diffusers 0.33.1__py3-none-any.whl → 0.34.0__py3-none-any.whl

diffusers/pipelines/omnigen/pipeline_omnigen.py

@@ -1,4 +1,4 @@
-# Copyright 2024 OmniGen team and The HuggingFace Team. All rights reserved.
+# Copyright 2025 OmniGen team 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.
@@ -23,12 +23,14 @@ from ...image_processor import PipelineImageInput, VaeImageProcessor
 from ...models.autoencoders import AutoencoderKL
 from ...models.transformers import OmniGenTransformer2DModel
 from ...schedulers import FlowMatchEulerDiscreteScheduler
-from ...utils import is_torch_xla_available, logging, replace_example_docstring
+from ...utils import is_torch_xla_available, is_torchvision_available, logging, replace_example_docstring
 from ...utils.torch_utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
-from .processor_omnigen import OmniGenMultiModalProcessor
 
 
+if is_torchvision_available():
+    from .processor_omnigen import OmniGenMultiModalProcessor
+
 if is_torch_xla_available():
     XLA_AVAILABLE = True
 else:
@@ -120,7 +122,7 @@ class OmniGenPipeline(
     r"""
     The OmniGen pipeline for multimodal-to-image generation.
 
-    Reference: https://arxiv.org/pdf/2409.11340
+    Reference: https://huggingface.co/papers/2409.11340
 
     Args:
         transformer ([`OmniGenTransformer2DModel`]):
@@ -176,7 +178,7 @@ class OmniGenPipeline(
         get the continue embedding of input images by VAE
 
         Args:
-            input_pixel_values: normlized pixel of input images
+            input_pixel_values: normalized pixel of input images
             device:
         Returns: torch.Tensor
         """
@@ -346,13 +348,13 @@ class OmniGenPipeline(
                 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 2.5):
-                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.
             img_guidance_scale (`float`, *optional*, defaults to 1.6):
-                Defined as equation 3 in [Instrucpix2pix](https://arxiv.org/pdf/2211.09800).
+                Defined as equation 3 in [Instrucpix2pix](https://huggingface.co/papers/2211.09800).
             use_input_image_size_as_output (bool, defaults to False):
                 whether to use the input image size as the output image size, which can be used for single-image input,
                 e.g., image editing task

diffusers/pipelines/omnigen/processor_omnigen.py

@@ -1,4 +1,4 @@
-# Copyright 2024 OmniGen team and The HuggingFace Team. All rights reserved.
+# Copyright 2025 OmniGen team 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.
@@ -18,7 +18,12 @@ from typing import Dict, List
 import numpy as np
 import torch
 from PIL import Image
-from torchvision import transforms
+
+from ...utils import is_torchvision_available
+
+
+if is_torchvision_available():
+    from torchvision import transforms
 
 
 def crop_image(pil_image, max_image_size):
@@ -198,7 +203,7 @@ class OmniGenCollator:
     def create_mask(self, attention_mask, num_tokens_for_output_images):
         """
         OmniGen applies causal attention to each element in the sequence, but applies bidirectional attention within
-        each image sequence References: [OmniGen](https://arxiv.org/pdf/2409.11340)
+        each image sequence References: [OmniGen](https://huggingface.co/papers/2409.11340)
         """
         extended_mask = []
         padding_images = []

diffusers/pipelines/onnx_utils.py

@@ -75,6 +75,11 @@ class OnnxRuntimeModel:
             logger.info("No onnxruntime provider specified, using CPUExecutionProvider")
             provider = "CPUExecutionProvider"
 
+        if provider_options is None:
+            provider_options = []
+        elif not isinstance(provider_options, list):
+            provider_options = [provider_options]
+
         return ort.InferenceSession(
             path, providers=[provider], sess_options=sess_options, provider_options=provider_options
         )
@@ -174,7 +179,10 @@ class OnnxRuntimeModel:
         # load model from local directory
         if os.path.isdir(model_id):
             model = OnnxRuntimeModel.load_model(
-                Path(model_id, model_file_name).as_posix(), provider=provider, sess_options=sess_options
+                Path(model_id, model_file_name).as_posix(),
+                provider=provider,
+                sess_options=sess_options,
+                provider_options=kwargs.pop("provider_options"),
             )
             kwargs["model_save_dir"] = Path(model_id)
         # load model from hub
@@ -190,7 +198,12 @@ class OnnxRuntimeModel:
             )
             kwargs["model_save_dir"] = Path(model_cache_path).parent
             kwargs["latest_model_name"] = Path(model_cache_path).name
-            model = OnnxRuntimeModel.load_model(model_cache_path, provider=provider, sess_options=sess_options)
+            model = OnnxRuntimeModel.load_model(
+                model_cache_path,
+                provider=provider,
+                sess_options=sess_options,
+                provider_options=kwargs.pop("provider_options"),
+            )
         return cls(model=model, **kwargs)
 
     @classmethod

diffusers/pipelines/pag/pag_utils.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.
@@ -31,7 +31,7 @@ logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 
 class PAGMixin:
-    r"""Mixin class for [Pertubed Attention Guidance](https://arxiv.org/abs/2403.17377v1)."""
+    r"""Mixin class for [Pertubed Attention Guidance](https://huggingface.co/papers/2403.17377v1)."""
 
     def _set_pag_attn_processor(self, pag_applied_layers, do_classifier_free_guidance):
         r"""

diffusers/pipelines/pag/pipeline_pag_controlnet_sd.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.
@@ -36,7 +36,7 @@ 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
@@ -541,7 +541,7 @@ class StableDiffusionControlNetPAGPipeline(
     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())
@@ -843,7 +843,7 @@ class StableDiffusionControlNetPAGPipeline(
         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):
@@ -933,8 +933,8 @@ class StableDiffusionControlNetPAGPipeline(
             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.
@@ -1228,7 +1228,11 @@ class StableDiffusionControlNetPAGPipeline(
             for i, t in enumerate(timesteps):
                 # Relevant thread:
                 # https://dev-discuss.pytorch.org/t/cudagraphs-in-pytorch-2-0/1428
-                if (is_unet_compiled and is_controlnet_compiled) and is_torch_higher_equal_2_1:
+                if (
+                    torch.cuda.is_available()
+                    and (is_unet_compiled and is_controlnet_compiled)
+                    and is_torch_higher_equal_2_1
+                ):
                     torch._inductor.cudagraph_mark_step_begin()
                 # expand the latents if we are doing classifier free guidance
                 latent_model_input = torch.cat([latents] * (prompt_embeds.shape[0] // latents.shape[0]))
@@ -1309,7 +1313,7 @@ class StableDiffusionControlNetPAGPipeline(
         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/pag/pipeline_pag_controlnet_sd_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.
@@ -37,7 +37,7 @@ from ...utils import (
     scale_lora_layers,
     unscale_lora_layers,
 )
-from ...utils.torch_utils import is_compiled_module, randn_tensor
+from ...utils.torch_utils import empty_device_cache, is_compiled_module, randn_tensor
 from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
 from ..stable_diffusion import StableDiffusionPipelineOutput
 from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker
@@ -520,7 +520,7 @@ class StableDiffusionControlNetPAGInpaintPipeline(
     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())
@@ -955,7 +955,7 @@ class StableDiffusionControlNetPAGInpaintPipeline(
         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):
@@ -1064,8 +1064,8 @@ class StableDiffusionControlNetPAGInpaintPipeline(
             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.
@@ -1521,7 +1521,7 @@ class StableDiffusionControlNetPAGInpaintPipeline(
         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/pag/pipeline_pag_controlnet_sd_xl.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.
@@ -619,7 +619,7 @@ class StableDiffusionXLControlNetPAGPipeline(
     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())
@@ -992,7 +992,7 @@ class StableDiffusionXLControlNetPAGPipeline(
         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):
@@ -1111,8 +1111,8 @@ class StableDiffusionXLControlNetPAGPipeline(
             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.
@@ -1498,7 +1498,11 @@ class StableDiffusionXLControlNetPAGPipeline(
             for i, t in enumerate(timesteps):
                 # Relevant thread:
                 # https://dev-discuss.pytorch.org/t/cudagraphs-in-pytorch-2-0/1428
-                if (is_unet_compiled and is_controlnet_compiled) and is_torch_higher_equal_2_1:
+                if (
+                    torch.cuda.is_available()
+                    and (is_unet_compiled and is_controlnet_compiled)
+                    and is_torch_higher_equal_2_1
+                ):
                     torch._inductor.cudagraph_mark_step_begin()
                 # expand the latents if we are doing classifier free guidance
                 latent_model_input = torch.cat([latents] * (prompt_embeds.shape[0] // latents.shape[0]))

diffusers/pipelines/pag/pipeline_pag_controlnet_sd_xl_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.
@@ -52,7 +52,7 @@ from ...utils import (
     scale_lora_layers,
     unscale_lora_layers,
 )
-from ...utils.torch_utils import is_compiled_module, randn_tensor
+from ...utils.torch_utils import empty_device_cache, is_compiled_module, randn_tensor
 from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
 from ..stable_diffusion_xl.pipeline_output import StableDiffusionXLPipelineOutput
 from .pag_utils import PAGMixin
@@ -115,7 +115,7 @@ EXAMPLE_DOC_STRING = """
         ...     with torch.no_grad(), torch.autocast("cuda"):
         ...         depth_map = depth_estimator(image).predicted_depth
 
-        ...     depth_map = torch.nn.fuctional.interpolate(
+        ...     depth_map = torch.nn.functional.interpolate(
         ...         depth_map.unsqueeze(1),
         ...         size=(1024, 1024),
         ...         mode="bicubic",
@@ -611,7 +611,7 @@ class StableDiffusionXLControlNetPAGImg2ImgPipeline(
     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())
@@ -926,7 +926,7 @@ class StableDiffusionXLControlNetPAGImg2ImgPipeline(
         # Offload text encoder if `enable_model_cpu_offload` was enabled
         if hasattr(self, "final_offload_hook") and self.final_offload_hook is not None:
             self.text_encoder_2.to("cpu")
-            torch.cuda.empty_cache()
+            empty_device_cache()
 
         image = image.to(device=device, dtype=dtype)
 
@@ -1074,7 +1074,7 @@ class StableDiffusionXLControlNetPAGImg2ImgPipeline(
         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):
@@ -1176,11 +1176,11 @@ class StableDiffusionXLControlNetPAGImg2ImgPipeline(
                 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` is
@@ -1191,8 +1191,8 @@ class StableDiffusionXLControlNetPAGImg2ImgPipeline(
             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 (`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.
@@ -1648,7 +1648,7 @@ class StableDiffusionXLControlNetPAGImg2ImgPipeline(
         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":
             # make sure the VAE is in float32 mode, as it overflows in float16

diffusers/pipelines/pag/pipeline_pag_hunyuandit.py

@@ -1,4 +1,4 @@
-# Copyright 2024 HunyuanDiT Authors and The HuggingFace Team. All rights reserved.
+# Copyright 2025 HunyuanDiT 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.
@@ -131,7 +131,7 @@ 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`):
@@ -443,7 +443,7 @@ class HunyuanDiTPAGPipeline(DiffusionPipeline, PAGMixin):
     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())
@@ -566,7 +566,7 @@ class HunyuanDiTPAGPipeline(DiffusionPipeline, PAGMixin):
         return self._guidance_rescale
 
     # 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):
@@ -638,8 +638,8 @@ class HunyuanDiTPAGPipeline(DiffusionPipeline, PAGMixin):
             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.
@@ -675,7 +675,7 @@ class HunyuanDiTPAGPipeline(DiffusionPipeline, PAGMixin):
                 inputs will be passed.
             guidance_rescale (`float`, *optional*, defaults to 0.0):
                 Rescale the noise_cfg according to `guidance_rescale`. Based on findings of [Common Diffusion Noise
-                Schedules and Sample Steps are Flawed](https://arxiv.org/pdf/2305.08891.pdf). See Section 3.4
+                Schedules and Sample Steps are Flawed](https://huggingface.co/papers/2305.08891). See Section 3.4
             original_size (`Tuple[int, int]`, *optional*, defaults to `(1024, 1024)`):
                 The original size of the image. Used to calculate the time ids.
             target_size (`Tuple[int, int]`, *optional*):
@@ -915,7 +915,7 @@ class HunyuanDiTPAGPipeline(DiffusionPipeline, PAGMixin):
                     noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
 
                 if self.do_classifier_free_guidance and 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=guidance_rescale)
 
                 # compute the previous noisy sample x_t -> x_t-1

diffusers/pipelines/pag/pipeline_pag_kolors.py

@@ -1,4 +1,4 @@
-# Copyright 2024 Stability AI, Kwai-Kolors Team and The HuggingFace Team. All rights reserved.
+# Copyright 2025 Stability AI, Kwai-Kolors Team 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.
@@ -453,7 +453,7 @@ class KolorsPAGPipeline(
     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())
@@ -651,7 +651,7 @@ class KolorsPAGPipeline(
         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):
@@ -749,11 +749,11 @@ class KolorsPAGPipeline(
                 "Mixture of Denoisers" multi-pipeline setup, as elaborated in [**Refining the Image
                 Output**](https://huggingface.co/docs/diffusers/api/pipelines/stable_diffusion/stable_diffusion_xl#refining-the-image-output)
             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.
             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` is
@@ -761,8 +761,8 @@ class KolorsPAGPipeline(
             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 (`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/pag/pipeline_pag_pixart_sigma.py

@@ -1,4 +1,4 @@
-# Copyright 2024 PixArt-Sigma Authors and The HuggingFace Team. All rights reserved.
+# Copyright 2025 PixArt-Sigma 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.
@@ -326,7 +326,7 @@ class PixArtSigmaPAGPipeline(DiffusionPipeline, PAGMixin):
     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())
@@ -488,7 +488,7 @@ class PixArtSigmaPAGPipeline(DiffusionPipeline, PAGMixin):
         # &amp
         caption = re.sub(r"&amp", "", caption)
 
-        # ip adresses:
+        # ip addresses:
         caption = re.sub(r"\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}", " ", caption)
 
         # article ids:
@@ -624,11 +624,11 @@ class PixArtSigmaPAGPipeline(DiffusionPipeline, PAGMixin):
                 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 4.5):
-                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.
             num_images_per_prompt (`int`, *optional*, defaults to 1):
                 The number of images to generate per prompt.
             height (`int`, *optional*, defaults to self.unet.config.sample_size):
@@ -636,8 +636,8 @@ class PixArtSigmaPAGPipeline(DiffusionPipeline, PAGMixin):
             width (`int`, *optional*, defaults to self.unet.config.sample_size):
                 The width in pixels of the generated image.
             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 (`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.
@@ -729,7 +729,7 @@ class PixArtSigmaPAGPipeline(DiffusionPipeline, PAGMixin):
         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