numpy()
 
         # 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
 
@@ -470,7 +470,7 @@ class OnnxStableDiffusionImg2ImgPipeline(DiffusionPipeline):
 
         # 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())
         extra_step_kwargs = {}

diffusers/pipelines/stable_diffusion/pipeline_onnx_stable_diffusion_inpaint.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.
@@ -360,19 +360,19 @@ class OnnxStableDiffusionInpaintPipeline(DiffusionPipeline):
                 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 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.
             negative_prompt (`str` or `List[str]`, *optional*):
                 The prompt or prompts not to guide the image generation. Ignored when not using guidance (i.e., ignored
                 if `guidance_scale` is less than `1`).
             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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
-                [`schedulers.DDIMScheduler`], will be ignored for others.
+                Corresponds to parameter eta (η) in the DDIM paper: https://huggingface.co/papers/2010.02502. Only
+                applies to [`schedulers.DDIMScheduler`], will be ignored for others.
             generator (`np.random.RandomState`, *optional*):
                 A np.random.RandomState to make generation deterministic.
             latents (`np.ndarray`, *optional*):
@@ -427,7 +427,7 @@ class OnnxStableDiffusionInpaintPipeline(DiffusionPipeline):
         self.scheduler.set_timesteps(num_inference_steps)
 
         # 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
 
@@ -483,11 +483,11 @@ class OnnxStableDiffusionInpaintPipeline(DiffusionPipeline):
         self.scheduler.set_timesteps(num_inference_steps)
 
         # scale the initial noise by the standard deviation required by the scheduler
-        latents = latents * np.float64(self.scheduler.init_noise_sigma)
+        latents = latents * self.scheduler.init_noise_sigma
 
         # 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())
         extra_step_kwargs = {}

diffusers/pipelines/stable_diffusion/pipeline_onnx_stable_diffusion_upscale.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.
@@ -378,11 +378,11 @@ class OnnxStableDiffusionUpscalePipeline(DiffusionPipeline):
                 The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                 expense of slower inference. This parameter will be modulated by `strength`.
             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.
             noise_level (`float`, defaults to 0.2):
                 Deteremines the amount of noise to add to the initial image before performing upscaling.
             negative_prompt (`str` or `List[str]`, *optional*):
@@ -391,8 +391,8 @@ class OnnxStableDiffusionUpscalePipeline(DiffusionPipeline):
             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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
-                [`schedulers.DDIMScheduler`], will be ignored for others.
+                Corresponds to parameter eta (η) in the DDIM paper: https://huggingface.co/papers/2010.02502. Only
+                applies to [`schedulers.DDIMScheduler`], will be ignored for others.
             generator (`np.random.RandomState`, *optional*):
                 A np.random.RandomState to make generation deterministic.
             latents (`torch.Tensor`, *optional*):
@@ -450,7 +450,7 @@ class OnnxStableDiffusionUpscalePipeline(DiffusionPipeline):
             generator = np.random
 
         # 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
 
@@ -481,7 +481,7 @@ class OnnxStableDiffusionUpscalePipeline(DiffusionPipeline):
         timesteps = self.scheduler.timesteps
 
         # Scale the initial noise by the standard deviation required by the scheduler
-        latents = latents * np.float64(self.scheduler.init_noise_sigma)
+        latents = latents * self.scheduler.init_noise_sigma
 
         # 5. Add noise to image
         noise_level = np.array([noise_level]).astype(np.int64)

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.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.
@@ -70,7 +70,7 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
     r"""
     Rescales `noise_cfg` tensor based on `guidance_rescale` to improve image quality and fix overexposure. Based on
     Section 3.4 from [Common Diffusion Noise Schedules and Sample Steps are
-    Flawed](https://arxiv.org/pdf/2305.08891.pdf).
+    Flawed](https://huggingface.co/papers/2305.08891).
 
     Args:
         noise_cfg (`torch.Tensor`):
@@ -608,7 +608,7 @@ class StableDiffusionPipeline(
     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())
@@ -757,7 +757,7 @@ class StableDiffusionPipeline(
         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):
@@ -836,8 +836,8 @@ class StableDiffusionPipeline(
             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.
@@ -867,7 +867,7 @@ class StableDiffusionPipeline(
                 [`self.processor`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
             guidance_rescale (`float`, *optional*, defaults to 0.0):
                 Guidance rescale factor from [Common Diffusion Noise Schedules and Sample Steps are
-                Flawed](https://arxiv.org/pdf/2305.08891.pdf). Guidance rescale factor should fix overexposure when
+                Flawed](https://huggingface.co/papers/2305.08891). Guidance rescale factor should fix overexposure when
                 using zero terminal SNR.
             clip_skip (`int`, *optional*):
                 Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that
@@ -1034,7 +1034,8 @@ class StableDiffusionPipeline(
 
                 # expand the latents if we are doing classifier free guidance
                 latent_model_input = torch.cat([latents] * 2) if self.do_classifier_free_guidance else latents
-                latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)
+                if hasattr(self.scheduler, "scale_model_input"):
+                    latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)
 
                 # predict the noise residual
                 noise_pred = self.unet(
@@ -1053,7 +1054,7 @@ class StableDiffusionPipeline(
                     noise_pred = noise_pred_uncond + self.guidance_scale * (noise_pred_text - noise_pred_uncond)
 
                 if self.do_classifier_free_guidance and self.guidance_rescale > 0.0:
-                    # Based on 3.4. in https://arxiv.org/pdf/2305.08891.pdf
+                    # Based on 3.4. in https://huggingface.co/papers/2305.08891
                     noise_pred = rescale_noise_cfg(noise_pred, noise_pred_text, guidance_rescale=self.guidance_rescale)
 
                 # compute the previous noisy sample x_t -> x_t-1

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_depth2img.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.
@@ -414,7 +414,7 @@ class StableDiffusionDepth2ImgPipeline(DiffusionPipeline, TextualInversionLoader
     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())
@@ -617,7 +617,7 @@ class StableDiffusionDepth2ImgPipeline(DiffusionPipeline, TextualInversionLoader
         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):
@@ -684,8 +684,8 @@ class StableDiffusionDepth2ImgPipeline(DiffusionPipeline, TextualInversionLoader
             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.