Diffusers documentation

Pipelines

You are viewing main version, which requires installation from source. If you'd like regular pip install, checkout the latest stable version (v0.27.1).
Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

to get started

Pipelines

The DiffusionPipeline is the quickest way to load any pretrained diffusion pipeline from the Hub for inference.

You shouldn’t use the DiffusionPipeline class for training or finetuning a diffusion model. Individual components (for example, UNet2DModel and UNet2DConditionModel) of diffusion pipelines are usually trained individually, so we suggest directly working with them instead.

The pipeline type (for example StableDiffusionPipeline) of any diffusion pipeline loaded with from_pretrained() is automatically detected and pipeline components are loaded and passed to the __init__ function of the pipeline.

Any pipeline object can be saved locally with save_pretrained().

DiffusionPipeline

class diffusers.DiffusionPipeline

< >

( )

Base class for all pipelines.

DiffusionPipeline stores all components (models, schedulers, and processors) for diffusion pipelines and provides methods for loading, downloading and saving models. It also includes methods to:

  • move all PyTorch modules to the device of your choice
  • enable/disable the progress bar for the denoising iteration

Class attributes:

  • config_name (str) — The configuration filename that stores the class and module names of all the diffusion pipeline’s components.
  • _optional_components (List[str]) — List of all optional components that don’t have to be passed to the pipeline to function (should be overridden by subclasses).

__call__

( *args **kwargs )

Call self as a function.

device

< >

( ) torch.device

Returns

torch.device

The torch device on which the pipeline is located.

to

< >

( *args **kwargs ) DiffusionPipeline

Parameters

  • dtype (torch.dtype, optional) — Returns a pipeline with the specified dtype
  • device (torch.Device, optional) — Returns a pipeline with the specified device
  • silence_dtype_warnings (str, optional, defaults to False) — Whether to omit warnings if the target dtype is not compatible with the target device.

Returns

DiffusionPipeline

The pipeline converted to specified dtype and/or dtype.

Performs Pipeline dtype and/or device conversion. A torch.dtype and torch.device are inferred from the arguments of self.to(*args, **kwargs).

If the pipeline already has the correct torch.dtype and torch.device, then it is returned as is. Otherwise, the returned pipeline is a copy of self with the desired torch.dtype and torch.device.

Here are the ways to call to:

  • to(dtype, silence_dtype_warnings=False) → DiffusionPipeline to return a pipeline with the specified dtype
  • to(device, silence_dtype_warnings=False) → DiffusionPipeline to return a pipeline with the specified device
  • to(device=None, dtype=None, silence_dtype_warnings=False) → DiffusionPipeline to return a pipeline with the specified device and dtype

components

< >

( )

The self.components property can be useful to run different pipelines with the same weights and configurations without reallocating additional memory.

Returns (dict): A dictionary containing all the modules needed to initialize the pipeline.

Examples:

>>> from diffusers import (
...     StableDiffusionPipeline,
...     StableDiffusionImg2ImgPipeline,
...     StableDiffusionInpaintPipeline,
... )

>>> text2img = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5")
>>> img2img = StableDiffusionImg2ImgPipeline(**text2img.components)
>>> inpaint = StableDiffusionInpaintPipeline(**text2img.components)

disable_attention_slicing

< >

( )

Disable sliced attention computation. If enable_attention_slicing was previously called, attention is computed in one step.

disable_xformers_memory_efficient_attention

< >

( )

Disable memory efficient attention from xFormers.

download

< >

( pretrained_model_name **kwargs ) os.PathLike

Parameters

  • pretrained_model_name (str or os.PathLike, optional) — A string, the repository id (for example CompVis/ldm-text2im-large-256) of a pretrained pipeline hosted on the Hub.
  • custom_pipeline (str, optional) — Can be either:

    • A string, the repository id (for example CompVis/ldm-text2im-large-256) of a pretrained pipeline hosted on the Hub. The repository must contain a file called pipeline.py that defines the custom pipeline.

    • A string, the file name of a community pipeline hosted on GitHub under Community. Valid file names must match the file name and not the pipeline script (clip_guided_stable_diffusion instead of clip_guided_stable_diffusion.py). Community pipelines are always loaded from the current main branch of GitHub.

    • A path to a directory (./my_pipeline_directory/) containing a custom pipeline. The directory must contain a file called pipeline.py that defines the custom pipeline.

    🧪 This is an experimental feature and may change in the future.

    For more information on how to load and create custom pipelines, take a look at How to contribute a community pipeline.

  • 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.
  • resume_download (bool, optional, defaults to False) — Whether or not to resume downloading the model weights and configuration files. If set to False, any incompletely downloaded files are deleted.
  • proxies (Dict[str, str], optional) — A dictionary of proxy servers to use by protocol or endpoint, for example, {'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}. The proxies are used on each request.
  • output_loading_info(bool, optional, defaults to False) — Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages.
  • local_files_only (bool, optional, defaults to False) — Whether to only load local model weights and configuration files or not. If set to True, the model won’t be downloaded from the Hub.
  • use_auth_token (str or bool, optional) — The token to use as HTTP bearer authorization for remote files. If True, the token generated from diffusers-cli login (stored in ~/.huggingface) is used.
  • revision (str, optional, defaults to "main") — The specific model version to use. It can be a branch name, a tag name, a commit id, or any identifier allowed by Git.
  • custom_revision (str, optional, defaults to "main") — The specific model version to use. It can be a branch name, a tag name, or a commit id similar to revision when loading a custom pipeline from the Hub. It can be a 🤗 Diffusers version when loading a custom pipeline from GitHub, otherwise it defaults to "main" when loading from the Hub.
  • mirror (str, optional) — Mirror source to resolve accessibility issues if you’re downloading a model in China. We do not guarantee the timeliness or safety of the source, and you should refer to the mirror site for more information.
  • variant (str, optional) — Load weights from a specified variant filename such as "fp16" or "ema". This is ignored when loading from_flax.
  • use_safetensors (bool, optional, defaults to None) — If set to None, the safetensors weights are downloaded if they’re available and if the safetensors library is installed. If set to True, the model is forcibly loaded from safetensors weights. If set to False, safetensors weights are not loaded.
  • use_onnx (bool, optional, defaults to False) — If set to True, ONNX weights will always be downloaded if present. If set to False, ONNX weights will never be downloaded. By default use_onnx defaults to the _is_onnx class attribute which is False for non-ONNX pipelines and True for ONNX pipelines. ONNX weights include both files ending with .onnx and .pb.

Returns

os.PathLike

A path to the downloaded pipeline.

Download and cache a PyTorch diffusion pipeline from pretrained pipeline weights.

To use private or gated models, log-in with huggingface-cli login.

enable_attention_slicing

< >

( slice_size: typing.Union[str, int, NoneType] = 'auto' )

Parameters

  • slice_size (str or int, optional, defaults to "auto") — When "auto", halves the input to the attention heads, so attention will be computed in two steps. If "max", maximum amount of memory will be saved by running only one slice at a time. If a number is provided, uses as many slices as attention_head_dim // slice_size. In this case, attention_head_dim must be a multiple of slice_size.

Enable sliced attention computation. When this option is enabled, the attention module splits the input tensor in slices to compute attention in several steps. For more than one attention head, the computation is performed sequentially over each head. This is useful to save some memory in exchange for a small speed decrease.

⚠️ Don’t enable attention slicing if you’re already using scaled_dot_product_attention (SDPA) from PyTorch 2.0 or xFormers. These attention computations are already very memory efficient so you won’t need to enable this function. If you enable attention slicing with SDPA or xFormers, it can lead to serious slow downs!

Examples:

>>> import torch
>>> from diffusers import StableDiffusionPipeline

>>> pipe = StableDiffusionPipeline.from_pretrained(
...     "runwayml/stable-diffusion-v1-5",
...     torch_dtype=torch.float16,
...     use_safetensors=True,
... )

>>> prompt = "a photo of an astronaut riding a horse on mars"
>>> pipe.enable_attention_slicing()
>>> image = pipe(prompt).images[0]

enable_model_cpu_offload

< >

( gpu_id: typing.Optional[int] = None device: typing.Union[torch.device, str] = 'cuda' )

Parameters

  • gpu_id (int, optional) — The ID of the accelerator that shall be used in inference. If not specified, it will default to 0.
  • device (torch.Device or str, optional, defaults to “cuda”) — The PyTorch device type of the accelerator that shall be used in inference. If not specified, it will default to “cuda”.

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 method is called, and the model remains in GPU until the next model runs. Memory savings are lower than with enable_sequential_cpu_offload, but performance is much better due to the iterative execution of the unet.

enable_sequential_cpu_offload

< >

( gpu_id: typing.Optional[int] = None device: typing.Union[torch.device, str] = 'cuda' )

Parameters

  • gpu_id (int, optional) — The ID of the accelerator that shall be used in inference. If not specified, it will default to 0.
  • device (torch.Device or str, optional, defaults to “cuda”) — The PyTorch device type of the accelerator that shall be used in inference. If not specified, it will default to “cuda”.

Offloads all models to CPU using 🤗 Accelerate, significantly reducing memory usage. When called, the state dicts of all torch.nn.Module components (except those in self._exclude_from_cpu_offload) are saved to CPU and then moved to torch.device('meta') and loaded to GPU only when their specific submodule has its forward method called. Offloading happens on a submodule basis. Memory savings are higher than with enable_model_cpu_offload, but performance is lower.

enable_xformers_memory_efficient_attention

< >

( attention_op: typing.Optional[typing.Callable] = None )

Parameters

  • attention_op (Callable, optional) — Override the default None operator for use as op argument to the memory_efficient_attention() function of xFormers.

Enable memory efficient attention from xFormers. When this option is enabled, you should observe lower GPU memory usage and a potential speed up during inference. Speed up during training is not guaranteed.

⚠️ When memory efficient attention and sliced attention are both enabled, memory efficient attention takes precedent.

Examples:

>>> import torch
>>> from diffusers import DiffusionPipeline
>>> from xformers.ops import MemoryEfficientAttentionFlashAttentionOp

>>> pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1", torch_dtype=torch.float16)
>>> pipe = pipe.to("cuda")
>>> pipe.enable_xformers_memory_efficient_attention(attention_op=MemoryEfficientAttentionFlashAttentionOp)
>>> # Workaround for not accepting attention shape using VAE for Flash Attention
>>> pipe.vae.enable_xformers_memory_efficient_attention(attention_op=None)

from_pretrained

< >

( pretrained_model_name_or_path: typing.Union[str, os.PathLike, NoneType] **kwargs )

Parameters

  • pretrained_model_name_or_path (str or os.PathLike, optional) — Can be either:

    • A string, the repo id (for example CompVis/ldm-text2im-large-256) of a pretrained pipeline hosted on the Hub.
    • A path to a directory (for example ./my_pipeline_directory/) containing pipeline weights saved using 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.
  • custom_pipeline (str, optional) —

    🧪 This is an experimental feature and may change in the future.

    Can be either:

    • A string, the repo id (for example hf-internal-testing/diffusers-dummy-pipeline) of a custom pipeline hosted on the Hub. The repository must contain a file called pipeline.py that defines the custom pipeline.
    • A string, the file name of a community pipeline hosted on GitHub under Community. Valid file names must match the file name and not the pipeline script (clip_guided_stable_diffusion instead of clip_guided_stable_diffusion.py). Community pipelines are always loaded from the current main branch of GitHub.
    • A path to a directory (./my_pipeline_directory/) containing a custom pipeline. The directory must contain a file called pipeline.py that defines the custom pipeline.

    For more information on how to load and create custom pipelines, please have a look at Loading and Adding Custom Pipelines

  • 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.
  • cache_dir (Union[str, os.PathLike], optional) — Path to a directory where a downloaded pretrained model configuration is cached if the standard cache is not used.
  • resume_download (bool, optional, defaults to False) — Whether or not to resume downloading the model weights and configuration files. If set to False, any incompletely downloaded files are deleted.
  • proxies (Dict[str, str], optional) — A dictionary of proxy servers to use by protocol or endpoint, for example, {'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}. The proxies are used on each request.
  • output_loading_info(bool, optional, defaults to False) — Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages.
  • local_files_only (bool, optional, defaults to False) — Whether to only load local model weights and configuration files or not. If set to True, the model won’t be downloaded from the Hub.
  • use_auth_token (str or bool, optional) — The token to use as HTTP bearer authorization for remote files. If True, the token generated from diffusers-cli login (stored in ~/.huggingface) is used.
  • revision (str, optional, defaults to "main") — The specific model version to use. It can be a branch name, a tag name, a commit id, or any identifier allowed by Git.
  • custom_revision (str, optional, defaults to "main") — The specific model version to use. It can be a branch name, a tag name, or a commit id similar to revision when loading a custom pipeline from the Hub. It can be a 🤗 Diffusers version when loading a custom pipeline from GitHub, otherwise it defaults to "main" when loading from the Hub.
  • mirror (str, optional) — Mirror source to resolve accessibility issues if you’re downloading a model in China. We do not guarantee the timeliness or safety of the source, and you should refer to the mirror site for more information.
  • device_map (str or Dict[str, Union[int, str, torch.device]], optional) — A map that specifies where each submodule should go. It doesn’t need to be defined for each parameter/buffer name; once a given module name is inside, every submodule of it will be sent to the same device.

    Set device_map="auto" to have 🤗 Accelerate automatically compute the most optimized device_map. For more information about each option see designing a device map.

  • max_memory (Dict, optional) — A dictionary device identifier for the maximum memory. Will default to the maximum memory available for each GPU and the available CPU RAM if unset.
  • offload_folder (str or os.PathLike, optional) — The path to offload weights if device_map contains the value "disk".
  • offload_state_dict (bool, optional) — If True, temporarily offloads the CPU state dict to the hard drive to avoid running out of CPU RAM if the weight of the CPU state dict + the biggest shard of the checkpoint does not fit. Defaults to True when there is some disk offload.
  • low_cpu_mem_usage (bool, optional, defaults to True if torch version >= 1.9.0 else False) — Speed up model loading only loading the pretrained weights and not initializing the weights. This also tries to not use more than 1x model size in CPU memory (including peak memory) while loading the model. Only supported for PyTorch >= 1.9.0. If you are using an older version of PyTorch, setting this argument to True will raise an error.
  • use_safetensors (bool, optional, defaults to None) — If set to None, the safetensors weights are downloaded if they’re available and if the safetensors library is installed. If set to True, the model is forcibly loaded from safetensors weights. If set to False, safetensors weights are not loaded.
  • use_onnx (bool, optional, defaults to None) — If set to True, ONNX weights will always be downloaded if present. If set to False, ONNX weights will never be downloaded. By default use_onnx defaults to the _is_onnx class attribute which is False for non-ONNX pipelines and True for ONNX pipelines. ONNX weights include both files ending with .onnx and .pb.
  • kwargs (remaining dictionary of keyword arguments, optional) — Can be used to overwrite load and saveable variables (the pipeline components of the specific pipeline class). The overwritten components are passed directly to the pipelines __init__ method. See example below for more information.
  • variant (str, optional) — Load weights from a specified variant filename such as "fp16" or "ema". This is ignored when loading from_flax.

Instantiate a PyTorch diffusion pipeline from pretrained pipeline weights.

The pipeline is set in evaluation mode (model.eval()) by default.

If you get the error message below, you need to finetune the weights for your downstream task:

Some weights of UNet2DConditionModel were not initialized from the model checkpoint at runwayml/stable-diffusion-v1-5 and are newly initialized because the shapes did not match:
- conv_in.weight: found shape torch.Size([320, 4, 3, 3]) in the checkpoint and torch.Size([320, 9, 3, 3]) in the model instantiated
You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.

To use private or gated models, log-in with huggingface-cli login.

Examples:

>>> from diffusers import DiffusionPipeline

>>> # Download pipeline from huggingface.co and cache.
>>> pipeline = DiffusionPipeline.from_pretrained("CompVis/ldm-text2im-large-256")

>>> # Download pipeline that requires an authorization token
>>> # For more information on access tokens, please refer to this section
>>> # of the documentation](https://huggingface.co/docs/hub/security-tokens)
>>> pipeline = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5")

>>> # Use a different scheduler
>>> from diffusers import LMSDiscreteScheduler

>>> scheduler = LMSDiscreteScheduler.from_config(pipeline.scheduler.config)
>>> pipeline.scheduler = scheduler

maybe_free_model_hooks

< >

( )

Function that offloads all components, removes all model hooks that were added when using enable_model_cpu_offload and then applies them again. In case the model has not been offloaded this function is a no-op. Make sure to add this function to the end of the __call__ function of your pipeline so that it functions correctly when applying enable_model_cpu_offload.

numpy_to_pil

< >

( images )

Convert a NumPy image or a batch of images to a PIL image.

save_pretrained

< >

( save_directory: typing.Union[str, os.PathLike] safe_serialization: bool = True variant: typing.Optional[str] = None push_to_hub: bool = False **kwargs )

Parameters

  • save_directory (str or os.PathLike) — Directory to save a pipeline to. Will be created if it doesn’t exist.
  • safe_serialization (bool, optional, defaults to True) — Whether to save the model using safetensors or the traditional PyTorch way with pickle.
  • variant (str, optional) — If specified, weights are saved in the format pytorch_model.<variant>.bin.
  • push_to_hub (bool, optional, defaults to False) — Whether or not to push your model to the Hugging Face model hub after saving it. You can specify the repository you want to push to with repo_id (will default to the name of save_directory in your namespace).
  • kwargs (Dict[str, Any], optional) — Additional keyword arguments passed along to the push_to_hub() method.

Save all saveable variables of the pipeline to a directory. A pipeline variable can be saved and loaded if its class implements both a save and loading method. The pipeline is easily reloaded using the from_pretrained() class method.