diffusers 0.33.0__py3-none-any.whl → 0.34.0__py3-none-any.whl

diffusers/pipelines/audioldm/pipeline_audioldm.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.
@@ -24,7 +24,7 @@ from ...models import AutoencoderKL, UNet2DConditionModel
 from ...schedulers import KarrasDiffusionSchedulers
 from ...utils import is_torch_xla_available, logging, replace_example_docstring
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import AudioPipelineOutput, DiffusionPipeline, StableDiffusionMixin
+from ..pipeline_utils import AudioPipelineOutput, DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin
 
 
 if is_torch_xla_available():
@@ -57,7 +57,7 @@ EXAMPLE_DOC_STRING = """
 """
 
 
-class AudioLDMPipeline(DiffusionPipeline, StableDiffusionMixin):
+class AudioLDMPipeline(DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin):
     r"""
     Pipeline for text-to-audio generation using AudioLDM.
 
@@ -81,6 +81,7 @@ class AudioLDMPipeline(DiffusionPipeline, StableDiffusionMixin):
             Vocoder of class `SpeechT5HifiGan`.
     """
 
+    _last_supported_version = "0.33.1"
     model_cpu_offload_seq = "text_encoder->unet->vae"
 
     def __init__(
@@ -261,7 +262,7 @@ class AudioLDMPipeline(DiffusionPipeline, StableDiffusionMixin):
     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())
@@ -397,8 +398,8 @@ class AudioLDMPipeline(DiffusionPipeline, StableDiffusionMixin):
             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.
@@ -472,7 +473,7 @@ class AudioLDMPipeline(DiffusionPipeline, StableDiffusionMixin):
 
         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/audioldm2/modeling_audioldm2.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/audioldm2/pipeline_audioldm2.py

@@ -1,4 +1,4 @@
-# Copyright 2024 CVSSP, ByteDance and The HuggingFace Team. All rights reserved.
+# Copyright 2025 CVSSP, ByteDance 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.
@@ -40,7 +40,8 @@ from ...utils import (
     logging,
     replace_example_docstring,
 )
-from ...utils.torch_utils import randn_tensor
+from ...utils.import_utils import is_transformers_version
+from ...utils.torch_utils import empty_device_cache, randn_tensor
 from ..pipeline_utils import AudioPipelineOutput, DiffusionPipeline
 from .modeling_audioldm2 import AudioLDM2ProjectionModel, AudioLDM2UNet2DConditionModel
 
@@ -266,9 +267,7 @@ class AudioLDM2Pipeline(DiffusionPipeline):
 
         if self.device.type != "cpu":
             self.to("cpu", silence_dtype_warnings=True)
-            device_mod = getattr(torch, device.type, None)
-            if hasattr(device_mod, "empty_cache") and device_mod.is_available():
-                device_mod.empty_cache()  # otherwise we don't see the memory savings (but they probably exist)
+            empty_device_cache(device.type)
 
         model_sequence = [
             self.text_encoder.text_model,
@@ -312,8 +311,19 @@ class AudioLDM2Pipeline(DiffusionPipeline):
             `inputs_embeds (`torch.Tensor` of shape `(batch_size, sequence_length, hidden_size)`):
                 The sequence of generated hidden-states.
         """
+        cache_position_kwargs = {}
+        if is_transformers_version("<", "4.52.0.dev0"):
+            cache_position_kwargs["input_ids"] = inputs_embeds
+            cache_position_kwargs["model_kwargs"] = model_kwargs
+        else:
+            cache_position_kwargs["seq_length"] = inputs_embeds.shape[0]
+            cache_position_kwargs["device"] = (
+                self.language_model.device if getattr(self, "language_model", None) is not None else self.device
+            )
+            cache_position_kwargs["model_kwargs"] = model_kwargs
         max_new_tokens = max_new_tokens if max_new_tokens is not None else self.language_model.config.max_new_tokens
-        model_kwargs = self.language_model._get_initial_cache_position(inputs_embeds, model_kwargs)
+        model_kwargs = self.language_model._get_initial_cache_position(**cache_position_kwargs)
+
         for _ in range(max_new_tokens):
             # prepare model inputs
             model_inputs = prepare_inputs_for_generation(inputs_embeds, **model_kwargs)
@@ -373,7 +383,7 @@ class AudioLDM2Pipeline(DiffusionPipeline):
                 *e.g.* prompt weighting. If not provided, negative_prompt_embeds will be computed from
                 `negative_prompt` input argument.
             generated_prompt_embeds (`torch.Tensor`, *optional*):
-                Pre-generated text embeddings from the GPT2 langauge model. Can be used to easily tweak text inputs,
+                Pre-generated text embeddings from the GPT2 language model. Can be used to easily tweak text inputs,
                  *e.g.* prompt weighting. If not provided, text embeddings will be generated from `prompt` input
                  argument.
             negative_generated_prompt_embeds (`torch.Tensor`, *optional*):
@@ -394,7 +404,7 @@ class AudioLDM2Pipeline(DiffusionPipeline):
             attention_mask (`torch.LongTensor`):
                 Attention mask to be applied to the `prompt_embeds`.
             generated_prompt_embeds (`torch.Tensor`):
-                Text embeddings generated from the GPT2 langauge model.
+                Text embeddings generated from the GPT2 language model.
 
         Example:
 
@@ -701,7 +711,7 @@ class AudioLDM2Pipeline(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())
@@ -888,8 +898,8 @@ class AudioLDM2Pipeline(DiffusionPipeline):
                 generated waveforms based on their cosine similarity with the text input in the joint text-audio
                 embedding space.
             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.
@@ -904,7 +914,7 @@ class AudioLDM2Pipeline(DiffusionPipeline):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If
                 not provided, `negative_prompt_embeds` are generated from the `negative_prompt` input argument.
             generated_prompt_embeds (`torch.Tensor`, *optional*):
-                Pre-generated text embeddings from the GPT2 langauge model. Can be used to easily tweak text inputs,
+                Pre-generated text embeddings from the GPT2 language model. Can be used to easily tweak text inputs,
                  *e.g.* prompt weighting. If not provided, text embeddings will be generated from `prompt` input
                  argument.
             negative_generated_prompt_embeds (`torch.Tensor`, *optional*):
@@ -987,7 +997,7 @@ class AudioLDM2Pipeline(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/aura_flow/pipeline_aura_flow.py

@@ -1,4 +1,4 @@
-# Copyright 2024 AuraFlow Authors and The HuggingFace Team. All rights reserved.
+# Copyright 2025 AuraFlow 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.
@@ -12,17 +12,25 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import inspect
-from typing import Callable, Dict, List, Optional, Tuple, Union
+from typing import Any, Callable, Dict, List, Optional, Tuple, Union
 
 import torch
 from transformers import T5Tokenizer, UMT5EncoderModel
 
 from ...callbacks import MultiPipelineCallbacks, PipelineCallback
 from ...image_processor import VaeImageProcessor
+from ...loaders import AuraFlowLoraLoaderMixin
 from ...models import AuraFlowTransformer2DModel, AutoencoderKL
 from ...models.attention_processor import AttnProcessor2_0, FusedAttnProcessor2_0, XFormersAttnProcessor
 from ...schedulers import FlowMatchEulerDiscreteScheduler
-from ...utils import is_torch_xla_available, logging, replace_example_docstring
+from ...utils import (
+    USE_PEFT_BACKEND,
+    is_torch_xla_available,
+    logging,
+    replace_example_docstring,
+    scale_lora_layers,
+    unscale_lora_layers,
+)
 from ...utils.torch_utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
@@ -112,7 +120,7 @@ def retrieve_timesteps(
     return timesteps, num_inference_steps
 
 
-class AuraFlowPipeline(DiffusionPipeline):
+class AuraFlowPipeline(DiffusionPipeline, AuraFlowLoraLoaderMixin):
     r"""
     Args:
         tokenizer (`T5TokenizerFast`):
@@ -233,6 +241,7 @@ class AuraFlowPipeline(DiffusionPipeline):
         prompt_attention_mask: Optional[torch.Tensor] = None,
         negative_prompt_attention_mask: Optional[torch.Tensor] = None,
         max_sequence_length: int = 256,
+        lora_scale: Optional[float] = None,
     ):
         r"""
         Encodes the prompt into text encoder hidden states.
@@ -259,10 +268,20 @@ class AuraFlowPipeline(DiffusionPipeline):
             negative_prompt_attention_mask (`torch.Tensor`, *optional*):
                 Pre-generated attention mask for negative text embeddings.
             max_sequence_length (`int`, defaults to 256): Maximum sequence length to use for the prompt.
+            lora_scale (`float`, *optional*):
+                A lora scale that will be applied to all LoRA layers of the text encoder if LoRA layers are loaded.
         """
+        # set lora scale so that monkey patched LoRA
+        # function of text encoder can correctly access it
+        if lora_scale is not None and isinstance(self, AuraFlowLoraLoaderMixin):
+            self._lora_scale = lora_scale
+
+            # dynamically adjust the LoRA scale
+            if self.text_encoder is not None and USE_PEFT_BACKEND:
+                scale_lora_layers(self.text_encoder, lora_scale)
+
         if device is None:
             device = self._execution_device
-
         if prompt is not None and isinstance(prompt, str):
             batch_size = 1
         elif prompt is not None and isinstance(prompt, list):
@@ -346,6 +365,11 @@ class AuraFlowPipeline(DiffusionPipeline):
             negative_prompt_embeds = None
             negative_prompt_attention_mask = None
 
+        if self.text_encoder is not None:
+            if isinstance(self, AuraFlowLoraLoaderMixin) and USE_PEFT_BACKEND:
+                # Retrieve the original scale by scaling back the LoRA layers
+                unscale_lora_layers(self.text_encoder, lora_scale)
+
         return prompt_embeds, prompt_attention_mask, negative_prompt_embeds, negative_prompt_attention_mask
 
     # Copied from diffusers.pipelines.stable_diffusion_3.pipeline_stable_diffusion_3.StableDiffusion3Pipeline.prepare_latents
@@ -403,6 +427,10 @@ class AuraFlowPipeline(DiffusionPipeline):
     def guidance_scale(self):
         return self._guidance_scale
 
+    @property
+    def attention_kwargs(self):
+        return self._attention_kwargs
+
     @property
     def num_timesteps(self):
         return self._num_timesteps
@@ -428,6 +456,7 @@ class AuraFlowPipeline(DiffusionPipeline):
         max_sequence_length: int = 256,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
+        attention_kwargs: Optional[Dict[str, Any]] = None,
         callback_on_step_end: Optional[
             Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
         ] = None,
@@ -455,11 +484,11 @@ class AuraFlowPipeline(DiffusionPipeline):
                 Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed,
                 `num_inference_steps` and `timesteps` must be `None`.
             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*):
@@ -486,6 +515,10 @@ class AuraFlowPipeline(DiffusionPipeline):
             return_dict (`bool`, *optional*, defaults to `True`):
                 Whether or not to return a [`~pipelines.stable_diffusion_xl.StableDiffusionXLPipelineOutput`] instead
                 of a plain tuple.
+            attention_kwargs (`dict`, *optional*):
+                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
+                `self.processor` in
+                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
             callback_on_step_end (`Callable`, *optional*):
                 A function that calls at the end of each denoising steps during the inference. The function is called
                 with the following arguments: `callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int,
@@ -520,6 +553,7 @@ class AuraFlowPipeline(DiffusionPipeline):
         )
 
         self._guidance_scale = guidance_scale
+        self._attention_kwargs = attention_kwargs
 
         # 2. Determine batch size.
         if prompt is not None and isinstance(prompt, str):
@@ -530,9 +564,10 @@ class AuraFlowPipeline(DiffusionPipeline):
             batch_size = prompt_embeds.shape[0]
 
         device = self._execution_device
+        lora_scale = self.attention_kwargs.get("scale", None) if self.attention_kwargs is not None else None
 
         # 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
 
@@ -553,6 +588,7 @@ class AuraFlowPipeline(DiffusionPipeline):
             prompt_attention_mask=prompt_attention_mask,
             negative_prompt_attention_mask=negative_prompt_attention_mask,
             max_sequence_length=max_sequence_length,
+            lora_scale=lora_scale,
         )
         if do_classifier_free_guidance:
             prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds], dim=0)
@@ -594,6 +630,7 @@ class AuraFlowPipeline(DiffusionPipeline):
                     encoder_hidden_states=prompt_embeds,
                     timestep=timestep,
                     return_dict=False,
+                    attention_kwargs=self.attention_kwargs,
                 )[0]
 
                 # perform guidance

diffusers/pipelines/auto_pipeline.py

@@ -21,6 +21,7 @@ from ..configuration_utils import ConfigMixin
 from ..models.controlnets import ControlNetUnionModel
 from ..utils import is_sentencepiece_available
 from .aura_flow import AuraFlowPipeline
+from .chroma import ChromaPipeline
 from .cogview3 import CogView3PlusPipeline
 from .cogview4 import CogView4ControlPipeline, CogView4Pipeline
 from .controlnet import (
@@ -143,6 +144,7 @@ AUTO_TEXT2IMAGE_PIPELINES_MAPPING = OrderedDict(
         ("flux-controlnet", FluxControlNetPipeline),
         ("lumina", LuminaPipeline),
         ("lumina2", Lumina2Pipeline),
+        ("chroma", ChromaPipeline),
         ("cogview3", CogView3PlusPipeline),
         ("cogview4", CogView4Pipeline),
         ("cogview4-control", CogView4ControlPipeline),
@@ -322,9 +324,8 @@ class AutoPipelineForText2Image(ConfigMixin):
                     - A path to a *directory* (for example `./my_pipeline_directory/`) containing pipeline weights
                       saved using
                     [`~DiffusionPipeline.save_pretrained`].
-            torch_dtype (`str` or `torch.dtype`, *optional*):
-                Override the default `torch.dtype` and load the model with another dtype. If "auto" is passed, the
-                dtype is automatically derived from the model's weights.
+            torch_dtype (`torch.dtype`, *optional*):
+                Override the default `torch.dtype` and load the model with another dtype.
             force_download (`bool`, *optional*, defaults to `False`):
                 Whether or not to force the (re-)download of the model weights and configuration files, overriding the
                 cached versions if they exist.
@@ -619,8 +620,7 @@ class AutoPipelineForImage2Image(ConfigMixin):
                       saved using
                     [`~DiffusionPipeline.save_pretrained`].
             torch_dtype (`str` or `torch.dtype`, *optional*):
-                Override the default `torch.dtype` and load the model with another dtype. If "auto" is passed, the
-                dtype is automatically derived from the model's weights.
+                Override the default `torch.dtype` and load the model with another dtype.
             force_download (`bool`, *optional*, defaults to `False`):
                 Whether or not to force the (re-)download of the model weights and configuration files, overriding the
                 cached versions if they exist.
@@ -930,8 +930,7 @@ class AutoPipelineForInpainting(ConfigMixin):
                       saved using
                     [`~DiffusionPipeline.save_pretrained`].
             torch_dtype (`str` or `torch.dtype`, *optional*):
-                Override the default `torch.dtype` and load the model with another dtype. If "auto" is passed, the
-                dtype is automatically derived from the model's weights.
+                Override the default `torch.dtype` and load the model with another dtype.
             force_download (`bool`, *optional*, defaults to `False`):
                 Whether or not to force the (re-)download of the model weights and configuration files, overriding the
                 cached versions if they exist.

diffusers/pipelines/blip_diffusion/modeling_blip2.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/blip_diffusion/modeling_ctx_clip.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.

diffusers/pipelines/blip_diffusion/pipeline_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.
 # You may obtain a copy of the License at
@@ -25,7 +25,7 @@ from ...utils import (
     replace_example_docstring,
 )
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, ImagePipelineOutput
 from .blip_image_processing import BlipImageProcessor
 from .modeling_blip2 import Blip2QFormerModel
 from .modeling_ctx_clip import ContextCLIPTextModel
@@ -81,7 +81,7 @@ EXAMPLE_DOC_STRING = """
 """
 
 
-class BlipDiffusionPipeline(DiffusionPipeline):
+class BlipDiffusionPipeline(DeprecatedPipelineMixin, DiffusionPipeline):
     """
     Pipeline for Zero-Shot Subject Driven Generation using Blip Diffusion.
 
@@ -107,6 +107,7 @@ class BlipDiffusionPipeline(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__(
@@ -138,7 +139,7 @@ class BlipDiffusionPipeline(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):
@@ -229,11 +230,11 @@ class BlipDiffusionPipeline(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):

diffusers/pipelines/chroma/__init__.py

@@ -0,0 +1,49 @@
+from typing import TYPE_CHECKING
+
+from ...utils import (
+    DIFFUSERS_SLOW_IMPORT,
+    OptionalDependencyNotAvailable,
+    _LazyModule,
+    get_objects_from_module,
+    is_torch_available,
+    is_transformers_available,
+)
+
+
+_dummy_objects = {}
+_additional_imports = {}
+_import_structure = {"pipeline_output": ["ChromaPipelineOutput"]}
+
+try:
+    if not (is_transformers_available() and is_torch_available()):
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    from ...utils import dummy_torch_and_transformers_objects  # noqa F403
+
+    _dummy_objects.update(get_objects_from_module(dummy_torch_and_transformers_objects))
+else:
+    _import_structure["pipeline_chroma"] = ["ChromaPipeline"]
+    _import_structure["pipeline_chroma_img2img"] = ["ChromaImg2ImgPipeline"]
+if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
+    try:
+        if not (is_transformers_available() and is_torch_available()):
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        from ...utils.dummy_torch_and_transformers_objects import *  # noqa F403
+    else:
+        from .pipeline_chroma import ChromaPipeline
+        from .pipeline_chroma_img2img import ChromaImg2ImgPipeline
+else:
+    import sys
+
+    sys.modules[__name__] = _LazyModule(
+        __name__,
+        globals()["__file__"],
+        _import_structure,
+        module_spec=__spec__,
+    )
+
+    for name, value in _dummy_objects.items():
+        setattr(sys.modules[__name__], name, value)
+    for name, value in _additional_imports.items():
+        setattr(sys.modules[__name__], name, value)