You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
-[CLIP text encoder](https://huggingface.co/docs/transformers/main/en/model_doc/clip#transformers.CLIPTextModel)
11
-
- a scheduler component, [scheduler](https://github.com/huggingface/diffusers/blob/main/src/diffusers/schedulers/scheduling_pndm.py),
11
+
- a scheduler component, [scheduler](https://github.com/huggingface/diffusers/blob/main/src/diffusers/schedulers/scheduling_pndm.py),
12
12
- a [CLIPImageProcessor](https://huggingface.co/docs/transformers/main/en/model_doc/clip#transformers.CLIPImageProcessor),
13
13
- as well as a [safety checker](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/safety_checker.py).
14
-
All of these components are necessary to run stable diffusion in inference even though they were trained
14
+
All of these components are necessary to run stable diffusion in inference even though they were trained
15
15
or created independently from each other.
16
16
17
-
To that end, we strive to offer all open-sourced, state-of-the-art diffusion system under a unified API.
17
+
To that end, we strive to offer all open-sourced, state-of-the-art diffusion system under a unified API.
18
18
More specifically, we strive to provide pipelines that
19
19
-1. can load the officially published weights and yield 1-to-1 the same outputs as the original implementation according to the corresponding paper (*e.g.*[LDMTextToImagePipeline](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines/latent_diffusion), uses the officially released weights of [High-Resolution Image Synthesis with Latent Diffusion Models](https://arxiv.org/abs/2112.10752)),
20
-
-2. have a simple user interface to run the model in inference (see the [Pipelines API](#pipelines-api) section),
20
+
-2. have a simple user interface to run the model in inference (see the [Pipelines API](#pipelines-api) section),
21
21
-3. are easy to understand with code that is self-explanatory and can be read along-side the official paper (see [Pipelines summary](#pipelines-summary)),
22
22
-4. can easily be contributed by the community (see the [Contribution](#contribution) section).
23
23
24
-
**Note** that pipelines do not (and should not) offer any training functionality.
24
+
**Note** that pipelines do not (and should not) offer any training functionality.
25
25
If you are looking for *official* training examples, please have a look at [examples](https://github.com/huggingface/diffusers/tree/main/examples).
26
26
27
27
28
28
## Pipelines Summary
29
29
30
-
The following table summarizes all officially supported pipelines, their corresponding paper, and if
30
+
The following table summarizes all officially supported pipelines, their corresponding paper, and if
31
31
available a colab notebook to directly try them out.
|[latent_diffusion](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/latent_diffusion)|[**High-Resolution Image Synthesis with Latent Diffusion Models**](https://arxiv.org/abs/2112.10752)|*Text-to-Image Generation*|
39
-
|[latent_diffusion_uncond](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/latent_diffusion_uncond)|[**High-Resolution Image Synthesis with Latent Diffusion Models**](https://arxiv.org/abs/2112.10752)|*Unconditional Image Generation*|
40
-
|[pndm](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/pndm)|[**Pseudo Numerical Methods for Diffusion Models on Manifolds**](https://arxiv.org/abs/2202.09778)|*Unconditional Image Generation*|
41
-
|[score_sde_ve](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/score_sde_ve)|[**Score-Based Generative Modeling through Stochastic Differential Equations**](https://openreview.net/forum?id=PxTIG12RRHS)|*Unconditional Image Generation*|
42
-
|[score_sde_vp](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/score_sde_vp)|[**Score-Based Generative Modeling through Stochastic Differential Equations**](https://openreview.net/forum?id=PxTIG12RRHS)|*Unconditional Image Generation*|
38
+
|[latent_diffusion](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/latent_diffusion)|[**High-Resolution Image Synthesis with Latent Diffusion Models**](https://arxiv.org/abs/2112.10752)|*Text-to-Image Generation*|
39
+
|[latent_diffusion_uncond](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/latent_diffusion_uncond)|[**High-Resolution Image Synthesis with Latent Diffusion Models**](https://arxiv.org/abs/2112.10752)|*Unconditional Image Generation*|
40
+
|[pndm](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/pndm)|[**Pseudo Numerical Methods for Diffusion Models on Manifolds**](https://arxiv.org/abs/2202.09778)|*Unconditional Image Generation*|
41
+
|[score_sde_ve](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/score_sde_ve)|[**Score-Based Generative Modeling through Stochastic Differential Equations**](https://openreview.net/forum?id=PxTIG12RRHS)|*Unconditional Image Generation*|
42
+
|[score_sde_vp](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/score_sde_vp)|[**Score-Based Generative Modeling through Stochastic Differential Equations**](https://openreview.net/forum?id=PxTIG12RRHS)|*Unconditional Image Generation*|
|[stochastic_karras_ve](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stochastic_karras_ve)|[**Elucidating the Design Space of Diffusion-Based Generative Models**](https://arxiv.org/abs/2206.00364)|*Unconditional Image Generation*|
46
+
|[stochastic_karras_ve](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stochastic_karras_ve)|[**Elucidating the Design Space of Diffusion-Based Generative Models**](https://arxiv.org/abs/2206.00364)|*Unconditional Image Generation*|
47
47
48
-
**Note**: Pipelines are simple examples of how to play around with the diffusion systems as described in the corresponding papers.
48
+
**Note**: Pipelines are simple examples of how to play around with the diffusion systems as described in the corresponding papers.
49
49
However, most of them can be adapted to use different scheduler components or even different model components. Some pipeline examples are shown in the [Examples](#examples) below.
50
50
51
51
## Pipelines API
52
52
53
-
Diffusion models often consist of multiple independently-trained models or other previously existing components.
53
+
Diffusion models often consist of multiple independently-trained models or other previously existing components.
54
54
55
55
56
-
Each model has been trained independently on a different task and the scheduler can easily be swapped out and replaced with a different one.
56
+
Each model has been trained independently on a different task and the scheduler can easily be swapped out and replaced with a different one.
57
57
During inference, we however want to be able to easily load all components and use them in inference - even if one component, *e.g.* CLIP's text encoder, originates from a different library, such as [Transformers](https://github.com/huggingface/transformers). To that end, all pipelines provide the following functionality:
58
58
59
59
-[`from_pretrained` method](https://github.com/huggingface/diffusers/blob/5cbed8e0d157f65d3ddc2420dfd09f2df630e978/src/diffusers/pipeline_utils.py#L139) that accepts a Hugging Face Hub repository id, *e.g.*[runwayml/stable-diffusion-v1-5](https://huggingface.co/runwayml/stable-diffusion-v1-5) or a path to a local directory, *e.g.*
60
-
"./stable-diffusion". To correctly retrieve which models and components should be loaded, one has to provide a `model_index.json` file, *e.g.*[runwayml/stable-diffusion-v1-5/model_index.json](https://huggingface.co/runwayml/stable-diffusion-v1-5/blob/main/model_index.json), which defines all components that should be
60
+
"./stable-diffusion". To correctly retrieve which models and components should be loaded, one has to provide a `model_index.json` file, *e.g.*[runwayml/stable-diffusion-v1-5/model_index.json](https://huggingface.co/runwayml/stable-diffusion-v1-5/blob/main/model_index.json), which defines all components that should be
61
61
loaded into the pipelines. More specifically, for each model/component one needs to define the format `<name>: ["<library>", "<class name>"]`. `<name>` is the attribute name given to the loaded instance of `<class name>` which can be found in the library or pipeline folder called `"<library>"`.
62
-
-[`save_pretrained`](https://github.com/huggingface/diffusers/blob/5cbed8e0d157f65d3ddc2420dfd09f2df630e978/src/diffusers/pipeline_utils.py#L90) that accepts a local path, *e.g.*`./stable-diffusion` under which all models/components of the pipeline will be saved. For each component/model a folder is created inside the local path that is named after the given attribute name, *e.g.*`./stable_diffusion/unet`.
63
-
In addition, a `model_index.json` file is created at the root of the local path, *e.g.*`./stable_diffusion/model_index.json` so that the complete pipeline can again be instantiated
62
+
-[`save_pretrained`](https://github.com/huggingface/diffusers/blob/5cbed8e0d157f65d3ddc2420dfd09f2df630e978/src/diffusers/pipeline_utils.py#L90) that accepts a local path, *e.g.*`./stable-diffusion` under which all models/components of the pipeline will be saved. For each component/model a folder is created inside the local path that is named after the given attribute name, *e.g.*`./stable_diffusion/unet`.
63
+
In addition, a `model_index.json` file is created at the root of the local path, *e.g.*`./stable_diffusion/model_index.json` so that the complete pipeline can again be instantiated
64
64
from the local path.
65
65
-[`to`](https://github.com/huggingface/diffusers/blob/5cbed8e0d157f65d3ddc2420dfd09f2df630e978/src/diffusers/pipeline_utils.py#L118) which accepts a `string` or `torch.device` to move all models that are of type `torch.nn.Module` to the passed device. The behavior is fully analogous to [PyTorch's `to` method](https://pytorch.org/docs/stable/generated/torch.nn.Module.html#torch.nn.Module.to).
66
-
-[`__call__`] method to use the pipeline in inference. `__call__` defines inference logic of the pipeline and should ideally encompass all aspects of it, from pre-processing to forwarding tensors to the different models and schedulers, as well as post-processing. The API of the `__call__` method can strongly vary from pipeline to pipeline. *E.g.* a text-to-image pipeline, such as [`StableDiffusionPipeline`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py) should accept among other things the text prompt to generate the image. A pure image generation pipeline, such as [DDPMPipeline](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines/ddpm) on the other hand can be run without providing any inputs. To better understand what inputs can be adapted for
66
+
-[`__call__`] method to use the pipeline in inference. `__call__` defines inference logic of the pipeline and should ideally encompass all aspects of it, from pre-processing to forwarding tensors to the different models and schedulers, as well as post-processing. The API of the `__call__` method can strongly vary from pipeline to pipeline. *E.g.* a text-to-image pipeline, such as [`StableDiffusionPipeline`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py) should accept among other things the text prompt to generate the image. A pure image generation pipeline, such as [DDPMPipeline](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines/ddpm) on the other hand can be run without providing any inputs. To better understand what inputs can be adapted for
67
67
each pipeline, one should look directly into the respective pipeline.
68
68
69
69
**Note**: All pipelines have PyTorch's autograd disabled by decorating the `__call__` method with a [`torch.no_grad`](https://pytorch.org/docs/stable/generated/torch.no_grad.html) decorator because pipelines should
70
70
not be used for training. If you want to store the gradients during the forward pass, we recommend writing your own pipeline, see also our [community-examples](https://github.com/huggingface/diffusers/tree/main/examples/community)
71
71
72
72
## Contribution
73
73
74
-
We are more than happy about any contribution to the officially supported pipelines 🤗. We aspire
74
+
We are more than happy about any contribution to the officially supported pipelines 🤗. We aspire
75
75
all of our pipelines to be **self-contained**, **easy-to-tweak**, **beginner-friendly** and for **one-purpose-only**.
76
76
77
-
-**Self-contained**: A pipeline shall be as self-contained as possible. More specifically, this means that all functionality should be either directly defined in the pipeline file itself, should be inherited from (and only from) the [`DiffusionPipeline` class](https://github.com/huggingface/diffusers/blob/5cbed8e0d157f65d3ddc2420dfd09f2df630e978/src/diffusers/pipeline_utils.py#L56) or be directly attached to the model and scheduler components of the pipeline.
78
-
-**Easy-to-use**: Pipelines should be extremely easy to use - one should be able to load the pipeline and
79
-
use it for its designated task, *e.g.* text-to-image generation, in just a couple of lines of code. Most
77
+
-**Self-contained**: A pipeline shall be as self-contained as possible. More specifically, this means that all functionality should be either directly defined in the pipeline file itself, should be inherited from (and only from) the [`DiffusionPipeline` class](https://github.com/huggingface/diffusers/blob/5cbed8e0d157f65d3ddc2420dfd09f2df630e978/src/diffusers/pipeline_utils.py#L56) or be directly attached to the model and scheduler components of the pipeline.
78
+
-**Easy-to-use**: Pipelines should be extremely easy to use - one should be able to load the pipeline and
79
+
use it for its designated task, *e.g.* text-to-image generation, in just a couple of lines of code. Most
80
80
logic including pre-processing, an unrolled diffusion loop, and post-processing should all happen inside the `__call__` method.
81
81
-**Easy-to-tweak**: Certain pipelines will not be able to handle all use cases and tasks that you might like them to. If you want to use a certain pipeline for a specific use case that is not yet supported, you might have to copy the pipeline file and tweak the code to your needs. We try to make the pipeline code as readable as possible so that each part –from pre-processing to diffusing to post-processing– can easily be adapted. If you would like the community to benefit from your customized pipeline, we would love to see a contribution to our [community-examples](https://github.com/huggingface/diffusers/tree/main/examples/community). If you feel that an important pipeline should be part of the official pipelines but isn't, a contribution to the [official pipelines](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines) would be even better.
82
82
-**One-purpose-only**: Pipelines should be used for one task and one task only. Even if two tasks are very similar from a modeling point of view, *e.g.* image2image translation and in-painting, pipelines shall be used for one task only to keep them *easy-to-tweak* and *readable*.
0 commit comments