[Don't Merge] Check

[DreamBooth] add text encoder LoRA support in the DreamBooth training script (#3130 )
* add: LoRA text encoder support for DreamBooth example. * fix initialization. * fix: modification call. * add: entry in the readme. * use dog dataset from hub. * fix: params to clip. * add entry to the LoRA doc. * add: tests for lora. * remove unnecessary list comprehension./
2023-04-20 15:53:26 +02:00 · 2023-04-20 17:25:17 +05:30 · 2023-04-20 12:13:47 +01:00 · 2023-04-20 13:09:33 +02:00 · 2023-04-20 13:09:20 +02:00 · 2023-04-20 11:55:42 +01:00
386 changed files with 29076 additions and 4349 deletions
@@ -21,26 +21,31 @@ jobs:
      fail-fast: false
      matrix:
        config:
-          - name: Fast PyTorch CPU tests on Ubuntu
-            framework: pytorch
+          - name: Fast PyTorch Pipeline CPU tests
+            framework: pytorch_pipelines
            runner: docker-cpu
            image: diffusers/diffusers-pytorch-cpu
-            report: torch_cpu
-          - name: Fast Flax CPU tests on Ubuntu
+            report: torch_cpu_pipelines
+          - name: Fast PyTorch Models & Schedulers CPU tests
+            framework: pytorch_models
+            runner: docker-cpu
+            image: diffusers/diffusers-pytorch-cpu
+            report: torch_cpu_models_schedulers
+          - name: Fast Flax CPU tests
            framework: flax
            runner: docker-cpu
            image: diffusers/diffusers-flax-cpu
            report: flax_cpu
-          - name: Fast ONNXRuntime CPU tests on Ubuntu
+          - name: Fast ONNXRuntime CPU tests
            framework: onnxruntime
            runner: docker-cpu
            image: diffusers/diffusers-onnxruntime-cpu
            report: onnx_cpu
-          - name: PyTorch Example CPU tests on Ubuntu
+          - name: PyTorch Example CPU tests
            framework: pytorch_examples
            runner: docker-cpu
            image: diffusers/diffusers-pytorch-cpu
-            report: torch_cpu
+            report: torch_example_cpu

    name: ${{ matrix.config.name }}

@@ -71,13 +76,21 @@ jobs:
      run: |
        python utils/print_env.py

-    - name: Run fast PyTorch CPU tests
-      if: ${{ matrix.config.framework == 'pytorch' }}
+    - name: Run fast PyTorch Pipeline CPU tests
+      if: ${{ matrix.config.framework == 'pytorch_pipelines' }}
      run: |
        python -m pytest -n 2 --max-worker-restart=0 --dist=loadfile \
          -s -v -k "not Flax and not Onnx" \
          --make-reports=tests_${{ matrix.config.report }} \
-          tests/
+          tests/pipelines
+
+    - name: Run fast PyTorch Model Scheduler CPU tests
+      if: ${{ matrix.config.framework == 'pytorch_models' }}
+      run: |
+        python -m pytest -n 2 --max-worker-restart=0 --dist=loadfile \
+          -s -v -k "not Flax and not Onnx" \
+          --make-reports=tests_${{ matrix.config.report }} \
+          tests/models tests/schedulers tests/others

    - name: Run fast Flax TPU tests
      if: ${{ matrix.config.framework == 'flax' }}
@@ -85,7 +98,7 @@ jobs:
        python -m pytest -n 2 --max-worker-restart=0 --dist=loadfile \
          -s -v -k "Flax" \
          --make-reports=tests_${{ matrix.config.report }} \
-          tests/
+          tests

    - name: Run fast ONNXRuntime CPU tests
      if: ${{ matrix.config.framework == 'onnxruntime' }}
@@ -38,7 +38,7 @@ jobs:
            framework: pytorch_examples
            runner: docker-cpu
            image: diffusers/diffusers-pytorch-cpu
-            report: torch_cpu
+            report: torch_example_cpu

    name: ${{ matrix.config.name }}

@@ -24,7 +24,7 @@ community include:
 * Accepting responsibility and apologizing to those affected by our mistakes,
  and learning from the experience
 * Focusing on what is best not just for us as individuals, but for the
-  overall community
+  overall diffusers community

 Examples of unacceptable behavior include:

@@ -34,6 +34,7 @@ Examples of unacceptable behavior include:
 * Public or private harassment
 * Publishing others' private information, such as a physical or email
  address, without their explicit permission
+* Spamming issues or PRs with links to projects unrelated to this library
 * Other conduct which could reasonably be considered inappropriate in a
  professional setting

@@ -1,94 +1,350 @@
-<!---
-Copyright 2023 The HuggingFace Team. All rights reserved.
+<!--Copyright 2023 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
+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

-    http://www.apache.org/licenses/LICENSE-2.0
+http://www.apache.org/licenses/LICENSE-2.0

-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
 -->

-# How to contribute to diffusers?
+# How to contribute to Diffusers 🧨

-Everyone is welcome to contribute, and we value everybody's contribution. Code
-is thus not the only way to help the community. Answering questions, helping
-others, reaching out and improving the documentations are immensely valuable to
-the community.
+We ❤️ contributions from the open-source community! Everyone is welcome, and all types of participation –not just code– are valued and appreciated. Answering questions, helping others, reaching out, and improving the documentation are all immensely valuable to the community, so don't be afraid and get involved if you're up for it!

-It also helps us if you spread the word: reference the library from blog posts
-on the awesome projects it made possible, shout out on Twitter every time it has
-helped you, or simply star the repo to say "thank you".
+Everyone is encouraged to start by saying 👋 in our public Discord channel. We discuss the latest trends in diffusion models, ask questions, show off personal projects, help each other with contributions, or just hang out ☕. <a href="https://Discord.gg/G7tWnz98XR"><img alt="Join us on Discord" src="https://img.shields.io/Discord/823813159592001537?color=5865F2&logo=Discord&logoColor=white"></a>

-Whichever way you choose to contribute, please be mindful to respect our
-[code of conduct](https://github.com/huggingface/diffusers/blob/main/CODE_OF_CONDUCT.md).
+Whichever way you choose to contribute, we strive to be part of an open, welcoming, and kind community. Please, read our [code of conduct](https://github.com/huggingface/diffusers/blob/main/CODE_OF_CONDUCT.md) and be mindful to respect it during your interactions. We also recommend you become familiar with the [ethical guidelines](https://huggingface.co/docs/diffusers/conceptual/ethical_guidelines) that guide our project and ask you to adhere to the same principles of transparency and responsibility.

-## You can contribute in so many ways!
+We enormously value feedback from the community, so please do not be afraid to speak up if you believe you have valuable feedback that can help improve the library - every message, comment, issue, and pull request (PR) is read and considered.

-There are 4 ways you can contribute to diffusers:
-* Fixing outstanding issues with the existing code;
-* Implementing [new diffusion pipelines](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines#contribution), [new schedulers](https://github.com/huggingface/diffusers/tree/main/src/diffusers/schedulers) or [new models](https://github.com/huggingface/diffusers/tree/main/src/diffusers/models)
-* [Contributing to the examples](https://github.com/huggingface/diffusers/tree/main/examples) or to the documentation;
-* Submitting issues related to bugs or desired new features.
+## Overview

-In particular there is a special [Good First Issue](https://github.com/huggingface/diffusers/contribute) listing. 
-It will give you a list of open Issues that are open to anybody to work on. Just comment in the issue that you'd like to work on it. 
-In that same listing you will also find some Issues with `Good Second Issue` label. These are
-typically slightly more complicated than the Issues with just `Good First Issue` label. But if you
-feel you know what you're doing, go for it.
+You can contribute in many ways ranging from answering questions on issues to adding new diffusion models to
+the core library.

-*All are equally valuable to the community.*
+In the following, we give an overview of different ways to contribute, ranked by difficulty in ascending order. All of them are valuable to the community.

-## Submitting a new issue or feature request
+* 1. Asking and answering questions on [the Diffusers discussion forum](https://discuss.huggingface.co/c/discussion-related-to-httpsgithubcomhuggingfacediffusers) or on [Discord](https://discord.gg/G7tWnz98XR).
+* 2. Opening new issues on [the GitHub Issues tab](https://github.com/huggingface/diffusers/issues/new/choose)
+* 3. Answering issues on [the GitHub Issues tab](https://github.com/huggingface/diffusers/issues)
+* 4. Fix a simple issue, marked by the "Good first issue" label, see [here](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22good+first+issue%22).
+* 5. Contribute to the [documentation](https://github.com/huggingface/diffusers/tree/main/docs/source).
+* 6. Contribute a [Community Pipeline](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3Acommunity-examples)
+* 7. Contribute to the [examples](https://github.com/huggingface/diffusers/tree/main/examples).
+* 8. Fix a more difficult issue, marked by the "Good second issue" label, see [here](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22Good+second+issue%22).
+* 9. Add a new pipeline, model, or scheduler, see ["New Pipeline/Model"](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+pipeline%2Fmodel%22) and ["New scheduler"](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+scheduler%22) issues. For this contribution, please have a look at [Design Philosophy](https://github.com/huggingface/diffusers/blob/main/PHILOSOPHY.md).

-Do your best to follow these guidelines when submitting an issue or a feature
-request. It will make it easier for us to come back to you quickly and with good
-feedback.
+As said before, **all contributions are valuable to the community**.
+In the following, we will explain each contribution a bit more in detail.

-### Did you find a bug?
+For all contributions 4.-9. you will need to open a PR. It is explained in detail how to do so in [Opening a pull requst](#how-to-open-a-pr)
+
+### 1. Asking and answering questions on the Diffusers discussion forum or on the Diffusers Discord
+
+Any question or comment related to the Diffusers library can be asked on the [discussion forum](https://discuss.huggingface.co/c/discussion-related-to-httpsgithubcomhuggingfacediffusers/) or on [Discord](https://discord.gg/G7tWnz98XR). Such questions and comments include (but are not limited to):
+- Reports of training or inference experiments in an attempt to share knowledge
+- Presentation of personal projects
+- Questions to non-official training examples
+- Project proposals
+- General feedback
+- Paper summaries
+- Asking for help on personal projects that build on top of the Diffusers library
+- General questions
+- Ethical questions regarding diffusion models
+- ...
+
+Every question that is asked on the forum or on Discord actively encourages the community to publicly
+share knowledge and might very well help a beginner in the future that has the same question you're
+having. Please do pose any questions you might have.
+In the same spirit, you are of immense help to the community by answering such questions because this way you are publicly documenting knowledge for everybody to learn from.
+
+**Please** keep in mind that the more effort you put into asking or answering a question, the higher
+the quality of the publicly documented knowledge. In the same way, well-posed and well-answered questions create a high-quality knowledge database accessible to everybody, while badly posed questions or answers reduce the overall quality of the public knowledge database.
+In short, a high quality question or answer is *precise*, *concise*, *relevant*, *easy-to-understand*, *accesible*, and *well-formated/well-posed*. For more information, please have a look through the [How to write a good issue](#how-to-write-a-good-issue) section.
+
+**NOTE about channels**:
+[*The forum*](https://discuss.huggingface.co/c/discussion-related-to-httpsgithubcomhuggingfacediffusers/63) is much better indexed by search engines, such as Google. Posts are ranked by popularity rather than chronologically. Hence, it's easier to look up questions and answers that we posted some time ago.
+In addition, questions and answers posted in the forum can easily be linked to.
+In contrast, *Discord* has a chat-like format that invites fast back-and-forth communication.
+While it will most likely take less time for you to get an answer to your question on Discord, your
+question won't be visible anymore over time. Also, it's much harder to find information that was posted a while back on Discord. We therefore strongly recommend using the forum for high-quality questions and answers in an attempt to create long-lasting knowledge for the community. If discussions on Discord lead to very interesting answers and conclusions, we recommend posting the results on the forum to make the information more available for future readers.
+
+### 2. Opening new issues on the GitHub issues tab

 The 🧨 Diffusers library is robust and reliable thanks to the users who notify us of
 the problems they encounter. So thank you for reporting an issue.

-First, we would really appreciate it if you could **make sure the bug was not
-already reported** (use the search bar on Github under Issues).
+Remember, GitHub issues are reserved for technical questions directly related to the Diffusers library, bug reports, feature requests, or feedback on the library design.

-### Do you want to implement a new diffusion pipeline / diffusion model?
+In a nutshell, this means that everything that is **not** related to the **code of the Diffusers library** (including the documentation) should **not** be asked on GitHub, but rather on either the [forum](https://discuss.huggingface.co/c/discussion-related-to-httpsgithubcomhuggingfacediffusers/63) or [Discord](https://discord.gg/G7tWnz98XR).

-Awesome! Please provide the following information:
+**Please consider the following guidelines when opening a new issue**:
+- Make sure you have searched whether your issue has already been asked before (use the search bar on GitHub under Issues).
+- Please never report a new issue on another (related) issue. If another issue is highly related, please
+open a new issue nevertheless and link to the related issue.
+- Make sure your issue is written in English. Please use one of the great, free online translation services, such as [DeepL](https://www.deepl.com/translator) to translate from your native language to English if you are not comfortable in English.
+- Check whether your issue might be solved by updating to the newest Diffusers version. Before posting your issue, please make sure that `python -c "import diffusers; print(diffusers.__version__)"` is higher or matches the latest Diffusers version.
+- Remember that the more effort you put into opening a new issue, the higher the quality of your answer will be and the better the overall quality of the Diffusers issues.

-* Short description of the diffusion pipeline and link to the paper;
-* Link to the implementation if it is open-source;
-* Link to the model weights if they are available.
+New issues usually include the following.

-If you are willing to contribute the model yourself, let us know so we can best
-guide you.
+#### 2.1. Reproducible, minimal bug reports.

-### Do you want a new feature (that is not a model)?
+A bug report should always have a reproducible code snippet and be as minimal and concise as possible.
+This means in more detail:
+- Narrow the bug down as much as you can, **do not just dump your whole code file**
+- Format your code
+- Do not include any external libraries except for Diffusers depending on them.
+- **Always** provide all necessary information about your environment; for this, you can run: `diffusers-cli env` in your shell and copy-paste the displayed information to the issue.
+- Explain the issue. If the reader doesn't know what the issue is and why it is an issue, she cannot solve it.
+- **Always** make sure the reader can reproduce your issue with as little effort as possible. If your code snippet cannot be run because of missing libraries or undefined variables, the reader cannot help you. Make sure your reproducible code snippet is as minimal as possible and can be copy-pasted into a simple Python shell.
+- If in order to reproduce your issue a model and/or dataset is required, make sure the reader has access to that model or dataset. You can always upload your model or dataset to the [Hub](https://huggingface.co) to make it easily downloadable. Try to keep your model and dataset as small as possible, to make the reproduction of your issue as effortless as possible.
+
+For more information, please have a look through the [How to write a good issue](#how-to-write-a-good-issue) section.
+
+You can open a bug report [here](https://github.com/huggingface/diffusers/issues/new/choose).
+
+#### 2.2. Feature requests.

 A world-class feature request addresses the following points:

 1. Motivation first:
-  * Is it related to a problem/frustration with the library? If so, please explain
-    why. Providing a code snippet that demonstrates the problem is best.
-  * Is it related to something you would need for a project? We'd love to hear
-    about it!
-  * Is it something you worked on and think could benefit the community?
-    Awesome! Tell us what problem it solved for you.
+* Is it related to a problem/frustration with the library? If so, please explain
+why. Providing a code snippet that demonstrates the problem is best.
+* Is it related to something you would need for a project? We'd love to hear
+about it!
+* Is it something you worked on and think could benefit the community?
+Awesome! Tell us what problem it solved for you.
 2. Write a *full paragraph* describing the feature;
 3. Provide a **code snippet** that demonstrates its future use;
 4. In case this is related to a paper, please attach a link;
 5. Attach any additional information (drawings, screenshots, etc.) you think may help.

-If your issue is well written we're already 80% of the way there by the time you
-post it.
+You can open a feature request [here](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=&template=feature_request.md&title=).

-## Start contributing! (Pull Requests)
+#### 2.3 Feedback. 
+
+Feedback about the library design and why it is good or not good helps the core maintainers immensely to build a user-friendly library. To understand the philosophy behind the current design philosophy, please have a look [here](https://huggingface.co/docs/diffusers/conceptual/philosophy). If you feel like a certain design choice does not fit with the current design philosophy, please explain why and how it should be changed. If a certain design choice follows the design philosophy too much, hence restricting use cases, explain why and how it should be changed.
+If a certain design choice is very useful for you, please also leave a note as this is great feedback for future design decisions.
+
+You can open an issue about feedback [here](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=&template=feedback.md&title=).
+
+#### 2.4 Technical questions. 
+
+Technical questions are mainly about why certain code of the library was written in a certain way, or what a certain part of the code does. Please make sure to link to the code in question and please provide detail on
+why this part of the code is difficult to understand.
+
+You can open an issue about a technical question [here](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=bug&template=bug-report.yml).
+
+#### 2.5 Proposal to add a new model, scheduler, or pipeline.
+
+If the diffusion model community released a new model, pipeline, or scheduler that you would like to see in the Diffusers library, please provide the following information:
+
+* Short description of the diffusion pipeline, model, or scheduler and link to the paper or public release.
+* Link to any of its open-source implementation.
+* Link to the model weights if they are available.
+
+If you are willing to contribute to the model yourself, let us know so we can best guide you. Also, don't forget
+to tag the original author of the component (model, scheduler, pipeline, etc.) by GitHub handle if you can find it.
+
+You can open a request for a model/pipeline/scheduler [here](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=New+model%2Fpipeline%2Fscheduler&template=new-model-addition.yml).
+
+### 3. Answering issues on the GitHub issues tab
+
+Answering issues on GitHub might require some technical knowledge of Diffusers, but we encourage everybody to give it a try even if you are not 100% certain that your answer is correct.
+Some tips to give a high-quality answer to an issue:
+- Be as concise and minimal as possible
+- Stay on topic. An answer to the issue should concern the issue and only the issue.
+- Provide links to code, papers, or other sources that prove or encourage your point.
+- Answer in code. If a simple code snippet is the answer to the issue or shows how the issue can be solved, please provide a fully reproducible code snippet.
+
+Also, many issues tend to be simply off-topic, duplicates of other issues, or irrelevant. It is of great
+help to the maintainers if you can answer such issues, encouraging the author of the issue to be
+more precise, provide the link to a duplicated issue or redirect them to [the forum](https://discuss.huggingface.co/c/discussion-related-to-httpsgithubcomhuggingfacediffusers/63) or [Discord](https://discord.gg/G7tWnz98XR)
+
+If you have verified that the issued bug report is correct and requires a correction in the source code,
+please have a look at the next sections.
+
+For all of the following contributions, you will need to open a PR. It is explained in detail how to do so in the [Opening a pull requst](#how-to-open-a-pr) section.
+
+### 4. Fixing a "Good first issue"
+
+*Good first issues* are marked by the [Good first issue](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22good+first+issue%22) label. Usually, the issue already
+explains how a potential solution should look so that it is easier to fix.
+If the issue hasn't been closed and you would like to try to fix this issue, you can just leave a message "I would like to try this issue.". There are usually three scenarios:
+- a.) The issue description already proposes a fix. In this case and if the solution makes sense to you, you can open a PR or draft PR to fix it.
+- b.) The issue description does not propose a fix. In this case, you can ask what a proposed fix could look like and someone from the Diffusers team should answer shortly. If you have a good idea of how to fix it, feel free to directly open a PR.
+- c.) There is already an open PR to fix the issue, but the issue hasn't been closed yet. If the PR has gone stale, you can simply open a new PR and link to the stale PR. PRs often go stale if the original contributor who wanted to fix the issue suddenly cannot find the time anymore to proceed. This often happens in open-source and is very normal. In this case, the community will be very happy if you give it a new try and leverage the knowledge of the existing PR. If there is already a PR and it is active, you can help the author by giving suggestions, reviewing the PR or even asking whether you can contribute to the PR.
+
+
+### 5. Contribute to the documentation
+
+A good library **always** has good documentation! The official documentation is often one of the first points of contact for new users of the library, and therefore contributing to the documentation is a **highly
+valuable contribution**.
+
+Contributing to the library can have many forms:
+
+- Correcting spelling or grammatical errors.
+- Correct incorrect formatting of the docstring. If you see that the official documentation is weirdly displayed or a link is broken, we are very happy if you take some time to correct it.
+- Correct the shape or dimensions of a docstring input or output tensor.
+- Clarify documentation that is hard to understand or incorrect.
+- Update outdated code examples.
+- Translating the documentation to another language.
+
+Anything displayed on [the official Diffusers doc page](https://huggingface.co/docs/diffusers/index) is part of the official documentation and can be corrected, adjusted in the respective [documentation source](https://github.com/huggingface/diffusers/tree/main/docs/source).
+
+Please have a look at [this page](https://github.com/huggingface/diffusers/tree/main/docs) on how to verify changes made to the documentation locally.
+
+
+### 6. Contribute a community pipeline
+
+[Pipelines](https://huggingface.co/docs/diffusers/api/pipelines/overview) are usually the first point of contact between the Diffusers library and the user.
+Pipelines are examples of how to use Diffusers [models](https://huggingface.co/docs/diffusers/api/models) and [schedulers](https://huggingface.co/docs/diffusers/api/schedulers/overview).
+We support two types of pipelines:
+
+- Official Pipelines
+- Community Pipelines
+
+Both official and community pipelines follow the same design and consist of the same type of components.
+
+Official pipelines are tested and maintained by the core maintainers of Diffusers. Their code
+resides in [src/diffusers/pipelines](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines).
+In contrast, community pipelines are contributed and maintained purely by the **community** and are **not** tested.
+They reside in [examples/community](https://github.com/huggingface/diffusers/tree/main/examples/community) and while they can be accessed via the [PyPI diffusers package](https://pypi.org/project/diffusers/), their code is not part of the PyPI distribution.
+
+The reason for the distinction is that the core maintainers of the Diffusers library cannot maintain and test all
+possible ways diffusion models can be used for inference, but some of them may be of interest to the community.
+Officially released diffusion pipelines,
+such as Stable Diffusion are added to the core src/diffusers/pipelines package which ensures
+high quality of maintenance, no backward-breaking code changes, and testing.
+More bleeding edge pipelines should be added as community pipelines. If usage for a community pipeline is high, the pipeline can be moved to the official pipelines upon request from the community. This is one of the ways we strive to be a community-driven library.
+
+To add a community pipeline, one should add a <name-of-the-community>.py file to [examples/community](https://github.com/huggingface/diffusers/tree/main/examples/community) and adapt the [examples/community/README.md](https://github.com/huggingface/diffusers/tree/main/examples/community/README.md) to include an example of the new pipeline.
+
+An example can be seen [here](https://github.com/huggingface/diffusers/pull/2400).
+
+Community pipeline PRs are only checked at a superficial level and ideally they should be maintained by their original authors.
+
+Contributing a community pipeline is a great way to understand how Diffusers models and schedulers work. Having contributed a community pipeline is usually the first stepping stone to contributing an official pipeline to the
+core package.
+
+### 7. Contribute to training examples
+
+Diffusers examples are a collection of training scripts that reside in [examples](https://github.com/huggingface/diffusers/tree/main/examples).
+
+We support two types of training examples:
+
+- Official training examples
+- Research training examples
+
+Research training examples are located in [examples/research_projects](https://github.com/huggingface/diffusers/tree/main/examples/research_projects) whereas official training examples include all folders under [examples](https://github.com/huggingface/diffusers/tree/main/examples) except the `research_projects` and `community` folders.
+The official training examples are maintained by the Diffusers' core maintainers whereas the research training examples are maintained by the community.
+This is because of the same reasons put forward in [6. Contribute a community pipeline](#contribute-a-community-pipeline) for official pipelines vs. community pipelines: It is not feasible for the core maintainers to maintain all possible training methods for diffusion models.
+If the Diffusers core maintainers and the community consider a certain training paradigm to be too experimental or not popular enough, the corresponding training code should be put in the `research_projects` folder and maintained by the author.
+
+Both official training and research examples consist of a directory that contains one or more training scripts, a requirements.txt file, and a README.md file. In order for the user to make use of the
+training examples, it is required to clone the repository:
+
+```
+git clone https://github.com/huggingface/diffusers
+```
+
+as well as to install all additional dependencies required for training:
+
+```
+pip install -r /examples/<your-example-folder>/requirements.txt
+```
+
+Therefore when adding an example, the `requirements.txt` file shall define all pip dependencies required for your training example so that once all those are installed, the user can run the example's training script. See, for example, the [DreamBooth `requirements.txt` file](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/requirements.txt).
+
+Training examples of the Diffusers library should adhere to the following philosophy:
+- All the code necessary to run the examples should be found in a single Python file
+- One should be able to run the example from the command line with `python <your-example>.py --args`
+- Examples should be kept simple and serve as **an example** on how to use Diffusers for training. The purpose of example scripts is **not** to create state-of-the-art diffusion models, but rather to reproduce known training schemes without adding too much custom logic. As a byproduct of this point, our examples also strive to serve as good educational materials.
+
+To contribute an example, it is highly recommended to look at already existing examples such as [dreambooth](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth.py) to get an idea of how they should look like.
+We strongly advise contributors to make use of the [Accelerate library](https://github.com/huggingface/accelerate) as it's tightly integrated
+with Diffusers.
+Once an example script works, please make sure to add a comprehensive `README.md` that states how to use the example exactly. This README should include:
+- An example command on how to run the example script as shown [here e.g.](https://github.com/huggingface/diffusers/tree/main/examples/dreambooth#running-locally-with-pytorch).
+- A link to some training results (logs, models, ...) that show what the user can expect as shown [here e.g.](https://api.wandb.ai/report/patrickvonplaten/xm6cd5q5).
+- If you are adding a non-official/research training example, **please don't forget** to add a sentence that you are maintaining this training example which includes your git handle as shown [here](https://github.com/huggingface/diffusers/tree/main/examples/research_projects/intel_opts#diffusers-examples-with-intel-optimizations).
+
+If you are contributing to the official training examples, please also make sure to add a test to [examples/test_examples.py](https://github.com/huggingface/diffusers/blob/main/examples/test_examples.py). This is not necessary for non-official training examples.
+
+### 8. Fixing a "Good second issue"
+
+*Good second issues* are marked by the [Good second issue](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22Good+second+issue%22) label. Good second issues are
+usually more complicated to solve than [Good first issues](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22good+first+issue%22).
+The issue description usually gives less guidance on how to fix the issue and requires
+a decent understanding of the library by the interested contributor.
+If you are interested in tackling a second good issue, feel free to open a PR to fix it and link the PR to the issue. If you see that a PR has already been opened for this issue but did not get merged, have a look to understand why it wasn't merged and try to open an improved PR.
+Good second issues are usually more difficult to get merged compared to good first issues, so don't hesitate to ask for help from the core maintainers. If your PR is almost finished the core maintainers can also jump into your PR and commit to it in order to get it merged.
+
+### 9. Adding pipelines, models, schedulers
+
+Pipelines, models, and schedulers are the most important pieces of the Diffusers library.
+They provide easy access to state-of-the-art diffusion technologies and thus allow the community to
+build powerful generative AI applications.
+
+By adding a new model, pipeline, or scheduler you might enable a new powerful use case for any of the user interfaces relying on Diffusers which can be of immense value for the whole generative AI ecosystem.
+
+Diffusers has a couple of open feature requests for all three components - feel free to gloss over them
+if you don't know yet what specific component you would like to add:
+- [Model or pipeline](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+pipeline%2Fmodel%22)
+- [Scheduler](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+scheduler%22)
+
+Before adding any of the three components, it is strongly recommended that you give the [Philosophy guide](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22Good+second+issue%22) a read to better understand the design of any of the three components. Please be aware that
+we cannot merge model, scheduler, or pipeline additions that strongly diverge from our design philosophy
+as it will lead to API inconsistencies. If you fundamentally disagree with a design choice, please
+open a [Feedback issue](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=&template=feedback.md&title=) instead so that it can be discussed whether a certain design
+pattern/design choice shall be changed everywhere in the library and whether we shall update our design philosophy. Consistency across the library is very important for us.
+
+Please make sure to add links to the original codebase/paper to the PR and ideally also ping the
+original author directly on the PR so that they can follow the progress and potentially help with questions.
+
+If you are unsure or stuck in the PR, don't hesitate to leave a message to ask for a first review or help.
+
+## How to write a good issue
+
+**The better your issue is written, the higher the chances that it will be quickly resolved.**
+
+1. Make sure that you've used the correct template for your issue. You can pick between *Bug Report*, *Feature Request*, *Feedback about API Design*, *New model/pipeline/scheduler addition*, *Forum*, or a blank issue. Make sure to pick the correct one when opening [a new issue](https://github.com/huggingface/diffusers/issues/new/choose).
+2. **Be precise**: Give your issue a fitting title. Try to formulate your issue description as simple as possible. The more precise you are when submitting an issue, the less time it takes to understand the issue and potentially solve it. Make sure to open an issue for one issue only and not for multiple issues. If you found multiple issues, simply open multiple issues. If your issue is a bug, try to be as precise as possible about what bug it is - you should not just write "Error in diffusers".
+3. **Reproducibility**: No reproducible code snippet == no solution. If you encounter a bug, maintainers **have to be able to reproduce** it. Make sure that you include a code snippet that can be copy-pasted into a Python interpreter to reproduce the issue. Make sure that your code snippet works, *i.e.* that there are no missing imports or missing links to images, ... Your issue should contain an error message **and** a code snippet that can be copy-pasted without any changes to reproduce the exact same error message. If your issue is using local model weights or local data that cannot be accessed by the reader, the issue cannot be solved. If you cannot share your data or model, try to make a dummy model or dummy data.
+4. **Minimalistic**: Try to help the reader as much as you can to understand the issue as quickly as possible by staying as concise as possible. Remove all code / all information that is irrelevant to the issue. If you have found a bug, try to create the easiest code example you can to demonstrate your issue, do not just dump your whole workflow into the issue as soon as you have found a bug. E.g., if you train a model and get an error at some point during the training, you should first try to understand what part of the training code is responsible for the error and try to reproduce it with a couple of lines. Try to use dummy data instead of full datasets.
+5. Add links. If you are referring to a certain naming, method, or model make sure to provide a link so that the reader can better understand what you mean. If you are referring to a specific PR or issue, make sure to link it to your issue. Do not assume that the reader knows what you are talking about. The more links you add to your issue the better.
+6. Formatting. Make sure to nicely format your issue by formatting code into Python code syntax, and error messages into normal code syntax. See the [official GitHub formatting docs](https://docs.github.com/en/get-started/writing-on-github/getting-started-with-writing-and-formatting-on-github/basic-writing-and-formatting-syntax) for more information.
+7. Think of your issue not as a ticket to be solved, but rather as a beautiful entry to a well-written encyclopedia. Every added issue is a contribution to publicly available knowledge. By adding a nicely written issue you not only make it easier for maintainers to solve your issue, but you are helping the whole community to better understand a certain aspect of the library.
+
+## How to write a good PR
+
+1. Be a chameleon. Understand existing design patterns and syntax and make sure your code additions flow seamlessly into the existing code base. Pull requests that significantly diverge from existing design patterns or user interfaces will not be merged.
+2. Be laser focused. A pull request should solve one problem and one problem only. Make sure to not fall into the trap of "also fixing another problem while we're adding it". It is much more difficult to review pull requests that solve multiple, unrelated problems at once.
+3. If helpful, try to add a code snippet that displays an example of how your addition can be used.
+4. The title of your pull request should be a summary of its contribution.
+5. If your pull request addresses an issue, please mention the issue number in
+the pull request description to make sure they are linked (and people
+consulting the issue know you are working on it);
+6. To indicate a work in progress please prefix the title with `[WIP]`. These
+are useful to avoid duplicated work, and to differentiate it from PRs ready
+to be merged;
+7. Try to formulate and format your text as explained in [How to write a good issue](#how-to-write-a-good-issue).
+8. Make sure existing tests pass;
+9. Add high-coverage tests. No quality testing = no merge.
+- If you are adding new `@slow` tests, make sure they pass using
+`RUN_SLOW=1 python -m pytest tests/test_my_new_model.py`.
+CircleCI does not run the slow tests, but GitHub actions does every night!
+10. All public methods must have informative docstrings that work nicely with markdown. See `[pipeline_latent_diffusion.py](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/latent_diffusion/pipeline_latent_diffusion.py)` for an example.
+11. Due to the rapidly growing repository, it is important to make sure that no files that would significantly weigh down the repository are added. This includes images, videos, and other non-text files. We prefer to leverage a hf.co hosted `dataset` like
+[`hf-internal-testing`](https://huggingface.co/hf-internal-testing) or [huggingface/documentation-images](https://huggingface.co/datasets/huggingface/documentation-images) to place these files.
+If an external contribution, feel free to add the images to your PR and ask a Hugging Face member to migrate your images
+to this dataset.
+
+## How to open a PR

 Before writing code, we strongly advise you to search through the existing PRs or
 issues to make sure that nobody is already working on the same thing. If you are
@@ -99,146 +355,105 @@ You will need basic `git` proficiency to be able to contribute to
 manual. Type `git --help` in a shell and enjoy. If you prefer books, [Pro
 Git](https://git-scm.com/book/en/v2) is a very good reference.

-Follow these steps to start contributing ([supported Python versions](https://github.com/huggingface/diffusers/blob/main/setup.py#L426)):
+Follow these steps to start contributing ([supported Python versions](https://github.com/huggingface/diffusers/blob/main/setup.py#L244)):

 1. Fork the [repository](https://github.com/huggingface/diffusers) by
-   clicking on the 'Fork' button on the repository's page. This creates a copy of the code
-   under your GitHub user account.
+clicking on the 'Fork' button on the repository's page. This creates a copy of the code
+under your GitHub user account.

 2. Clone your fork to your local disk, and add the base repository as a remote:

-   ```bash
-   $ git clone git@github.com:<your Github handle>/diffusers.git
-   $ cd diffusers
-   $ git remote add upstream https://github.com/huggingface/diffusers.git
-   ```
+ ```bash
+ $ git clone git@github.com:<your Github handle>/diffusers.git
+ $ cd diffusers
+ $ git remote add upstream https://github.com/huggingface/diffusers.git
+ ```

 3. Create a new branch to hold your development changes:

-   ```bash
-   $ git checkout -b a-descriptive-name-for-my-changes
-   ```
+ ```bash
+ $ git checkout -b a-descriptive-name-for-my-changes
+ ```

-   **Do not** work on the `main` branch.
+**Do not** work on the `main` branch.

 4. Set up a development environment by running the following command in a virtual environment:

-   ```bash
-   $ pip install -e ".[dev]"
-   ```
+ ```bash
+ $ pip install -e ".[dev]"
+ ```

-   (If diffusers was already installed in the virtual environment, remove
-   it with `pip uninstall diffusers` before reinstalling it in editable
-   mode with the `-e` flag.)
-
-   To run the full test suite, you might need the additional dependency on `transformers` and `datasets` which requires a separate source
-   install:
-
-   ```bash
-   $ git clone https://github.com/huggingface/transformers
-   $ cd transformers
-   $ pip install -e .
-   ```
-
-   ```bash
-   $ git clone https://github.com/huggingface/datasets
-   $ cd datasets
-   $ pip install -e .
-   ```
-
-   If you have already cloned that repo, you might need to `git pull` to get the most recent changes in the `datasets`
-   library.
+If you have already cloned the repo, you might need to `git pull` to get the most recent changes in the
+library.

 5. Develop the features on your branch.

-   As you work on the features, you should make sure that the test suite
-   passes. You should run the tests impacted by your changes like this:
+As you work on the features, you should make sure that the test suite
+passes. You should run the tests impacted by your changes like this:

-   ```bash
-   $ pytest tests/<TEST_TO_RUN>.py
-   ```
+ ```bash
+ $ pytest tests/<TEST_TO_RUN>.py
+ ```
+ 
+Before you run the tests, please make sure you install the dependencies required for testing. You can do so 
+with this command:

-   You can also run the full suite with the following command, but it takes
-   a beefy machine to produce a result in a decent amount of time now that
-   Diffusers has grown a lot. Here is the command for it:
+ ```bash
+ $ pip install -e ".[test]"
+ ```

-   ```bash
-   $ make test
-   ```
+You can run the full test suite with the following command, but it takes
+a beefy machine to produce a result in a decent amount of time now that
+Diffusers has grown a lot. Here is the command for it:

-   For more information about tests, check out the
-   [dedicated documentation](https://huggingface.co/docs/diffusers/testing)
+ ```bash
+ $ make test
+ ```

-   🧨 Diffusers relies on `black` and `isort` to format its source code
-   consistently. After you make changes, apply automatic style corrections and code verifications
-   that can't be automated in one go with:
+🧨 Diffusers relies on `black` and `isort` to format its source code
+consistently. After you make changes, apply automatic style corrections and code verifications
+that can't be automated in one go with:

-   ```bash
-   $ make style
-   ```
+ ```bash
+ $ make style
+ ```

-   🧨 Diffusers also uses `ruff` and a few custom scripts to check for coding mistakes. Quality
-   control runs in CI, however you can also run the same checks with:
+🧨 Diffusers also uses `ruff` and a few custom scripts to check for coding mistakes. Quality
+control runs in CI, however, you can also run the same checks with:

-   ```bash
-   $ make quality
-   ```
+ ```bash
+ $ make quality
+ ```

-   Once you're happy with your changes, add changed files using `git add` and
-   make a commit with `git commit` to record your changes locally:
+Once you're happy with your changes, add changed files using `git add` and
+make a commit with `git commit` to record your changes locally:

-   ```bash
-   $ git add modified_file.py
-   $ git commit
-   ```
+ ```bash
+ $ git add modified_file.py
+ $ git commit
+ ```

-   It is a good idea to sync your copy of the code with the original
-   repository regularly. This way you can quickly account for changes:
+It is a good idea to sync your copy of the code with the original
+repository regularly. This way you can quickly account for changes:

-   ```bash
-   $ git fetch upstream
-   $ git rebase upstream/main
-   ```
+ ```bash
+ $ git pull upstream main
+ ```

-   Push the changes to your account using:
+Push the changes to your account using:

-   ```bash
-   $ git push -u origin a-descriptive-name-for-my-changes
-   ```
+ ```bash
+ $ git push -u origin a-descriptive-name-for-my-changes
+ ```

-6. Once you are satisfied (**and the checklist below is happy too**), go to the
-   webpage of your fork on GitHub. Click on 'Pull request' to send your changes
-   to the project maintainers for review.
+6. Once you are satisfied, go to the
+webpage of your fork on GitHub. Click on 'Pull request' to send your changes
+to the project maintainers for review.

 7. It's ok if maintainers ask you for changes. It happens to core contributors
-   too! So everyone can see the changes in the Pull request, work in your local
-   branch and push the changes to your fork. They will automatically appear in
-   the pull request.
-
-
-### Checklist
-
-1. The title of your pull request should be a summary of its contribution;
-2. If your pull request addresses an issue, please mention the issue number in
-   the pull request description to make sure they are linked (and people
-   consulting the issue know you are working on it);
-3. To indicate a work in progress please prefix the title with `[WIP]`. These
-   are useful to avoid duplicated work, and to differentiate it from PRs ready
-   to be merged;
-4. Make sure existing tests pass;
-5. Add high-coverage tests. No quality testing = no merge.
-   - If you are adding new `@slow` tests, make sure they pass using
-     `RUN_SLOW=1 python -m pytest tests/test_my_new_model.py`.
-   - If you are adding a new tokenizer, write tests, and make sure
-     `RUN_SLOW=1 python -m pytest tests/test_tokenization_{your_model_name}.py` passes.
-   CircleCI does not run the slow tests, but github actions does every night!
-6. All public methods must have informative docstrings that work nicely with sphinx. See `modeling_bert.py` for an
-   example.
-7. Due to the rapidly growing repository, it is important to make sure that no files that would significantly weigh down the repository are added. This includes images, videos and other non-text files. We prefer to leverage a hf.co hosted `dataset` like
-   the ones hosted on [`hf-internal-testing`](https://huggingface.co/hf-internal-testing) in which to place these files and reference 
-   them by URL. We recommend putting them in the following dataset: [huggingface/documentation-images](https://huggingface.co/datasets/huggingface/documentation-images).
-   If an external contribution, feel free to add the images to your PR and ask a Hugging Face member to migrate your images
-   to this dataset.
+too! So everyone can see the changes in the Pull request, work in your local
+branch and push the changes to your fork. They will automatically appear in
+the pull request.

 ### Tests

@@ -252,7 +467,7 @@ repository, here's how to run tests with `pytest` for the library:
 $ python -m pytest -n auto --dist=loadfile -s -v ./tests/
 ```

-In fact, that's how `make test` is implemented (sans the `pip install` line)!
+In fact, that's how `make test` is implemented!

 You can specify a smaller set of tests in order to test only the feature
 you're working on.
@@ -265,26 +480,18 @@ have enough disk space and a good Internet connection, or a lot of patience!
 $ RUN_SLOW=yes python -m pytest -n auto --dist=loadfile -s -v ./tests/
 ```

-This means `unittest` is fully supported. Here's how to run tests with
-`unittest`:
+`unittest` is fully supported, here's how to run tests with it:

 ```bash
 $ python -m unittest discover -s tests -t . -v
 $ python -m unittest discover -s examples -t examples -v
 ```

-
-### Style guide
-
-For documentation strings, 🧨 Diffusers follows the [google style](https://google.github.io/styleguide/pyguide.html).
-
-**This guide was heavily inspired by the awesome [scikit-learn guide to contributing](https://github.com/scikit-learn/scikit-learn/blob/main/CONTRIBUTING.md).**
-
 ### Syncing forked main with upstream (HuggingFace) main

 To avoid pinging the upstream repository which adds reference notes to each upstream PR and sends unnecessary notifications to the developers involved in these PRs,
 when syncing the main branch of a forked repository, please, follow these steps:
-1. When possible, avoid syncing with the upstream using a branch and PR on the forked repository. Instead merge directly into the forked main.
+1. When possible, avoid syncing with the upstream using a branch and PR on the forked repository. Instead, merge directly into the forked main.
 2. If a PR is absolutely necessary, use the following steps after checking out your branch:
 ```
 $ git checkout -b your-branch-for-syncing
@@ -292,3 +499,7 @@ $ git pull --squash --no-commit upstream main
 $ git commit -m '<your message without GitHub references>'
 $ git push --set-upstream origin your-branch-for-syncing
 ```
+
+### Style guide
+
+For documentation strings, 🧨 Diffusers follows the [google style](https://google.github.io/styleguide/pyguide.html).
@@ -0,0 +1,110 @@
+<!--Copyright 2023 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
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Philosophy
+
+🧨 Diffusers provides **state-of-the-art** pretrained diffusion models across multiple modalities.
+Its purpose is to serve as a **modular toolbox** for both inference and training.
+
+We aim at building a library that stands the test of time and therefore take API design very seriously.
+
+In a nutshell, Diffusers is built to be a natural extension of PyTorch. Therefore, most of our design choices are based on [PyTorch's Design Principles](https://pytorch.org/docs/stable/community/design.html#pytorch-design-philosophy). Let's go over the most important ones:
+
+## Usability over Performance
+
+- While Diffusers has many built-in performance-enhancing features (see [Memory and Speed](https://huggingface.co/docs/diffusers/optimization/fp16)), models are always loaded with the highest precision and lowest optimization. Therefore, by default diffusion pipelines are always instantiated on CPU with float32 precision if not otherwise defined by the user. This ensures usability across different platforms and accelerators and means that no complex installations are required to run the library.
+- Diffusers aim at being a **light-weight** package and therefore has very few required dependencies, but many soft dependencies that can improve performance (such as `accelerate`, `safetensors`, `onnx`, etc...). We strive to keep the library as lightweight as possible so that it can be added without much concern as a dependency on other packages.
+- Diffusers prefers simple, self-explainable code over condensed, magic code. This means that short-hand code syntaxes such as lambda functions, and advanced PyTorch operators are often not desired.
+
+## Simple over easy
+
+As PyTorch states, **explicit is better than implicit** and **simple is better than complex**. This design philosophy is reflected in multiple parts of the library: 
+- We follow PyTorch's API with methods like [`DiffusionPipeline.to`](https://huggingface.co/docs/diffusers/main/en/api/diffusion_pipeline#diffusers.DiffusionPipeline.to) to let the user handle device management.
+- Raising concise error messages is preferred to silently correct erroneous input. Diffusers aims at teaching the user, rather than making the library as easy to use as possible.
+- Complex model vs. scheduler logic is exposed instead of magically handled inside. Schedulers/Samplers are separated from diffusion models with minimal dependencies on each other. This forces the user to write the unrolled denoising loop. However, the separation allows for easier debugging and gives the user more control over adapting the denoising process or switching out diffusion models or schedulers.
+- Separately trained components of the diffusion pipeline, *e.g.* the text encoder, the unet, and the variational autoencoder, each have their own model class. This forces the user to handle the interaction between the different model components, and the serialization format separates the model components into different files. However, this allows for easier debugging and customization. Dreambooth or textual inversion training 
+is very simple thanks to diffusers' ability to separate single components of the diffusion pipeline.
+
+## Tweakable, contributor-friendly over abstraction
+
+For large parts of the library, Diffusers adopts an important design principle of the [Transformers library](https://github.com/huggingface/transformers), which is to prefer copy-pasted code over hasty abstractions. This design principle is very opinionated and stands in stark contrast to popular design principles such as [Don't repeat yourself (DRY)](https://en.wikipedia.org/wiki/Don%27t_repeat_yourself). 
+In short, just like Transformers does for modeling files, diffusers prefers to keep an extremely low level of abstraction and very self-contained code for pipelines and schedulers.
+Functions, long code blocks, and even classes can be copied across multiple files which at first can look like a bad, sloppy design choice that makes the library unmaintainable. 
+**However**, this design has proven to be extremely successful for Transformers and makes a lot of sense for community-driven, open-source machine learning libraries because:
+- Machine Learning is an extremely fast-moving field in which paradigms, model architectures, and algorithms are changing rapidly, which therefore makes it very difficult to define long-lasting code abstractions.
+- Machine Learning practitioners like to be able to quickly tweak existing code for ideation and research and therefore prefer self-contained code over one that contains many abstractions.
+- Open-source libraries rely on community contributions and therefore must build a library that is easy to contribute to. The more abstract the code, the more dependencies, the harder to read, and the harder to contribute to. Contributors simply stop contributing to very abstract libraries out of fear of breaking vital functionality. If contributing to a library cannot break other fundamental code, not only is it more inviting for potential new contributors, but it is also easier to review and contribute to multiple parts in parallel.
+
+At Hugging Face, we call this design the **single-file policy** which means that almost all of the code of a certain class should be written in a single, self-contained file. To read more about the philosophy, you can have a look
+at [this blog post](https://huggingface.co/blog/transformers-design-philosophy).
+
+In diffusers, we follow this philosophy for both pipelines and schedulers, but only partly for diffusion models. The reason we don't follow this design fully for diffusion models is because almost all diffusion pipelines, such 
+as [DDPM](https://huggingface.co/docs/diffusers/v0.12.0/en/api/pipelines/ddpm), [Stable Diffusion](https://huggingface.co/docs/diffusers/v0.12.0/en/api/pipelines/stable_diffusion/overview#stable-diffusion-pipelines), [UnCLIP (Dalle-2)](https://huggingface.co/docs/diffusers/v0.12.0/en/api/pipelines/unclip#overview) and [Imagen](https://imagen.research.google/) all rely on the same diffusion model, the [UNet](https://huggingface.co/docs/diffusers/api/models#diffusers.UNet2DConditionModel).
+
+Great, now you should have generally understood why 🧨 Diffusers is designed the way it is 🤗. 
+We try to apply these design principles consistently across the library. Nevertheless, there are some minor exceptions to the philosophy or some unlucky design choices. If you have feedback regarding the design, we would ❤️  to hear it [directly on GitHub](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=&template=feedback.md&title=).
+
+## Design Philosophy in Details
+
+Now, let's look a bit into the nitty-gritty details of the design philosophy. Diffusers essentially consist of three major classes, [pipelines](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines), [models](https://github.com/huggingface/diffusers/tree/main/src/diffusers/models), and [schedulers](https://github.com/huggingface/diffusers/tree/main/src/diffusers/schedulers).
+Let's walk through more in-detail design decisions for each class.
+
+### Pipelines
+
+Pipelines are designed to be easy to use (therefore do not follow [*Simple over easy*](#simple-over-easy) 100%)), are not feature complete, and should loosely be seen as examples of how to use [models](#models) and [schedulers](#schedulers) for inference.
+
+The following design principles are followed:
+- Pipelines follow the single-file policy. All pipelines can be found in individual directories under src/diffusers/pipelines. One pipeline folder corresponds to one diffusion paper/project/release. Multiple pipeline files can be gathered in one pipeline folder, as it’s done for [`src/diffusers/pipelines/stable-diffusion`](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines/stable_diffusion). If pipelines share similar functionality, one can make use of the [#Copied from mechanism](https://github.com/huggingface/diffusers/blob/125d783076e5bd9785beb05367a2d2566843a271/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py#L251).
+- Pipelines all inherit from [`DiffusionPipeline`]
+- Every pipeline consists of different model and scheduler components, that are documented in the [`model_index.json` file](https://huggingface.co/runwayml/stable-diffusion-v1-5/blob/main/model_index.json), are accessible under the same name as attributes of the pipeline and can be shared between pipelines with [`DiffusionPipeline.components`](https://huggingface.co/docs/diffusers/main/en/api/diffusion_pipeline#diffusers.DiffusionPipeline.components) function.
+- Every pipeline should be loadable via the [`DiffusionPipeline.from_pretrained`](https://huggingface.co/docs/diffusers/main/en/api/diffusion_pipeline#diffusers.DiffusionPipeline.from_pretrained) function.
+- Pipelines should be used **only** for inference.
+- Pipelines should be very readable, self-explanatory, and easy to tweak.
+- Pipelines should be designed to build on top of each other and be easy to integrate into higher-level APIs.
+- Pipelines are **not** intended to be feature-complete user interfaces. For future complete user interfaces one should rather have a look at [InvokeAI](https://github.com/invoke-ai/InvokeAI), [Diffuzers](https://github.com/abhishekkrthakur/diffuzers), and [lama-cleaner](https://github.com/Sanster/lama-cleaner)
+- Every pipeline should have one and only one way to run it via a `__call__` method. The naming of the `__call__` arguments should be shared across all pipelines.
+- Pipelines should be named after the task they are intended to solve.
+- In almost all cases, novel diffusion pipelines shall be implemented in a new pipeline folder/file.
+
+### Models
+
+Models are designed as configurable toolboxes that are natural extensions of [PyTorch's Module class](https://pytorch.org/docs/stable/generated/torch.nn.Module.html). They only partly follow the **single-file policy**.
+
+The following design principles are followed:
+- Models correspond to **a type of model architecture**. *E.g.* the [`UNet2DConditionModel`] class is used for all UNet variations that expect 2D image inputs and are conditioned on some context.
+- All models can be found in [`src/diffusers/models`](https://github.com/huggingface/diffusers/tree/main/src/diffusers/models) and every model architecture shall be defined in its file, e.g. [`unet_2d_condition.py`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/unet_2d_condition.py), [`transformer_2d.py`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/transformer_2d.py), etc...
+- Models **do not** follow the single-file policy and should make use of smaller model building blocks, such as [`attention.py`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention.py), [`resnet.py`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/resnet.py), [`embeddings.py`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/embeddings.py), etc... **Note**: This is in stark contrast to Transformers' modeling files and shows that models do not really follow the single-file policy.
+- Models intend to expose complexity, just like PyTorch's module does, and give clear error messages.
+- Models all inherit from `ModelMixin` and `ConfigMixin`.
+- Models can be optimized for performance when it doesn’t demand major code changes, keeps backward compatibility, and gives significant memory or compute gain.
+- Models should by default have the highest precision and lowest performance setting.
+- To integrate new model checkpoints whose general architecture can be classified as an architecture that already exists in Diffusers, the existing model architecture shall be adapted to make it work with the new checkpoint. One should only create a new file if the model architecture is fundamentally different.
+- Models should be designed to be easily extendable to future changes. This can be achieved by limiting public function arguments, configuration arguments, and "foreseeing" future changes, *e.g.* it is usually better to add `string` "...type" arguments that can easily be extended to new future types instead of boolean `is_..._type` arguments. Only the minimum amount of changes shall be made to existing architectures to make a new model checkpoint work.
+- The model design is a difficult trade-off between keeping code readable and concise and supporting many model checkpoints. For most parts of the modeling code, classes shall be adapted for new model checkpoints, while there are some exceptions where it is preferred to add new classes to make sure the code is kept concise and 
+readable longterm, such as [UNet blocks](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/unet_2d_blocks.py) and [Attention processors](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/cross_attention.py).
+
+### Schedulers
+
+Schedulers are responsible to guide the denoising process for inference as well as to define a noise schedule for training. They are designed as individual classes with loadable configuration files and strongly follow the **single-file policy**.
+
+The following design principles are followed:
+- All schedulers are found in [`src/diffusers/schedulers`](https://github.com/huggingface/diffusers/tree/main/src/diffusers/schedulers). 
+- Schedulers are **not** allowed to import from large utils files and shall be kept very self-contained. 
+- One scheduler python file corresponds to one scheduler algorithm (as might be defined in a paper). 
+- If schedulers share similar functionalities, we can make use of the `#Copied from` mechanism.
+- Schedulers all inherit from `SchedulerMixin` and `ConfigMixin`.
+- Schedulers can be easily swapped out with the [`ConfigMixin.from_config`](https://huggingface.co/docs/diffusers/main/en/api/configuration#diffusers.ConfigMixin.from_config) method as explained in detail [here](./using-diffusers/schedulers.mdx).
+- Every scheduler has to have a `set_num_inference_steps`, and a `step` function. `set_num_inference_steps(...)` has to be called before every denoising process, *i.e.* before `step(...)` is called.
+- Every scheduler exposes the timesteps to be "looped over" via a `timesteps` attribute, which is an array of timesteps the model will be called upon
+- The `step(...)` function takes a predicted model output and the "current" sample (x_t) and returns the "previous", slightly more denoised sample (x_t-1).
+- Given the complexity of diffusion schedulers, the `step` function does not expose all the complexity and can be a bit of a "black box".
+- In almost all cases, novel schedulers shall be implemented in a new scheduling file.
@@ -148,6 +148,18 @@ Check out the [Quickstart](https://huggingface.co/docs/diffusers/quicktour) to l
 | [versatile_diffusion](./api/pipelines/versatile_diffusion) | [Versatile Diffusion: Text, Images and Variations All in One Diffusion Model](https://arxiv.org/abs/2211.08332) | Dual Image and Text Guided Generation |
 | [vq_diffusion](./api/pipelines/vq_diffusion) | [Vector Quantized Diffusion Model for Text-to-Image Synthesis](https://arxiv.org/abs/2111.14822) | Text-to-Image Generation |

+## Contribution
+
+We ❤️  contributions from the open-source community! 
+If you want to contribute to this library, please check out our [Contribution guide](https://github.com/huggingface/diffusers/blob/main/CONTRIBUTING.md).
+You can look out for [issues](https://github.com/huggingface/diffusers/issues) you'd like to tackle to contribute to the library.
+- See [Good first issues](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22good+first+issue%22) for general opportunities to contribute
+- See [New model/pipeline](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+pipeline%2Fmodel%22) to contribute exciting new diffusion models / diffusion pipelines
+- See [New scheduler](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+scheduler%22)
+
+Also, say 👋 in our public Discord channel <a href="https://discord.gg/G7tWnz98XR"><img alt="Join us on Discord" src="https://img.shields.io/discord/823813159592001537?color=5865F2&logo=discord&logoColor=white"></a>. We discuss the hottest trends about diffusion models, help each other with contributions, personal projects or
+just hang out ☕.
+
 ## Credits

 This library concretizes previous work by many different authors and would not have been possible without their great research and implementations. We'd like to thank, in particular, the following implementations which have helped us in our development and without which the API could not have been as polished today:
@@ -4,7 +4,7 @@
  - local: quicktour
    title: Quicktour
  - local: stable_diffusion
-    title: Stable Diffusion
+    title: Effective and efficient diffusion
  - local: installation
    title: Installation
  title: Get started
@@ -25,7 +25,7 @@
    - local: using-diffusers/schedulers
      title: Load and compare different schedulers
    - local: using-diffusers/custom_pipeline_overview
-      title: Load and add custom pipelines
+      title: Load community pipelines
    - local: using-diffusers/kerascv
      title: Load KerasCV Stable Diffusion checkpoints
    title: Loading & Hub
@@ -33,25 +33,27 @@
    - local: using-diffusers/pipeline_overview
      title: Overview
    - local: using-diffusers/unconditional_image_generation
-      title: Unconditional Image Generation
+      title: Unconditional image generation
    - local: using-diffusers/conditional_image_generation
-      title: Text-to-Image Generation
+      title: Text-to-image generation
    - local: using-diffusers/img2img
-      title: Text-Guided Image-to-Image
+      title: Text-guided image-to-image
    - local: using-diffusers/inpaint
-      title: Text-Guided Image-Inpainting
+      title: Text-guided image-inpainting
    - local: using-diffusers/depth2img
-      title: Text-Guided Depth-to-Image
+      title: Text-guided depth-to-image
    - local: using-diffusers/reusing_seeds
-      title: Reusing seeds for deterministic generation
+      title: Improve image quality with deterministic generation
    - local: using-diffusers/reproducibility
-      title: Reproducibility
+      title: Create reproducible pipelines
    - local: using-diffusers/custom_pipeline_examples
-      title: Community Pipelines
+      title: Community pipelines
    - local: using-diffusers/contribute_pipeline
-      title: How to contribute a Pipeline
+      title: How to contribute a community pipeline
    - local: using-diffusers/using_safetensors
      title: Using safetensors
+    - local: using-diffusers/stable_diffusion_jax_how_to
+      title: Stable Diffusion in JAX/Flax
    - local: using-diffusers/weighted_prompts
      title: Weighting Prompts
    title: Pipelines for Inference
@@ -68,6 +70,12 @@
      title: Text-to-image
    - local: training/lora
      title: Low-Rank Adaptation of Large Language Models (LoRA)
+    - local: training/controlnet
+      title: ControlNet
+    - local: training/instructpix2pix
+      title: InstructPix2Pix Training
+    - local: training/custom_diffusion
+      title: Custom Diffusion
    title: Training
  - sections:
    - local: using-diffusers/rl
@@ -91,6 +99,8 @@
    title: ONNX
  - local: optimization/open_vino
    title: OpenVINO
+  - local: optimization/coreml
+    title: Core ML
  - local: optimization/mps
    title: MPS
  - local: optimization/habana
@@ -130,6 +140,8 @@
      title: AltDiffusion
    - local: api/pipelines/audio_diffusion
      title: Audio Diffusion
+    - local: api/pipelines/audioldm
+      title: AudioLDM
    - local: api/pipelines/cycle_diffusion
      title: Cycle Diffusion
    - local: api/pipelines/dance_diffusion
@@ -154,6 +166,8 @@
      title: Score SDE VE
    - local: api/pipelines/semantic_stable_diffusion
      title: Semantic Guidance
+    - local: api/pipelines/spectrogram_diffusion
+      title: "Spectrogram Diffusion"
    - sections:
      - local: api/pipelines/stable_diffusion/overview
        title: Overview
@@ -183,6 +197,8 @@
        title: MultiDiffusion Panorama
      - local: api/pipelines/stable_diffusion/controlnet
        title: Text-to-Image Generation with ControlNet Conditioning
+      - local: api/pipelines/stable_diffusion/model_editing
+        title: Text-to-Image Model Editing
      title: Stable Diffusion
    - local: api/pipelines/stable_diffusion_2
      title: Stable Diffusion 2
@@ -190,6 +206,10 @@
      title: Stable unCLIP
    - local: api/pipelines/stochastic_karras_ve
      title: Stochastic Karras VE
+    - local: api/pipelines/text_to_video
+      title: Text-to-Video
+    - local: api/pipelines/text_to_video_zero
+      title: Text-to-Video Zero
    - local: api/pipelines/unclip
      title: UnCLIP
    - local: api/pipelines/latent_diffusion_uncond
@@ -28,3 +28,15 @@ API to load such adapter neural networks via the [`loaders.py` module](https://g
 ### UNet2DConditionLoadersMixin

 [[autodoc]] loaders.UNet2DConditionLoadersMixin
+
+### TextualInversionLoaderMixin
+
+[[autodoc]] loaders.TextualInversionLoaderMixin
+
+### LoraLoaderMixin
+
+[[autodoc]] loaders.LoraLoaderMixin
+
+### FromCkptMixin
+
+[[autodoc]] loaders.FromCkptMixin
@@ -37,6 +37,12 @@ The models are built on the base class ['ModelMixin'] that is a `torch.nn.module
 ## UNet2DConditionModel
 [[autodoc]] UNet2DConditionModel

+## UNet3DConditionOutput
+[[autodoc]] models.unet_3d_condition.UNet3DConditionOutput
+
+## UNet3DConditionModel
+[[autodoc]] UNet3DConditionModel
+
 ## DecoderOutput
 [[autodoc]] models.vae.DecoderOutput

@@ -58,6 +64,12 @@ The models are built on the base class ['ModelMixin'] that is a `torch.nn.module
 ## Transformer2DModelOutput
 [[autodoc]] models.transformer_2d.Transformer2DModelOutput

+## TransformerTemporalModel
+[[autodoc]] models.transformer_temporal.TransformerTemporalModel
+
+## Transformer2DModelOutput
+[[autodoc]] models.transformer_temporal.TransformerTemporalModelOutput
+
 ## PriorTransformer
 [[autodoc]] models.prior_transformer.PriorTransformer

@@ -87,3 +99,9 @@ The models are built on the base class ['ModelMixin'] that is a `torch.nn.module

 ## FlaxAutoencoderKL
 [[autodoc]] FlaxAutoencoderKL
+
+## FlaxControlNetOutput
+[[autodoc]] models.controlnet_flax.FlaxControlNetOutput
+
+## FlaxControlNetModel
+[[autodoc]] FlaxControlNetModel
@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.

 # AltDiffusion

-AltDiffusion was proposed in [AltCLIP: Altering the Language Encoder in CLIP for Extended Language Capabilities](https://arxiv.org/abs/2211.06679) by Zhongzhi Chen, Guang Liu, Bo-Wen Zhang, Fulong Ye, Qinghong Yang, Ledell Wu
+AltDiffusion was proposed in [AltCLIP: Altering the Language Encoder in CLIP for Extended Language Capabilities](https://arxiv.org/abs/2211.06679) by Zhongzhi Chen, Guang Liu, Bo-Wen Zhang, Fulong Ye, Qinghong Yang, Ledell Wu.

 The abstract of the paper is the following:

@@ -28,11 +28,11 @@ The abstract of the paper is the following:

 ## Tips

- AltDiffusion is conceptually exaclty the same as [Stable Diffusion](./api/pipelines/stable_diffusion/overview).
+- AltDiffusion is conceptually exactly the same as [Stable Diffusion](./stable_diffusion/overview).

 - *Run AltDiffusion*

-AltDiffusion can be tested very easily with the [`AltDiffusionPipeline`], [`AltDiffusionImg2ImgPipeline`] and the `"BAAI/AltDiffusion-m9"` checkpoint exactly in the same way it is shown in the [Conditional Image Generation Guide](./using-diffusers/conditional_image_generation) and the [Image-to-Image Generation Guide](./using-diffusers/img2img).
+AltDiffusion can be tested very easily with the [`AltDiffusionPipeline`], [`AltDiffusionImg2ImgPipeline`] and the `"BAAI/AltDiffusion-m9"` checkpoint exactly in the same way it is shown in the [Conditional Image Generation Guide](../../using-diffusers/conditional_image_generation) and the [Image-to-Image Generation Guide](../../using-diffusers/img2img).

 - *How to load and use different schedulers.*

@@ -0,0 +1,82 @@
+<!--Copyright 2023 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
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# AudioLDM
+
+## Overview
+
+AudioLDM was proposed in [AudioLDM: Text-to-Audio Generation with Latent Diffusion Models](https://arxiv.org/abs/2301.12503) by Haohe Liu et al.
+
+Inspired by [Stable Diffusion](https://huggingface.co/docs/diffusers/api/pipelines/stable_diffusion/overview), AudioLDM
+is a text-to-audio _latent diffusion model (LDM)_ that learns continuous audio representations from [CLAP](https://huggingface.co/docs/transformers/main/model_doc/clap)
+latents. AudioLDM takes a text prompt as input and predicts the corresponding audio. It can generate text-conditional
+sound effects, human speech and music.
+
+This pipeline was contributed by [sanchit-gandhi](https://huggingface.co/sanchit-gandhi). The original codebase can be found [here](https://github.com/haoheliu/AudioLDM).
+
+## Text-to-Audio
+
+The [`AudioLDMPipeline`] can be used to load pre-trained weights from [cvssp/audioldm](https://huggingface.co/cvssp/audioldm) and generate text-conditional audio outputs:
+
+```python
+from diffusers import AudioLDMPipeline
+import torch
+import scipy
+
+repo_id = "cvssp/audioldm"
+pipe = AudioLDMPipeline.from_pretrained(repo_id, torch_dtype=torch.float16)
+pipe = pipe.to("cuda")
+
+prompt = "Techno music with a strong, upbeat tempo and high melodic riffs"
+audio = pipe(prompt, num_inference_steps=10, audio_length_in_s=5.0).audios[0]
+
+# save the audio sample as a .wav file
+scipy.io.wavfile.write("techno.wav", rate=16000, data=audio)
+```
+
+### Tips
+
+Prompts:
+* Descriptive prompt inputs work best: you can use adjectives to describe the sound (e.g. "high quality" or "clear") and make the prompt context specific (e.g., "water stream in a forest" instead of "stream").
+* It's best to use general terms like 'cat' or 'dog' instead of specific names or abstract objects that the model may not be familiar with.
+
+Inference:
+* The _quality_ of the predicted audio sample can be controlled by the `num_inference_steps` argument: higher steps give higher quality audio at the expense of slower inference.
+* The _length_ of the predicted audio sample can be controlled by varying the `audio_length_in_s` argument.
+
+### How to load and use different schedulers
+
+The AudioLDM pipeline uses [`DDIMScheduler`] scheduler by default. But `diffusers` provides many other schedulers
+that can be used with the AudioLDM pipeline such as [`PNDMScheduler`], [`LMSDiscreteScheduler`], [`EulerDiscreteScheduler`], 
+[`EulerAncestralDiscreteScheduler`] etc. We recommend using the [`DPMSolverMultistepScheduler`] as it's currently the fastest
+scheduler there is.
+
+To use a different scheduler, you can either change it via the [`ConfigMixin.from_config`]
+method, or pass the `scheduler` argument to the `from_pretrained` method of the pipeline. For example, to use the
+[`DPMSolverMultistepScheduler`], you can do the following:
+
+```python
+>>> from diffusers import AudioLDMPipeline, DPMSolverMultistepScheduler
+>>> import torch
+
+>>> pipeline = AudioLDMPipeline.from_pretrained("cvssp/audioldm", torch_dtype=torch.float16)
+>>> pipeline.scheduler = DPMSolverMultistepScheduler.from_config(pipeline.scheduler.config)
+
+>>> # or
+>>> dpm_scheduler = DPMSolverMultistepScheduler.from_pretrained("cvssp/audioldm", subfolder="scheduler")
+>>> pipeline = AudioLDMPipeline.from_pretrained("cvssp/audioldm", scheduler=dpm_scheduler, torch_dtype=torch.float16)
+```
+
+## AudioLDMPipeline
+[[autodoc]] AudioLDMPipeline
+	- all
+	- __call__
@@ -19,9 +19,9 @@ components - all of which are needed to have a functioning end-to-end diffusion
 As an example, [Stable Diffusion](https://huggingface.co/blog/stable_diffusion) has three independently trained models:
 - [Autoencoder](./api/models#vae)
 - [Conditional Unet](./api/models#UNet2DConditionModel)
- [CLIP text encoder](https://huggingface.co/docs/transformers/v4.21.2/en/model_doc/clip#transformers.CLIPTextModel)
+- [CLIP text encoder](https://huggingface.co/docs/transformers/v4.27.1/en/model_doc/clip#transformers.CLIPTextModel)
 - a scheduler component, [scheduler](./api/scheduler#pndm), 
- a [CLIPFeatureExtractor](https://huggingface.co/docs/transformers/v4.21.2/en/model_doc/clip#transformers.CLIPFeatureExtractor),
+- a [CLIPImageProcessor](https://huggingface.co/docs/transformers/v4.27.1/en/model_doc/clip#transformers.CLIPImageProcessor),
 - as well as a [safety checker](./stable_diffusion#safety_checker).
 All of these components are necessary to run stable diffusion in inference even though they were trained 
 or created independently from each other.
@@ -77,11 +77,13 @@ available a colab notebook to directly try them out.
 | [stable_unclip](./stable_unclip) | **Stable unCLIP** | Text-to-Image Generation |
 | [stable_unclip](./stable_unclip) | **Stable unCLIP** | Image-to-Image Text-Guided Generation |
 | [stochastic_karras_ve](./stochastic_karras_ve) | [**Elucidating the Design Space of Diffusion-Based Generative Models**](https://arxiv.org/abs/2206.00364) | Unconditional Image Generation |
+| [text_to_video_sd](./api/pipelines/text_to_video) | [Modelscope's Text-to-video-synthesis Model in Open Domain](https://modelscope.cn/models/damo/text-to-video-synthesis/summary) | Text-to-Video Generation |
 | [unclip](./unclip) | [Hierarchical Text-Conditional Image Generation with CLIP Latents](https://arxiv.org/abs/2204.06125) | Text-to-Image Generation |
 | [versatile_diffusion](./versatile_diffusion) | [Versatile Diffusion: Text, Images and Variations All in One Diffusion Model](https://arxiv.org/abs/2211.08332) | Text-to-Image Generation | 
 | [versatile_diffusion](./versatile_diffusion) | [Versatile Diffusion: Text, Images and Variations All in One Diffusion Model](https://arxiv.org/abs/2211.08332) | Image Variations Generation | 
 | [versatile_diffusion](./versatile_diffusion) | [Versatile Diffusion: Text, Images and Variations All in One Diffusion Model](https://arxiv.org/abs/2211.08332) | Dual Image and Text Guided Generation | 
 | [vq_diffusion](./vq_diffusion) | [Vector Quantized Diffusion Model for Text-to-Image Synthesis](https://arxiv.org/abs/2111.14822) | Text-to-Image Generation | 
+| [text_to_video_zero](./text_to_video_zero) | [Text2Video-Zero: Text-to-Image Diffusion Models are Zero-Shot Video Generators](https://arxiv.org/abs/2303.13439) | Text-to-Video Generation |


 **Note**: Pipelines are simple examples of how to play around with the diffusion systems as described in the corresponding papers. 
@@ -107,7 +109,7 @@ from the local path.
 each pipeline, one should look directly into the respective pipeline.

 **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
-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)
+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).

 ## Contribution

@@ -172,7 +174,7 @@ You can also run this example on colab [![Open In Colab](https://colab.research.

 ### Tweak prompts reusing seeds and latents

-You can generate your own latents to reproduce results, or tweak your prompt on a specific result you liked. [This notebook](https://github.com/pcuenca/diffusers-examples/blob/main/notebooks/stable-diffusion-seeds.ipynb) shows how to do it step by step. You can also run it in Google Colab [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/pcuenca/diffusers-examples/blob/main/notebooks/stable-diffusion-seeds.ipynb).
+You can generate your own latents to reproduce results, or tweak your prompt on a specific result you liked. [This notebook](https://github.com/pcuenca/diffusers-examples/blob/main/notebooks/stable-diffusion-seeds.ipynb) shows how to do it step by step. You can also run it in Google Colab [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/pcuenca/diffusers-examples/blob/main/notebooks/stable-diffusion-seeds.ipynb)


 ### In-painting using Stable Diffusion
@@ -14,7 +14,7 @@ specific language governing permissions and limitations under the License.

 ## Overview

-[Paint by Example: Exemplar-based Image Editing with Diffusion Models](https://arxiv.org/abs/2211.13227) by Binxin Yang, Shuyang Gu, Bo Zhang, Ting Zhang, Xuejin Chen, Xiaoyan Sun, Dong Chen, Fang Wen
+[Paint by Example: Exemplar-based Image Editing with Diffusion Models](https://arxiv.org/abs/2211.13227) by Binxin Yang, Shuyang Gu, Bo Zhang, Ting Zhang, Xuejin Chen, Xiaoyan Sun, Dong Chen, Fang Wen.

 The abstract of the paper is the following:

@@ -24,11 +24,11 @@ The abstract of the paper is the following:

 | Pipeline | Tasks | Colab | Demo
 |---|---|:---:|:---:|
-| [pipeline_semantic_stable_diffusion.py](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/semantic_stable_diffusion/pipeline_semantic_stable_diffusion) | *Text-to-Image Generation* |  [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/ml-research/semantic-image-editing/blob/main/examples/SemanticGuidance.ipynb) | [Coming Soon](https://huggingface.co/AIML-TUDA)
+| [pipeline_semantic_stable_diffusion.py](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/semantic_stable_diffusion/pipeline_semantic_stable_diffusion.py) | *Text-to-Image Generation* |  [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/ml-research/semantic-image-editing/blob/main/examples/SemanticGuidance.ipynb) | [Coming Soon](https://huggingface.co/AIML-TUDA)

 ## Tips

- The Semantic Guidance pipeline can be used with any [Stable Diffusion](./api/pipelines/stable_diffusion/text2img) checkpoint.
+- The Semantic Guidance pipeline can be used with any [Stable Diffusion](./stable_diffusion/text2img) checkpoint.

 ### Run Semantic Guidance

@@ -67,7 +67,7 @@ out = pipe(
 )
 ```

-For more examples check the colab notebook.
+For more examples check the Colab notebook.

 ## StableDiffusionSafePipelineOutput
 [[autodoc]] pipelines.semantic_stable_diffusion.SemanticStableDiffusionPipelineOutput
@@ -0,0 +1,54 @@
+<!--Copyright 2023 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
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Multi-instrument Music Synthesis with Spectrogram Diffusion
+
+## Overview
+
+[Spectrogram Diffusion](https://arxiv.org/abs/2206.05408) by Curtis Hawthorne, Ian Simon, Adam Roberts, Neil Zeghidour, Josh Gardner, Ethan Manilow, and Jesse Engel.
+
+An ideal music synthesizer should be both interactive and expressive, generating high-fidelity audio in realtime for arbitrary combinations of instruments and notes. Recent neural synthesizers have exhibited a tradeoff between domain-specific models that offer detailed control of only specific instruments, or raw waveform models that can train on any music but with minimal control and slow generation. In this work, we focus on a middle ground of neural synthesizers that can generate audio from MIDI sequences with arbitrary combinations of instruments in realtime. This enables training on a wide range of transcription datasets with a single model, which in turn offers note-level control of composition and instrumentation across a wide range of instruments. We use a simple two-stage process: MIDI to spectrograms with an encoder-decoder Transformer, then spectrograms to audio with a generative adversarial network (GAN) spectrogram inverter. We compare training the decoder as an autoregressive model and as a Denoising Diffusion Probabilistic Model (DDPM) and find that the DDPM approach is superior both qualitatively and as measured by audio reconstruction and Fréchet distance metrics. Given the interactivity and generality of this approach, we find this to be a promising first step towards interactive and expressive neural synthesis for arbitrary combinations of instruments and notes.
+
+The original codebase of this implementation can be found at [magenta/music-spectrogram-diffusion](https://github.com/magenta/music-spectrogram-diffusion).
+
+## Model
+
+![img](https://storage.googleapis.com/music-synthesis-with-spectrogram-diffusion/architecture.png)
+
+As depicted above the model takes as input a MIDI file and tokenizes it into a sequence of 5 second intervals. Each tokenized interval then together with positional encodings is passed through the Note Encoder and its representation is concatenated with the previous window's generated spectrogram representation obtained via the Context Encoder. For the initial 5 second window this is set to zero. The resulting context is then used as conditioning to sample the denoised Spectrogram from the MIDI window and we concatenate this spectrogram to the final output as well as use it for the context of the next MIDI window. The process repeats till we have gone over all the MIDI inputs. Finally a MelGAN decoder converts the potentially long spectrogram to audio which is the final result of this pipeline.
+
+## Available Pipelines:
+
+| Pipeline | Tasks | Colab
+|---|---|:---:|
+| [pipeline_spectrogram_diffusion.py](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/spectrogram_diffusion/pipeline_spectrogram_diffusion.py) | *Unconditional Audio Generation* | - |
+
+
+## Example usage
+
+```python
+from diffusers import SpectrogramDiffusionPipeline, MidiProcessor
+
+pipe = SpectrogramDiffusionPipeline.from_pretrained("google/music-spectrogram-diffusion")
+pipe = pipe.to("cuda")
+processor = MidiProcessor()
+
+# Download MIDI from: wget http://www.piano-midi.de/midis/beethoven/beethoven_hammerklavier_2.mid
+output = pipe(processor("beethoven_hammerklavier_2.mid"))
+
+audio = output.audios[0]
+```
+
+## SpectrogramDiffusionPipeline
+[[autodoc]] SpectrogramDiffusionPipeline
+	- all
+	- __call__
@@ -131,10 +131,153 @@ This should take only around 3-4 seconds on GPU (depending on hardware). The out
 ![img](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/vermeer_disco_dancing.png)


-**Note**: To see how to run all other ControlNet checkpoints, please have a look at [ControlNet with Stable Diffusion 1.5](#controlnet-with-stable-diffusion-1.5)
+**Note**: To see how to run all other ControlNet checkpoints, please have a look at [ControlNet with Stable Diffusion 1.5](#controlnet-with-stable-diffusion-1.5).

 <!-- TODO: add space -->

+## Combining multiple conditionings
+
+Multiple ControlNet conditionings can be combined for a single image generation. Pass a list of ControlNets to the pipeline's constructor and a corresponding list of conditionings to `__call__`.
+
+When combining conditionings, it is helpful to mask conditionings such that they do not overlap. In the example, we mask the middle of the canny map where the pose conditioning is located.
+
+It can also be helpful to vary the `controlnet_conditioning_scales` to emphasize one conditioning over the other.
+
+### Canny conditioning
+
+The original image:
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/landscape.png"/>
+
+Prepare the conditioning:
+
+```python 
+from diffusers.utils import load_image
+from PIL import Image
+import cv2
+import numpy as np
+from diffusers.utils import load_image
+
+canny_image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/landscape.png"
+)
+canny_image = np.array(canny_image)
+
+low_threshold = 100
+high_threshold = 200
+
+canny_image = cv2.Canny(canny_image, low_threshold, high_threshold)
+
+# zero out middle columns of image where pose will be overlayed
+zero_start = canny_image.shape[1] // 4
+zero_end = zero_start + canny_image.shape[1] // 2
+canny_image[:, zero_start:zero_end] = 0
+
+canny_image = canny_image[:, :, None]
+canny_image = np.concatenate([canny_image, canny_image, canny_image], axis=2)
+canny_image = Image.fromarray(canny_image)
+```
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/controlnet/landscape_canny_masked.png"/>
+
+### Openpose conditioning
+
+The original image:
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/person.png" width=600/>
+
+Prepare the conditioning:
+
+```python
+from controlnet_aux import OpenposeDetector
+from diffusers.utils import load_image
+
+openpose = OpenposeDetector.from_pretrained("lllyasviel/ControlNet")
+
+openpose_image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/person.png"
+)
+openpose_image = openpose(openpose_image)
+```
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/controlnet/person_pose.png" width=600/>
+
+### Running ControlNet with multiple conditionings
+
+```python
+from diffusers import StableDiffusionControlNetPipeline, ControlNetModel, UniPCMultistepScheduler
+import torch
+
+controlnet = [
+    ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-openpose", torch_dtype=torch.float16),
+    ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny", torch_dtype=torch.float16),
+]
+
+pipe = StableDiffusionControlNetPipeline.from_pretrained(
+    "runwayml/stable-diffusion-v1-5", controlnet=controlnet, torch_dtype=torch.float16
+)
+pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config)
+
+pipe.enable_xformers_memory_efficient_attention()
+pipe.enable_model_cpu_offload()
+
+prompt = "a giant standing in a fantasy landscape, best quality"
+negative_prompt = "monochrome, lowres, bad anatomy, worst quality, low quality"
+
+generator = torch.Generator(device="cpu").manual_seed(1)
+
+images = [openpose_image, canny_image]
+
+image = pipe(
+    prompt,
+    images,
+    num_inference_steps=20,
+    generator=generator,
+    negative_prompt=negative_prompt,
+    controlnet_conditioning_scale=[1.0, 0.8],
+).images[0]
+
+image.save("./multi_controlnet_output.png")
+```
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/controlnet/multi_controlnet_output.png" width=600/>
+
+### Guess Mode
+
+Guess Mode is [a ControlNet feature that was implemented](https://github.com/lllyasviel/ControlNet#guess-mode--non-prompt-mode) after the publication of [the paper](https://arxiv.org/abs/2302.05543). The description states:
+
+>In this mode, the ControlNet encoder will try best to recognize the content of the input control map, like depth map, edge map, scribbles, etc, even if you remove all prompts.
+
+#### The core implementation:
+
+It adjusts the scale of the output residuals from ControlNet by a fixed ratio depending on the block depth. The shallowest DownBlock corresponds to `0.1`. As the blocks get deeper, the scale increases exponentially, and the scale for the output of the MidBlock becomes `1.0`. 
+
+Since the core implementation is just this, **it does not have any impact on prompt conditioning**. While it is common to use it without specifying any prompts, it is also possible to provide prompts if desired.
+
+#### Usage:
+
+Just specify `guess_mode=True` in the pipe() function. A `guidance_scale` between 3.0 and 5.0 is [recommended](https://github.com/lllyasviel/ControlNet#guess-mode--non-prompt-mode).
+```py
+from diffusers import StableDiffusionControlNetPipeline, ControlNetModel
+import torch
+
+controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny")
+pipe = StableDiffusionControlNetPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", controlnet=controlnet).to(
+    "cuda"
+)
+image = pipe("", image=canny_image, guess_mode=True, guidance_scale=3.0).images[0]
+image.save("guess_mode_generated.png")
+```
+
+#### Output image comparison:
+Canny Control Example
+
+|no guess_mode with prompt|guess_mode without prompt|
+|---|---|
+|<a href="https://huggingface.co/takuma104/controlnet_dev/resolve/main/gen_compare_guess_mode/output_images/diffusers/output_bird_canny_0.png"><img width="128" src="https://huggingface.co/takuma104/controlnet_dev/resolve/main/gen_compare_guess_mode/output_images/diffusers/output_bird_canny_0.png"/></a>|<a href="https://huggingface.co/takuma104/controlnet_dev/resolve/main/gen_compare_guess_mode/output_images/diffusers/output_bird_canny_0_gm.png"><img width="128" src="https://huggingface.co/takuma104/controlnet_dev/resolve/main/gen_compare_guess_mode/output_images/diffusers/output_bird_canny_0_gm.png"/></a>|
+
+
+
 ## Available checkpoints

 ControlNet requires a *control image* in addition to the text-to-image *prompt*. 
@@ -165,3 +308,10 @@ All checkpoints can be found under the authors' namespace [lllyasviel](https://h
 	- disable_vae_slicing
 	- enable_xformers_memory_efficient_attention
 	- disable_xformers_memory_efficient_attention
+	- load_textual_inversion
+
+## FlaxStableDiffusionControlNetPipeline
+[[autodoc]] FlaxStableDiffusionControlNetPipeline
+	- all
+	- __call__
+
@@ -30,4 +30,7 @@ Available Checkpoints are:
 	- enable_attention_slicing
 	- disable_attention_slicing
 	- enable_xformers_memory_efficient_attention
-	- disable_xformers_memory_efficient_attention
+	- disable_xformers_memory_efficient_attention
+	- load_textual_inversion
+	- load_lora_weights
+	- save_lora_weights
@@ -14,7 +14,7 @@ specific language governing permissions and limitations under the License.

 ## StableDiffusionImageVariationPipeline

-[`StableDiffusionImageVariationPipeline`] lets you generate variations from an input image using Stable Diffusion. It uses a fine-tuned version of Stable Diffusion model, trained by  [Justin Pinkney](https://www.justinpinkney.com/) (@Buntworthy) at [Lambda](https://lambdalabs.com/)
+[`StableDiffusionImageVariationPipeline`] lets you generate variations from an input image using Stable Diffusion. It uses a fine-tuned version of Stable Diffusion model, trained by  [Justin Pinkney](https://www.justinpinkney.com/) (@Buntworthy) at [Lambda](https://lambdalabs.com/).

 The original codebase can be found here:
 [Stable Diffusion Image Variations](https://github.com/LambdaLabsML/lambda-diffusers#stable-diffusion-image-variations)
@@ -28,4 +28,4 @@ Available Checkpoints are:
 	- enable_attention_slicing
 	- disable_attention_slicing
 	- enable_xformers_memory_efficient_attention
-	- disable_xformers_memory_efficient_attention
+	- disable_xformers_memory_efficient_attention
@@ -30,7 +30,11 @@ proposed by Chenlin Meng, Yutong He, Yang Song, Jiaming Song, Jiajun Wu, Jun-Yan
 	- disable_attention_slicing
 	- enable_xformers_memory_efficient_attention
 	- disable_xformers_memory_efficient_attention
+	- load_textual_inversion
+	- from_ckpt
+	- load_lora_weights
+	- save_lora_weights

 [[autodoc]] FlaxStableDiffusionImg2ImgPipeline
 	- all
-	- __call__
+	- __call__
@@ -31,7 +31,10 @@ Available checkpoints are:
 	- disable_attention_slicing
 	- enable_xformers_memory_efficient_attention
 	- disable_xformers_memory_efficient_attention
+	- load_textual_inversion
+	- load_lora_weights
+	- save_lora_weights

 [[autodoc]] FlaxStableDiffusionInpaintPipeline
 	- all
-	- __call__
+	- __call__
@@ -0,0 +1,61 @@
+<!--Copyright 2023 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
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Editing Implicit Assumptions in Text-to-Image Diffusion Models
+
+## Overview
+
+[Editing Implicit Assumptions in Text-to-Image Diffusion Models](https://arxiv.org/abs/2303.08084) by Hadas Orgad, Bahjat Kawar, and Yonatan Belinkov.
+
+The abstract of the paper is the following:
+
+*Text-to-image diffusion models often make implicit assumptions about the world when generating images. While some assumptions are useful (e.g., the sky is blue), they can also be outdated, incorrect, or reflective of social biases present in the training data. Thus, there is a need to control these assumptions without requiring explicit user input or costly re-training. In this work, we aim to edit a given implicit assumption in a pre-trained diffusion model. Our Text-to-Image Model Editing method, TIME for short, receives a pair of inputs: a "source" under-specified prompt for which the model makes an implicit assumption (e.g., "a pack of roses"), and a "destination" prompt that describes the same setting, but with a specified desired attribute (e.g., "a pack of blue roses"). TIME then updates the model's cross-attention layers, as these layers assign visual meaning to textual tokens. We edit the projection matrices in these layers such that the source prompt is projected close to the destination prompt. Our method is highly efficient, as it modifies a mere 2.2% of the model's parameters in under one second. To evaluate model editing approaches, we introduce TIMED (TIME Dataset), containing 147 source and destination prompt pairs from various domains. Our experiments (using Stable Diffusion) show that TIME is successful in model editing, generalizes well for related prompts unseen during editing, and imposes minimal effect on unrelated generations.*
+
+Resources:
+
+* [Project Page](https://time-diffusion.github.io/).
+* [Paper](https://arxiv.org/abs/2303.08084).
+* [Original Code](https://github.com/bahjat-kawar/time-diffusion).
+* [Demo](https://huggingface.co/spaces/bahjat-kawar/time-diffusion).
+
+## Available Pipelines:
+
+| Pipeline | Tasks | Demo
+|---|---|:---:|
+| [StableDiffusionModelEditingPipeline](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_model_editing.py) | *Text-to-Image Model Editing* | [🤗 Space](https://huggingface.co/spaces/bahjat-kawar/time-diffusion)) |
+
+This pipeline enables editing the diffusion model weights, such that its assumptions on a given concept are changed. The resulting change is expected to take effect in all prompt generations pertaining to the edited concept.
+
+## Usage example
+
+```python
+import torch
+from diffusers import StableDiffusionModelEditingPipeline
+
+model_ckpt = "CompVis/stable-diffusion-v1-4"
+pipe = StableDiffusionModelEditingPipeline.from_pretrained(model_ckpt)
+
+pipe = pipe.to("cuda")
+
+source_prompt = "A pack of roses"
+destination_prompt = "A pack of blue roses"
+pipe.edit_model(source_prompt, destination_prompt)
+
+prompt = "A field of roses"
+image = pipe(prompt).images[0]
+image.save("field_of_roses.png")
+```
+
+## StableDiffusionModelEditingPipeline
+[[autodoc]] StableDiffusionModelEditingPipeline
+	- __call__
+	- all
@@ -35,6 +35,7 @@ For more details about how Stable Diffusion works and how it differs from the ba
 | [StableDiffusionInstructPix2PixPipeline](./pix2pix) | **Experimental** – *Text-Based Image Editing * | | [InstructPix2Pix: Learning to Follow Image Editing Instructions](https://huggingface.co/spaces/timbrooks/instruct-pix2pix)
 | [StableDiffusionAttendAndExcitePipeline](./attend_and_excite) | **Experimental** – *Text-to-Image Generation * | | [Attend-and-Excite: Attention-Based Semantic Guidance for Text-to-Image Diffusion Models](https://huggingface.co/spaces/AttendAndExcite/Attend-and-Excite)
 | [StableDiffusionPix2PixZeroPipeline](./pix2pix_zero) | **Experimental** – *Text-Based Image Editing * | | [Zero-shot Image-to-Image Translation](https://arxiv.org/abs/2302.03027)
+| [StableDiffusionModelEditingPipeline](./model_editing) | **Experimental** – *Text-to-Image Model Editing * | | [Editing Implicit Assumptions in Text-to-Image Diffusion Models](https://arxiv.org/abs/2303.08084)



@@ -68,3 +68,6 @@ images[0].save("snowy_mountains.png")
 [[autodoc]] StableDiffusionInstructPix2PixPipeline
 	- __call__
 	- all
+	- load_textual_inversion
+	- load_lora_weights
+	- save_lora_weights
@@ -14,25 +14,26 @@ specific language governing permissions and limitations under the License.

 ## Overview

-[Self-Attention Guidance](https://arxiv.org/abs/2210.00939) by Susung Hong et al.
+[Improving Sample Quality of Diffusion Models Using Self-Attention Guidance](https://arxiv.org/abs/2210.00939) by Susung Hong et al.

 The abstract of the paper is the following:

-*Denoising diffusion models (DDMs) have been drawing much attention for their appreciable sample quality and diversity. Despite their remarkable performance, DDMs remain black boxes on which further study is necessary to take a profound step. Motivated by this, we delve into the design of conventional U-shaped diffusion models. More specifically, we investigate the self-attention modules within these models through carefully designed experiments and explore their characteristics. In addition, inspired by the studies that substantiate the effectiveness of the guidance schemes, we present plug-and-play diffusion guidance, namely Self-Attention Guidance (SAG), that can drastically boost the performance of existing diffusion models. Our method, SAG, extracts the intermediate attention map from a diffusion model at every iteration and selects tokens above a certain attention score for masking and blurring to obtain a partially blurred input. Subsequently, we measure the dissimilarity between the predicted noises obtained from feeding the blurred and original input to the diffusion model and leverage it as guidance. With this guidance, we observe apparent improvements in a wide range of diffusion models, e.g., ADM, IDDPM, and Stable Diffusion, and show that the results further improve by combining our method with the conventional guidance scheme. We provide extensive ablation studies to verify our choices.*
+*Denoising diffusion models (DDMs) have attracted attention for their exceptional generation quality and diversity. This success is largely attributed to the use of class- or text-conditional diffusion guidance methods, such as classifier and classifier-free guidance. In this paper, we present a more comprehensive perspective that goes beyond the traditional guidance methods. From this generalized perspective, we introduce novel condition- and training-free strategies to enhance the quality of generated images. As a simple solution, blur guidance improves the suitability of intermediate samples for their fine-scale information and structures, enabling diffusion models to generate higher quality samples with a moderate guidance scale. Improving upon this, Self-Attention Guidance (SAG) uses the intermediate self-attention maps of diffusion models to enhance their stability and efficacy. Specifically, SAG adversarially blurs only the regions that diffusion models attend to at each iteration and guides them accordingly. Our experimental results show that our SAG improves the performance of various diffusion models, including ADM, IDDPM, Stable Diffusion, and DiT. Moreover, combining SAG with conventional guidance methods leads to further improvement.*

 Resources:

 * [Project Page](https://ku-cvlab.github.io/Self-Attention-Guidance).
 * [Paper](https://arxiv.org/abs/2210.00939).
 * [Original Code](https://github.com/KU-CVLAB/Self-Attention-Guidance).
-* [Demo](https://colab.research.google.com/github/SusungHong/Self-Attention-Guidance/blob/main/SAG_Stable.ipynb).
+* [Hugging Face Demo](https://huggingface.co/spaces/susunghong/Self-Attention-Guidance).
+* [Colab Demo](https://colab.research.google.com/github/SusungHong/Self-Attention-Guidance/blob/main/SAG_Stable.ipynb).


 ## Available Pipelines:

 | Pipeline | Tasks | Demo
 |---|---|:---:|
-| [StableDiffusionSAGPipeline](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_sag.py) | *Text-to-Image Generation* | [Colab](https://colab.research.google.com/github/SusungHong/Self-Attention-Guidance/blob/main/SAG_Stable.ipynb) |
+| [StableDiffusionSAGPipeline](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_sag.py) | *Text-to-Image Generation* | [🤗 Space](https://huggingface.co/spaces/susunghong/Self-Attention-Guidance) |

 ## Usage example

@@ -39,6 +39,10 @@ Available Checkpoints are:
 	- disable_xformers_memory_efficient_attention
 	- enable_vae_tiling
 	- disable_vae_tiling
+	- load_textual_inversion
+	- from_ckpt
+	- load_lora_weights
+	- save_lora_weights

 [[autodoc]] FlaxStableDiffusionPipeline
 	- all
@@ -28,15 +28,15 @@ The abstract of the paper is the following:

 ## Tips

- Safe Stable Diffusion may also be used with weights of [Stable Diffusion](./api/pipelines/stable_diffusion/text2img).
+- Safe Stable Diffusion may also be used with weights of [Stable Diffusion](./stable_diffusion/text2img).

 ### Run Safe Stable Diffusion

-Safe Stable Diffusion can be tested very easily with the [`StableDiffusionPipelineSafe`], and the `"AIML-TUDA/stable-diffusion-safe"` checkpoint exactly in the same way it is shown in the [Conditional Image Generation Guide](./using-diffusers/conditional_image_generation).
+Safe Stable Diffusion can be tested very easily with the [`StableDiffusionPipelineSafe`], and the `"AIML-TUDA/stable-diffusion-safe"` checkpoint exactly in the same way it is shown in the [Conditional Image Generation Guide](../../using-diffusers/conditional_image_generation).

 ### Interacting with the Safety Concept

-To check and edit the currently used safety concept, use the `safety_concept` property of [`StableDiffusionPipelineSafe`]
+To check and edit the currently used safety concept, use the `safety_concept` property of [`StableDiffusionPipelineSafe`]:
 ```python
 >>> from diffusers import StableDiffusionPipelineSafe

@@ -60,7 +60,7 @@ You may use the 4 configurations defined in the [Safe Latent Diffusion paper](ht

 The following configurations are available: `SafetyConfig.WEAK`, `SafetyConfig.MEDIUM`, `SafetyConfig.STRONG`, and `SafetyConfig.MAX`.

-### How to load and use different schedulers.
+### How to load and use different schedulers

 The safe stable diffusion pipeline uses [`PNDMScheduler`] scheduler by default. But `diffusers` provides many other schedulers that can be used with the stable diffusion pipeline such as [`DDIMScheduler`], [`LMSDiscreteScheduler`], [`EulerDiscreteScheduler`], [`EulerAncestralDiscreteScheduler`] etc.
 To use a different scheduler, you can either change it via the [`ConfigMixin.from_config`] method or pass the `scheduler` argument to the `from_pretrained` method of the pipeline. For example, to use the [`EulerDiscreteScheduler`], you can do the following:
@@ -16,6 +16,10 @@ Stable unCLIP checkpoints are finetuned from [stable diffusion 2.1](./stable_dif
 Stable unCLIP also still conditions on text embeddings. Given the two separate conditionings, stable unCLIP can be used
 for text guided image variation. When combined with an unCLIP prior, it can also be used for full text to image generation.

+To know more about the unCLIP process, check out the following paper:
+
+[Hierarchical Text-Conditional Image Generation with CLIP Latents](https://arxiv.org/abs/2204.06125) by Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, Mark Chen.
+
 ## Tips

 Stable unCLIP takes a `noise_level` as input during inference. `noise_level` determines how much noise is added 
@@ -24,50 +28,124 @@ we do not add any additional noise to the image embeddings i.e. `noise_level = 0

 ### Available checkpoints:

-TODO
+* Image variation
+	* [stabilityai/stable-diffusion-2-1-unclip](https://hf.co/stabilityai/stable-diffusion-2-1-unclip)
+	* [stabilityai/stable-diffusion-2-1-unclip-small](https://hf.co/stabilityai/stable-diffusion-2-1-unclip-small)
+* Text-to-image 
+	* [stabilityai/stable-diffusion-2-1-unclip-small](https://hf.co/stabilityai/stable-diffusion-2-1-unclip-small)

 ### Text-to-Image Generation
+Stable unCLIP can be leveraged for text-to-image generation by pipelining it with the prior model of KakaoBrain's open source DALL-E 2 replication [Karlo](https://huggingface.co/kakaobrain/karlo-v1-alpha)

 ```python
 import torch
-from diffusers import StableUnCLIPPipeline
+from diffusers import UnCLIPScheduler, DDPMScheduler, StableUnCLIPPipeline
+from diffusers.models import PriorTransformer
+from transformers import CLIPTokenizer, CLIPTextModelWithProjection
+
+prior_model_id = "kakaobrain/karlo-v1-alpha"
+data_type = torch.float16
+prior = PriorTransformer.from_pretrained(prior_model_id, subfolder="prior", torch_dtype=data_type)
+
+prior_text_model_id = "openai/clip-vit-large-patch14"
+prior_tokenizer = CLIPTokenizer.from_pretrained(prior_text_model_id)
+prior_text_model = CLIPTextModelWithProjection.from_pretrained(prior_text_model_id, torch_dtype=data_type)
+prior_scheduler = UnCLIPScheduler.from_pretrained(prior_model_id, subfolder="prior_scheduler")
+prior_scheduler = DDPMScheduler.from_config(prior_scheduler.config)
+
+stable_unclip_model_id = "stabilityai/stable-diffusion-2-1-unclip-small"

 pipe = StableUnCLIPPipeline.from_pretrained(
-    "fusing/stable-unclip-2-1-l", torch_dtype=torch.float16
-)  # TODO update model path
+    stable_unclip_model_id,
+    torch_dtype=data_type,
+    variant="fp16",
+    prior_tokenizer=prior_tokenizer,
+    prior_text_encoder=prior_text_model,
+    prior=prior,
+    prior_scheduler=prior_scheduler,
+)
+
 pipe = pipe.to("cuda")
+wave_prompt = "dramatic wave, the Oceans roar, Strong wave spiral across the oceans as the waves unfurl into roaring crests; perfect wave form; perfect wave shape; dramatic wave shape; wave shape unbelievable; wave; wave shape spectacular"

-prompt = "a photo of an astronaut riding a horse on mars"
-images = pipe(prompt).images
-images[0].save("astronaut_horse.png")
+images = pipe(prompt=wave_prompt).images
+images[0].save("waves.png")
 ```
+<Tip warning={true}>

+For text-to-image we use `stabilityai/stable-diffusion-2-1-unclip-small` as it was trained on CLIP ViT-L/14 embedding, the same as the Karlo model prior. [stabilityai/stable-diffusion-2-1-unclip](https://hf.co/stabilityai/stable-diffusion-2-1-unclip) was trained on OpenCLIP ViT-H, so we don't recommend its use. 
+
+</Tip>

 ### Text guided Image-to-Image Variation

 ```python
-import requests
-import torch
-from PIL import Image
-from io import BytesIO
-
 from diffusers import StableUnCLIPImg2ImgPipeline
+from diffusers.utils import load_image
+import torch

 pipe = StableUnCLIPImg2ImgPipeline.from_pretrained(
-    "fusing/stable-unclip-2-1-l-img2img", torch_dtype=torch.float16
-)  # TODO update model path
+    "stabilityai/stable-diffusion-2-1-unclip", torch_dtype=torch.float16, variation="fp16"
+)
 pipe = pipe.to("cuda")

-url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg"
+url = "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/stable_unclip/tarsila_do_amaral.png"
+init_image = load_image(url)

-response = requests.get(url)
-init_image = Image.open(BytesIO(response.content)).convert("RGB")
-init_image = init_image.resize((768, 512))
+images = pipe(init_image).images
+images[0].save("variation_image.png")
+```

+Optionally, you can also pass a prompt to `pipe` such as:
+
+```python 
 prompt = "A fantasy landscape, trending on artstation"

-images = pipe(prompt, init_image).images
-images[0].save("fantasy_landscape.png")
+images = pipe(init_image, prompt=prompt).images
+images[0].save("variation_image_two.png")
+```
+
+### Memory optimization
+
+If you are short on GPU memory, you can enable smart CPU offloading so that models that are not needed
+immediately for a computation can be offloaded to CPU:
+
+```python 
+from diffusers import StableUnCLIPImg2ImgPipeline
+from diffusers.utils import load_image
+import torch
+
+pipe = StableUnCLIPImg2ImgPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-2-1-unclip", torch_dtype=torch.float16, variation="fp16"
+)
+# Offload to CPU.
+pipe.enable_model_cpu_offload()
+
+url = "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/stable_unclip/tarsila_do_amaral.png"
+init_image = load_image(url)
+
+images = pipe(init_image).images
+images[0]
+```
+
+Further memory optimizations are possible by enabling VAE slicing on the pipeline: 
+
+```python 
+from diffusers import StableUnCLIPImg2ImgPipeline
+from diffusers.utils import load_image
+import torch
+
+pipe = StableUnCLIPImg2ImgPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-2-1-unclip", torch_dtype=torch.float16, variation="fp16"
+)
+pipe.enable_model_cpu_offload()
+pipe.enable_vae_slicing()
+
+url = "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/stable_unclip/tarsila_do_amaral.png"
+init_image = load_image(url)
+
+images = pipe(init_image).images
+images[0]
 ```

 ### StableUnCLIPPipeline
@@ -0,0 +1,130 @@
+<!--Copyright 2023 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
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+<Tip warning={true}>
+
+This pipeline is for research purposes only. 
+
+</Tip>
+
+# Text-to-video synthesis
+
+## Overview
+
+[VideoFusion: Decomposed Diffusion Models for High-Quality Video Generation](https://arxiv.org/abs/2303.08320) by Zhengxiong Luo, Dayou Chen, Yingya Zhang, Yan Huang, Liang Wang, Yujun Shen, Deli Zhao, Jingren Zhou, Tieniu Tan.
+
+The abstract of the paper is the following:
+
+*A diffusion probabilistic model (DPM), which constructs a forward diffusion process by gradually adding noise to data points and learns the reverse denoising process to generate new samples, has been shown to handle complex data distribution. Despite its recent success in image synthesis, applying DPMs to video generation is still challenging due to high-dimensional data spaces. Previous methods usually adopt a standard diffusion process, where frames in the same video clip are destroyed with independent noises, ignoring the content redundancy and temporal correlation. This work presents a decomposed diffusion process via resolving the per-frame noise into a base noise that is shared among all frames and a residual noise that varies along the time axis. The denoising pipeline employs two jointly-learned networks to match the noise decomposition accordingly. Experiments on various datasets confirm that our approach, termed as VideoFusion, surpasses both GAN-based and diffusion-based alternatives in high-quality video generation. We further show that our decomposed formulation can benefit from pre-trained image diffusion models and well-support text-conditioned video creation.*
+
+Resources:
+
+* [Website](https://modelscope.cn/models/damo/text-to-video-synthesis/summary)
+* [GitHub repository](https://github.com/modelscope/modelscope/)
+* [🤗 Spaces](https://huggingface.co/spaces/damo-vilab/modelscope-text-to-video-synthesis)
+
+## Available Pipelines:
+
+| Pipeline | Tasks | Demo
+|---|---|:---:|
+| [TextToVideoSDPipeline](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth.py) | *Text-to-Video Generation* | [🤗 Spaces](https://huggingface.co/spaces/damo-vilab/modelscope-text-to-video-synthesis)
+
+## Usage example 
+
+Let's start by generating a short video with the default length of 16 frames (2s at 8 fps):
+
+```python 
+import torch
+from diffusers import DiffusionPipeline
+from diffusers.utils import export_to_video
+
+pipe = DiffusionPipeline.from_pretrained("damo-vilab/text-to-video-ms-1.7b", torch_dtype=torch.float16, variant="fp16")
+pipe = pipe.to("cuda")
+
+prompt = "Spiderman is surfing"
+video_frames = pipe(prompt).frames
+video_path = export_to_video(video_frames)
+video_path
+```
+
+Diffusers supports different optimization techniques to improve the latency
+and memory footprint of a pipeline. Since videos are often more memory-heavy than images,
+we can enable CPU offloading and VAE slicing to keep the memory footprint at bay.
+
+Let's generate a video of 8 seconds (64 frames) on the same GPU using CPU offloading and VAE slicing:
+
+```python
+import torch
+from diffusers import DiffusionPipeline
+from diffusers.utils import export_to_video
+
+pipe = DiffusionPipeline.from_pretrained("damo-vilab/text-to-video-ms-1.7b", torch_dtype=torch.float16, variant="fp16")
+pipe.enable_model_cpu_offload()
+
+# memory optimization
+pipe.enable_vae_slicing()
+
+prompt = "Darth Vader surfing a wave"
+video_frames = pipe(prompt, num_frames=64).frames
+video_path = export_to_video(video_frames)
+video_path
+```
+
+It just takes **7 GBs of GPU memory** to generate the 64 video frames using PyTorch 2.0, "fp16" precision and the techniques mentioned above.
+
+We can also use a different scheduler easily, using the same method we'd use for Stable Diffusion:
+
+```python
+import torch
+from diffusers import DiffusionPipeline, DPMSolverMultistepScheduler
+from diffusers.utils import export_to_video
+
+pipe = DiffusionPipeline.from_pretrained("damo-vilab/text-to-video-ms-1.7b", torch_dtype=torch.float16, variant="fp16")
+pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)
+pipe.enable_model_cpu_offload()
+
+prompt = "Spiderman is surfing"
+video_frames = pipe(prompt, num_inference_steps=25).frames
+video_path = export_to_video(video_frames)
+video_path
+```
+
+Here are some sample outputs: 
+
+<table>
+    <tr>
+        <td><center>
+        An astronaut riding a horse.
+        <br>
+        <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/astr.gif"
+            alt="An astronaut riding a horse."
+            style="width: 300px;" />
+        </center></td>
+        <td ><center>
+        Darth vader surfing in waves.
+        <br>
+        <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/vader.gif"
+            alt="Darth vader surfing in waves."
+            style="width: 300px;" />
+        </center></td>
+    </tr>
+</table>
+
+## Available checkpoints 
+
+* [damo-vilab/text-to-video-ms-1.7b](https://huggingface.co/damo-vilab/text-to-video-ms-1.7b/)
+* [damo-vilab/text-to-video-ms-1.7b-legacy](https://huggingface.co/damo-vilab/text-to-video-ms-1.7b-legacy)
+
+## TextToVideoSDPipeline
+[[autodoc]] TextToVideoSDPipeline
+	- all
+	- __call__
@@ -0,0 +1,240 @@
+<!--Copyright 2023 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
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Zero-Shot Text-to-Video Generation 
+
+## Overview
+
+
+[Text2Video-Zero: Text-to-Image Diffusion Models are Zero-Shot Video Generators](https://arxiv.org/abs/2303.13439) by
+Levon Khachatryan,
+Andranik Movsisyan,
+Vahram Tadevosyan,
+Roberto Henschel,
+[Zhangyang Wang](https://www.ece.utexas.edu/people/faculty/atlas-wang), Shant Navasardyan, [Humphrey Shi](https://www.humphreyshi.com).
+
+Our method Text2Video-Zero enables zero-shot video generation using either
+1. A textual prompt, or
+2. A prompt combined with guidance from poses or edges, or 
+3. Video Instruct-Pix2Pix, i.e., instruction-guided video editing.
+
+Results are temporally consistent and follow closely the guidance and textual prompts.
+
+![teaser-img](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/t2v_zero_teaser.png)
+
+The abstract of the paper is the following:
+
+*Recent text-to-video generation approaches rely on computationally heavy training and require large-scale video datasets. In this paper, we introduce a new task of zero-shot text-to-video generation and propose a low-cost approach (without any training or optimization) by leveraging the power of existing text-to-image synthesis methods (e.g., Stable Diffusion), making them suitable for the video domain.
+Our key modifications include (i) enriching the latent codes of the generated frames with motion dynamics to keep the global scene and the background time consistent; and (ii) reprogramming frame-level self-attention using a new cross-frame attention of each frame on the first frame, to preserve the context, appearance, and identity of the foreground object.
+Experiments show that this leads to low overhead, yet high-quality and remarkably consistent video generation. Moreover, our approach is not limited to text-to-video synthesis but is also applicable to other tasks such as conditional and content-specialized video generation, and Video Instruct-Pix2Pix, i.e., instruction-guided video editing.
+As experiments show, our method performs comparably or sometimes better than recent approaches, despite not being trained on additional video data.*
+
+
+
+Resources:
+
+* [Project Page](https://text2video-zero.github.io/)
+* [Paper](https://arxiv.org/abs/2303.13439)
+* [Original Code](https://github.com/Picsart-AI-Research/Text2Video-Zero)
+
+
+## Available Pipelines:
+
+| Pipeline | Tasks | Demo
+|---|---|:---:|
+| [TextToVideoZeroPipeline](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_zero.py) | *Zero-shot Text-to-Video Generation* | [🤗 Space](https://huggingface.co/spaces/PAIR/Text2Video-Zero)
+
+
+## Usage example
+
+### Text-To-Video
+
+To generate a video from prompt, run the following python command
+```python
+import torch
+import imageio
+from diffusers import TextToVideoZeroPipeline
+
+model_id = "runwayml/stable-diffusion-v1-5"
+pipe = TextToVideoZeroPipeline.from_pretrained(model_id, torch_dtype=torch.float16).to("cuda")
+
+prompt = "A panda is playing guitar on times square"
+result = pipe(prompt=prompt).images
+result = [(r * 255).astype("uint8") for r in result]
+imageio.mimsave("video.mp4", result, fps=4)
+```
+You can change these parameters in the pipeline call:
+* Motion field strength (see the [paper](https://arxiv.org/abs/2303.13439), Sect. 3.3.1):
+    * `motion_field_strength_x` and `motion_field_strength_y`. Default: `motion_field_strength_x=12`, `motion_field_strength_y=12`
+* `T` and `T'` (see the [paper](https://arxiv.org/abs/2303.13439), Sect. 3.3.1)
+    * `t0` and `t1` in the range `{0, ..., num_inference_steps}`. Default: `t0=45`, `t1=48`
+* Video length:
+    * `video_length`, the number of frames video_length to be generated. Default: `video_length=8`
+
+
+### Text-To-Video with Pose Control
+To generate a video from prompt with additional pose control
+
+1. Download a demo video
+
+    ```python
+    from huggingface_hub import hf_hub_download
+
+    filename = "__assets__/poses_skeleton_gifs/dance1_corr.mp4"
+    repo_id = "PAIR/Text2Video-Zero"
+    video_path = hf_hub_download(repo_type="space", repo_id=repo_id, filename=filename)
+    ```
+
+
+2. Read video containing extracted pose images
+    ```python
+    from PIL import Image
+    import imageio
+
+    reader = imageio.get_reader(video_path, "ffmpeg")
+    frame_count = 8
+    pose_images = [Image.fromarray(reader.get_data(i)) for i in range(frame_count)]
+    ```
+    To extract pose from actual video, read [ControlNet documentation](./stable_diffusion/controlnet).
+
+3. Run `StableDiffusionControlNetPipeline` with our custom attention processor
+
+    ```python
+    import torch
+    from diffusers import StableDiffusionControlNetPipeline, ControlNetModel
+    from diffusers.pipelines.text_to_video_synthesis.pipeline_text_to_video_zero import CrossFrameAttnProcessor
+
+    model_id = "runwayml/stable-diffusion-v1-5"
+    controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-openpose", torch_dtype=torch.float16)
+    pipe = StableDiffusionControlNetPipeline.from_pretrained(
+        model_id, controlnet=controlnet, torch_dtype=torch.float16
+    ).to("cuda")
+
+    # Set the attention processor
+    pipe.unet.set_attn_processor(CrossFrameAttnProcessor(batch_size=2))
+    pipe.controlnet.set_attn_processor(CrossFrameAttnProcessor(batch_size=2))
+
+    # fix latents for all frames
+    latents = torch.randn((1, 4, 64, 64), device="cuda", dtype=torch.float16).repeat(len(pose_images), 1, 1, 1)
+
+    prompt = "Darth Vader dancing in a desert"
+    result = pipe(prompt=[prompt] * len(pose_images), image=pose_images, latents=latents).images
+    imageio.mimsave("video.mp4", result, fps=4)
+    ```
+
+
+### Text-To-Video with Edge Control
+
+To generate a video from prompt with additional pose control,
+follow the steps described above for pose-guided generation using [Canny edge ControlNet model](https://huggingface.co/lllyasviel/sd-controlnet-canny).
+
+
+### Video Instruct-Pix2Pix
+
+To perform text-guided video editing (with [InstructPix2Pix](./stable_diffusion/pix2pix)):
+
+1. Download a demo video
+
+    ```python
+    from huggingface_hub import hf_hub_download
+
+    filename = "__assets__/pix2pix video/camel.mp4"
+    repo_id = "PAIR/Text2Video-Zero"
+    video_path = hf_hub_download(repo_type="space", repo_id=repo_id, filename=filename)
+    ```
+
+2. Read video from path
+    ```python
+    from PIL import Image
+    import imageio
+
+    reader = imageio.get_reader(video_path, "ffmpeg")
+    frame_count = 8
+    video = [Image.fromarray(reader.get_data(i)) for i in range(frame_count)]
+    ```
+
+3. Run `StableDiffusionInstructPix2PixPipeline` with our custom attention processor
+    ```python
+    import torch
+    from diffusers import StableDiffusionInstructPix2PixPipeline
+    from diffusers.pipelines.text_to_video_synthesis.pipeline_text_to_video_zero import CrossFrameAttnProcessor
+
+    model_id = "timbrooks/instruct-pix2pix"
+    pipe = StableDiffusionInstructPix2PixPipeline.from_pretrained(model_id, torch_dtype=torch.float16).to("cuda")
+    pipe.unet.set_attn_processor(CrossFrameAttnProcessor(batch_size=3))
+
+    prompt = "make it Van Gogh Starry Night style"
+    result = pipe(prompt=[prompt] * len(video), image=video).images
+    imageio.mimsave("edited_video.mp4", result, fps=4)
+    ```
+
+
+### DreamBooth specialization 
+
+Methods **Text-To-Video**, **Text-To-Video with Pose Control** and **Text-To-Video with Edge Control**
+can run with custom [DreamBooth](../training/dreambooth) models, as shown below for
+[Canny edge ControlNet model](https://huggingface.co/lllyasviel/sd-controlnet-canny) and
+[Avatar style DreamBooth](https://huggingface.co/PAIR/text2video-zero-controlnet-canny-avatar) model
+
+1. Download a demo video
+
+    ```python
+    from huggingface_hub import hf_hub_download
+
+    filename = "__assets__/canny_videos_mp4/girl_turning.mp4"
+    repo_id = "PAIR/Text2Video-Zero"
+    video_path = hf_hub_download(repo_type="space", repo_id=repo_id, filename=filename)
+    ```
+
+2. Read video from path
+    ```python
+    from PIL import Image
+    import imageio
+
+    reader = imageio.get_reader(video_path, "ffmpeg")
+    frame_count = 8
+    video = [Image.fromarray(reader.get_data(i)) for i in range(frame_count)]
+    ```
+
+3. Run `StableDiffusionControlNetPipeline` with custom trained DreamBooth model
+    ```python
+    import torch
+    from diffusers import StableDiffusionControlNetPipeline, ControlNetModel
+    from diffusers.pipelines.text_to_video_synthesis.pipeline_text_to_video_zero import CrossFrameAttnProcessor
+
+    # set model id to custom model
+    model_id = "PAIR/text2video-zero-controlnet-canny-avatar"
+    controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny", torch_dtype=torch.float16)
+    pipe = StableDiffusionControlNetPipeline.from_pretrained(
+        model_id, controlnet=controlnet, torch_dtype=torch.float16
+    ).to("cuda")
+
+    # Set the attention processor
+    pipe.unet.set_attn_processor(CrossFrameAttnProcessor(batch_size=2))
+    pipe.controlnet.set_attn_processor(CrossFrameAttnProcessor(batch_size=2))
+
+    # fix latents for all frames
+    latents = torch.randn((1, 4, 64, 64), device="cuda", dtype=torch.float16).repeat(len(pose_images), 1, 1, 1)
+
+    prompt = "oil painting of a beautiful girl avatar style"
+    result = pipe(prompt=[prompt] * len(pose_images), image=pose_images, latents=latents).images
+    imageio.mimsave("video.mp4", result, fps=4)
+    ```
+
+You can filter out some available DreamBooth-trained models with [this link](https://huggingface.co/models?search=dreambooth).
+
+
+
+## TextToVideoZeroPipeline
+[[autodoc]] TextToVideoZeroPipeline
+	- all
+	- __call__
@@ -20,7 +20,7 @@ The abstract of the paper is the following:

 ## Tips

- VersatileDiffusion is conceptually very similar as [Stable Diffusion](./api/pipelines/stable_diffusion/overview),  but instead of providing just a image data stream conditioned on text, VersatileDiffusion provides both a image and text data stream and can be conditioned on both text and image.
+- VersatileDiffusion is conceptually very similar as [Stable Diffusion](./stable_diffusion/overview),  but instead of providing just a image data stream conditioned on text, VersatileDiffusion provides both a image and text data stream and can be conditioned on both text and image.

 ### *Run VersatileDiffusion*

@@ -10,7 +10,7 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Denoising diffusion implicit models (DDIM)
+# Denoising Diffusion Implicit Models (DDIM)

 ## Overview

@@ -24,4 +24,4 @@ The original codebase of this paper can be found here: [ermongroup/ddim](https:/
 For questions, feel free to contact the author on [tsong.me](https://tsong.me/).

 ## DDIMScheduler
-[[autodoc]] DDIMScheduler
+[[autodoc]] DDIMScheduler
@@ -10,7 +10,7 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Denoising diffusion probabilistic models (DDPM)
+# Denoising Diffusion Probabilistic Models (DDPM)

 ## Overview

@@ -24,4 +24,4 @@ We present high quality image synthesis results using diffusion probabilistic mo
 The original paper can be found [here](https://arxiv.org/abs/2010.02502).

 ## DDPMScheduler
-[[autodoc]] DDPMScheduler
+[[autodoc]] DDPMScheduler
@@ -14,8 +14,8 @@ specific language governing permissions and limitations under the License.

 ## Overview

-Ancestral sampling with Euler method steps. Based on the original (k-diffusion)[https://github.com/crowsonkb/k-diffusion/blob/481677d114f6ea445aa009cf5bd7a9cdee909e47/k_diffusion/sampling.py#L72] implementation by Katherine Crowson.
+Ancestral sampling with Euler method steps. Based on the original [k-diffusion](https://github.com/crowsonkb/k-diffusion/blob/481677d114f6ea445aa009cf5bd7a9cdee909e47/k_diffusion/sampling.py#L72) implementation by Katherine Crowson.
 Fast scheduler which often times generates good outputs with 20-30 steps.

 ## EulerAncestralDiscreteScheduler
-[[autodoc]] EulerAncestralDiscreteScheduler
+[[autodoc]] EulerAncestralDiscreteScheduler
@@ -10,11 +10,11 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# variance exploding stochastic differential equation (VE-SDE) scheduler
+# Variance Exploding Stochastic Differential Equation (VE-SDE) scheduler

 ## Overview

 Original paper can be found [here](https://arxiv.org/abs/2011.13456).

 ## ScoreSdeVeScheduler
-[[autodoc]] ScoreSdeVeScheduler
+[[autodoc]] ScoreSdeVeScheduler
@@ -10,7 +10,7 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Variance preserving stochastic differential equation (VP-SDE) scheduler
+# Variance Preserving Stochastic Differential Equation (VP-SDE) scheduler

 ## Overview

@@ -23,4 +23,4 @@ Score SDE-VP is under construction.
 </Tip>

 ## ScoreSdeVpScheduler
-[[autodoc]] schedulers.scheduling_sde_vp.ScoreSdeVpScheduler
+[[autodoc]] schedulers.scheduling_sde_vp.ScoreSdeVpScheduler
@@ -16,7 +16,7 @@ specific language governing permissions and limitations under the License.

 UniPC is a training-free framework designed for the fast sampling of diffusion models, which consists of a corrector (UniC) and a predictor (UniP) that share a unified analytical form and support arbitrary orders.

-For more details about the method, please refer to the [[paper]](https://arxiv.org/abs/2302.04867) and the [[code]](https://github.com/wl-zhao/UniPC).
+For more details about the method, please refer to the [paper](https://arxiv.org/abs/2302.04867) and the [code](https://github.com/wl-zhao/UniPC).

 Fast Sampling of Diffusion Models with Exponential Integrator.

@@ -12,83 +12,339 @@ specific language governing permissions and limitations under the License.

 # How to contribute to Diffusers 🧨

-We ❤️ contributions from the open-source community! Everyone is welcome, and all types of participation –not just code– are valued and appreciated. Answering questions, helping others, reaching out and improving the documentation are all immensely valuable to the community, so don't be afraid and get involved if you're up for it!
+We ❤️ contributions from the open-source community! Everyone is welcome, and all types of participation –not just code– are valued and appreciated. Answering questions, helping others, reaching out, and improving the documentation are all immensely valuable to the community, so don't be afraid and get involved if you're up for it!

-It also helps us if you spread the word: reference the library from blog posts
-on the awesome projects it made possible, shout out on Twitter every time it has
-helped you, or simply star the repo to say "thank you".
+Everyone is encouraged to start by saying 👋 in our public Discord channel. We discuss the latest trends in diffusion models, ask questions, show off personal projects, help each other with contributions, or just hang out ☕. <a href="https://Discord.gg/G7tWnz98XR"><img alt="Join us on Discord" src="https://img.shields.io/Discord/823813159592001537?color=5865F2&logo=Discord&logoColor=white"></a>

-We encourage everyone to start by saying 👋 in our public Discord channel. We discuss the hottest trends about diffusion models, ask questions, show-off personal projects, help each other with contributions, or just hang out ☕. <a href="https://discord.gg/G7tWnz98XR"><img alt="Join us on Discord" src="https://img.shields.io/discord/823813159592001537?color=5865F2&logo=discord&logoColor=white"></a>
-
-Whichever way you choose to contribute, we strive to be part of an open, welcoming and kind community. Please, read our [code of conduct](https://github.com/huggingface/diffusers/blob/main/CODE_OF_CONDUCT.md) and be mindful to respect it during your interactions.
+Whichever way you choose to contribute, we strive to be part of an open, welcoming, and kind community. Please, read our [code of conduct](https://github.com/huggingface/diffusers/blob/main/CODE_OF_CONDUCT.md) and be mindful to respect it during your interactions. We also recommend you become familiar with the [ethical guidelines](https://huggingface.co/docs/diffusers/conceptual/ethical_guidelines) that guide our project and ask you to adhere to the same principles of transparency and responsibility.

+We enormously value feedback from the community, so please do not be afraid to speak up if you believe you have valuable feedback that can help improve the library - every message, comment, issue, and pull request (PR) is read and considered.

 ## Overview

-You can contribute in so many ways! Just to name a few:
+You can contribute in many ways ranging from answering questions on issues to adding new diffusion models to
+the core library.

-* Fixing outstanding issues with the existing code.
-* Implementing [new diffusion pipelines](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines#contribution), [new schedulers](https://github.com/huggingface/diffusers/tree/main/src/diffusers/schedulers) or [new models](https://github.com/huggingface/diffusers/tree/main/src/diffusers/models).
-* [Contributing to the examples](https://github.com/huggingface/diffusers/tree/main/examples).
-* [Contributing to the documentation](https://github.com/huggingface/diffusers/tree/main/docs/source).
-* Submitting issues related to bugs or desired new features.
+In the following, we give an overview of different ways to contribute, ranked by difficulty in ascending order. All of them are valuable to the community.

-*All are equally valuable to the community.*
+* 1. Asking and answering questions on [the Diffusers discussion forum](https://discuss.huggingface.co/c/discussion-related-to-httpsgithubcomhuggingfacediffusers) or on [Discord](https://discord.gg/G7tWnz98XR).
+* 2. Opening new issues on [the GitHub Issues tab](https://github.com/huggingface/diffusers/issues/new/choose)
+* 3. Answering issues on [the GitHub Issues tab](https://github.com/huggingface/diffusers/issues)
+* 4. Fix a simple issue, marked by the "Good first issue" label, see [here](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22good+first+issue%22).
+* 5. Contribute to the [documentation](https://github.com/huggingface/diffusers/tree/main/docs/source).
+* 6. Contribute a [Community Pipeline](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3Acommunity-examples)
+* 7. Contribute to the [examples](https://github.com/huggingface/diffusers/tree/main/examples).
+* 8. Fix a more difficult issue, marked by the "Good second issue" label, see [here](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22Good+second+issue%22).
+* 9. Add a new pipeline, model, or scheduler, see ["New Pipeline/Model"](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+pipeline%2Fmodel%22) and ["New scheduler"](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+scheduler%22) issues. For this contribution, please have a look at [Design Philosophy](https://github.com/huggingface/diffusers/blob/main/PHILOSOPHY.md).

-### Browse GitHub issues for suggestions
+As said before, **all contributions are valuable to the community**.
+In the following, we will explain each contribution a bit more in detail.

-If you need inspiration, you can look out for [issues](https://github.com/huggingface/diffusers/issues) you'd like to tackle to contribute to the library. There are a few filters that can be helpful:
+For all contributions 4.-9. you will need to open a PR. It is explained in detail how to do so in [Opening a pull requst](#how-to-open-a-pr)

- See [Good first issues](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22good+first+issue%22) for general opportunities to contribute and getting started with the codebase.
- See [New pipeline/model](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+pipeline%2Fmodel%22) to contribute exciting new diffusion models or diffusion pipelines.
- See [New scheduler](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+scheduler%22) to work on new samplers and schedulers.
+### 1. Asking and answering questions on the Diffusers discussion forum or on the Diffusers Discord

+Any question or comment related to the Diffusers library can be asked on the [discussion forum](https://discuss.huggingface.co/c/discussion-related-to-httpsgithubcomhuggingfacediffusers/) or on [Discord](https://discord.gg/G7tWnz98XR). Such questions and comments include (but are not limited to):
+- Reports of training or inference experiments in an attempt to share knowledge
+- Presentation of personal projects
+- Questions to non-official training examples
+- Project proposals
+- General feedback
+- Paper summaries
+- Asking for help on personal projects that build on top of the Diffusers library
+- General questions
+- Ethical questions regarding diffusion models
+- ...

-## Submitting a new issue or feature request
+Every question that is asked on the forum or on Discord actively encourages the community to publicly
+share knowledge and might very well help a beginner in the future that has the same question you're
+having. Please do pose any questions you might have.
+In the same spirit, you are of immense help to the community by answering such questions because this way you are publicly documenting knowledge for everybody to learn from.

-Do your best to follow these guidelines when submitting an issue or a feature
-request. It will make it easier for us to come back to you quickly and with good
-feedback.
+**Please** keep in mind that the more effort you put into asking or answering a question, the higher
+the quality of the publicly documented knowledge. In the same way, well-posed and well-answered questions create a high-quality knowledge database accessible to everybody, while badly posed questions or answers reduce the overall quality of the public knowledge database.
+In short, a high quality question or answer is *precise*, *concise*, *relevant*, *easy-to-understand*, *accesible*, and *well-formated/well-posed*. For more information, please have a look through the [How to write a good issue](#how-to-write-a-good-issue) section.

-### Did you find a bug?
+**NOTE about channels**:
+[*The forum*](https://discuss.huggingface.co/c/discussion-related-to-httpsgithubcomhuggingfacediffusers/63) is much better indexed by search engines, such as Google. Posts are ranked by popularity rather than chronologically. Hence, it's easier to look up questions and answers that we posted some time ago.
+In addition, questions and answers posted in the forum can easily be linked to.
+In contrast, *Discord* has a chat-like format that invites fast back-and-forth communication.
+While it will most likely take less time for you to get an answer to your question on Discord, your
+question won't be visible anymore over time. Also, it's much harder to find information that was posted a while back on Discord. We therefore strongly recommend using the forum for high-quality questions and answers in an attempt to create long-lasting knowledge for the community. If discussions on Discord lead to very interesting answers and conclusions, we recommend posting the results on the forum to make the information more available for future readers.
+
+### 2. Opening new issues on the GitHub issues tab

 The 🧨 Diffusers library is robust and reliable thanks to the users who notify us of
 the problems they encounter. So thank you for reporting an issue.

-First, we would really appreciate it if you could **make sure the bug was not
-already reported** (use the search bar on GitHub under Issues).
+Remember, GitHub issues are reserved for technical questions directly related to the Diffusers library, bug reports, feature requests, or feedback on the library design.

-### Do you want to implement a new diffusion pipeline / diffusion model?
+In a nutshell, this means that everything that is **not** related to the **code of the Diffusers library** (including the documentation) should **not** be asked on GitHub, but rather on either the [forum](https://discuss.huggingface.co/c/discussion-related-to-httpsgithubcomhuggingfacediffusers/63) or [Discord](https://discord.gg/G7tWnz98XR).

-Awesome! Please provide the following information:
+**Please consider the following guidelines when opening a new issue**:
+- Make sure you have searched whether your issue has already been asked before (use the search bar on GitHub under Issues).
+- Please never report a new issue on another (related) issue. If another issue is highly related, please
+open a new issue nevertheless and link to the related issue.
+- Make sure your issue is written in English. Please use one of the great, free online translation services, such as [DeepL](https://www.deepl.com/translator) to translate from your native language to English if you are not comfortable in English.
+- Check whether your issue might be solved by updating to the newest Diffusers version. Before posting your issue, please make sure that `python -c "import diffusers; print(diffusers.__version__)"` is higher or matches the latest Diffusers version.
+- Remember that the more effort you put into opening a new issue, the higher the quality of your answer will be and the better the overall quality of the Diffusers issues.

-* Short description of the diffusion pipeline and link to the paper;
-* Link to the implementation if it is open-source;
-* Link to the model weights if they are available.
+New issues usually include the following.

-If you are willing to contribute the model yourself, let us know so we can best
-guide you.
+#### 2.1. Reproducible, minimal bug reports.

-### Do you want a new feature (that is not a model)?
+A bug report should always have a reproducible code snippet and be as minimal and concise as possible.
+This means in more detail:
+- Narrow the bug down as much as you can, **do not just dump your whole code file**
+- Format your code
+- Do not include any external libraries except for Diffusers depending on them.
+- **Always** provide all necessary information about your environment; for this, you can run: `diffusers-cli env` in your shell and copy-paste the displayed information to the issue.
+- Explain the issue. If the reader doesn't know what the issue is and why it is an issue, she cannot solve it.
+- **Always** make sure the reader can reproduce your issue with as little effort as possible. If your code snippet cannot be run because of missing libraries or undefined variables, the reader cannot help you. Make sure your reproducible code snippet is as minimal as possible and can be copy-pasted into a simple Python shell.
+- If in order to reproduce your issue a model and/or dataset is required, make sure the reader has access to that model or dataset. You can always upload your model or dataset to the [Hub](https://huggingface.co) to make it easily downloadable. Try to keep your model and dataset as small as possible, to make the reproduction of your issue as effortless as possible.
+
+For more information, please have a look through the [How to write a good issue](#how-to-write-a-good-issue) section.
+
+You can open a bug report [here](https://github.com/huggingface/diffusers/issues/new/choose).
+
+#### 2.2. Feature requests.

 A world-class feature request addresses the following points:

 1. Motivation first:
-  * Is it related to a problem/frustration with the library? If so, please explain
-    why. Providing a code snippet that demonstrates the problem is best.
-  * Is it related to something you would need for a project? We'd love to hear
-    about it!
-  * Is it something you worked on and think could benefit the community?
-    Awesome! Tell us what problem it solved for you.
+* Is it related to a problem/frustration with the library? If so, please explain
+why. Providing a code snippet that demonstrates the problem is best.
+* Is it related to something you would need for a project? We'd love to hear
+about it!
+* Is it something you worked on and think could benefit the community?
+Awesome! Tell us what problem it solved for you.
 2. Write a *full paragraph* describing the feature;
 3. Provide a **code snippet** that demonstrates its future use;
 4. In case this is related to a paper, please attach a link;
 5. Attach any additional information (drawings, screenshots, etc.) you think may help.

-If your issue is well written we're already 80% of the way there by the time you
-post it.
+You can open a feature request [here](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=&template=feature_request.md&title=).

-## Start contributing! (Pull Requests)
+#### 2.3 Feedback. 
+
+Feedback about the library design and why it is good or not good helps the core maintainers immensely to build a user-friendly library. To understand the philosophy behind the current design philosophy, please have a look [here](https://huggingface.co/docs/diffusers/conceptual/philosophy). If you feel like a certain design choice does not fit with the current design philosophy, please explain why and how it should be changed. If a certain design choice follows the design philosophy too much, hence restricting use cases, explain why and how it should be changed.
+If a certain design choice is very useful for you, please also leave a note as this is great feedback for future design decisions.
+
+You can open an issue about feedback [here](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=&template=feedback.md&title=).
+
+#### 2.4 Technical questions. 
+
+Technical questions are mainly about why certain code of the library was written in a certain way, or what a certain part of the code does. Please make sure to link to the code in question and please provide detail on
+why this part of the code is difficult to understand.
+
+You can open an issue about a technical question [here](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=bug&template=bug-report.yml).
+
+#### 2.5 Proposal to add a new model, scheduler, or pipeline.
+
+If the diffusion model community released a new model, pipeline, or scheduler that you would like to see in the Diffusers library, please provide the following information:
+
+* Short description of the diffusion pipeline, model, or scheduler and link to the paper or public release.
+* Link to any of its open-source implementation.
+* Link to the model weights if they are available.
+
+If you are willing to contribute to the model yourself, let us know so we can best guide you. Also, don't forget
+to tag the original author of the component (model, scheduler, pipeline, etc.) by GitHub handle if you can find it.
+
+You can open a request for a model/pipeline/scheduler [here](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=New+model%2Fpipeline%2Fscheduler&template=new-model-addition.yml).
+
+### 3. Answering issues on the GitHub issues tab
+
+Answering issues on GitHub might require some technical knowledge of Diffusers, but we encourage everybody to give it a try even if you are not 100% certain that your answer is correct.
+Some tips to give a high-quality answer to an issue:
+- Be as concise and minimal as possible
+- Stay on topic. An answer to the issue should concern the issue and only the issue.
+- Provide links to code, papers, or other sources that prove or encourage your point.
+- Answer in code. If a simple code snippet is the answer to the issue or shows how the issue can be solved, please provide a fully reproducible code snippet.
+
+Also, many issues tend to be simply off-topic, duplicates of other issues, or irrelevant. It is of great
+help to the maintainers if you can answer such issues, encouraging the author of the issue to be
+more precise, provide the link to a duplicated issue or redirect them to [the forum](https://discuss.huggingface.co/c/discussion-related-to-httpsgithubcomhuggingfacediffusers/63) or [Discord](https://discord.gg/G7tWnz98XR)
+
+If you have verified that the issued bug report is correct and requires a correction in the source code,
+please have a look at the next sections.
+
+For all of the following contributions, you will need to open a PR. It is explained in detail how to do so in the [Opening a pull requst](#how-to-open-a-pr) section.
+
+### 4. Fixing a `Good first issue`
+
+*Good first issues* are marked by the [Good first issue](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22good+first+issue%22) label. Usually, the issue already
+explains how a potential solution should look so that it is easier to fix.
+If the issue hasn't been closed and you would like to try to fix this issue, you can just leave a message "I would like to try this issue.". There are usually three scenarios:
+- a.) The issue description already proposes a fix. In this case and if the solution makes sense to you, you can open a PR or draft PR to fix it.
+- b.) The issue description does not propose a fix. In this case, you can ask what a proposed fix could look like and someone from the Diffusers team should answer shortly. If you have a good idea of how to fix it, feel free to directly open a PR.
+- c.) There is already an open PR to fix the issue, but the issue hasn't been closed yet. If the PR has gone stale, you can simply open a new PR and link to the stale PR. PRs often go stale if the original contributor who wanted to fix the issue suddenly cannot find the time anymore to proceed. This often happens in open-source and is very normal. In this case, the community will be very happy if you give it a new try and leverage the knowledge of the existing PR. If there is already a PR and it is active, you can help the author by giving suggestions, reviewing the PR or even asking whether you can contribute to the PR.
+
+
+### 5. Contribute to the documentation
+
+A good library **always** has good documentation! The official documentation is often one of the first points of contact for new users of the library, and therefore contributing to the documentation is a **highly
+valuable contribution**.
+
+Contributing to the library can have many forms:
+
+- Correcting spelling or grammatical errors.
+- Correct incorrect formatting of the docstring. If you see that the official documentation is weirdly displayed or a link is broken, we are very happy if you take some time to correct it.
+- Correct the shape or dimensions of a docstring input or output tensor.
+- Clarify documentation that is hard to understand or incorrect.
+- Update outdated code examples.
+- Translating the documentation to another language.
+
+Anything displayed on [the official Diffusers doc page](https://huggingface.co/docs/diffusers/index) is part of the official documentation and can be corrected, adjusted in the respective [documentation source](https://github.com/huggingface/diffusers/tree/main/docs/source).
+
+Please have a look at [this page](https://github.com/huggingface/diffusers/tree/main/docs) on how to verify changes made to the documentation locally.
+
+
+### 6. Contribute a community pipeline
+
+[Pipelines](https://huggingface.co/docs/diffusers/api/pipelines/overview) are usually the first point of contact between the Diffusers library and the user.
+Pipelines are examples of how to use Diffusers [models](https://huggingface.co/docs/diffusers/api/models) and [schedulers](https://huggingface.co/docs/diffusers/api/schedulers/overview).
+We support two types of pipelines:
+
+- Official Pipelines
+- Community Pipelines
+
+Both official and community pipelines follow the same design and consist of the same type of components.
+
+Official pipelines are tested and maintained by the core maintainers of Diffusers. Their code
+resides in [src/diffusers/pipelines](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines).
+In contrast, community pipelines are contributed and maintained purely by the **community** and are **not** tested.
+They reside in [examples/community](https://github.com/huggingface/diffusers/tree/main/examples/community) and while they can be accessed via the [PyPI diffusers package](https://pypi.org/project/diffusers/), their code is not part of the PyPI distribution.
+
+The reason for the distinction is that the core maintainers of the Diffusers library cannot maintain and test all
+possible ways diffusion models can be used for inference, but some of them may be of interest to the community.
+Officially released diffusion pipelines,
+such as Stable Diffusion are added to the core src/diffusers/pipelines package which ensures
+high quality of maintenance, no backward-breaking code changes, and testing.
+More bleeding edge pipelines should be added as community pipelines. If usage for a community pipeline is high, the pipeline can be moved to the official pipelines upon request from the community. This is one of the ways we strive to be a community-driven library.
+
+To add a community pipeline, one should add a <name-of-the-community>.py file to [examples/community](https://github.com/huggingface/diffusers/tree/main/examples/community) and adapt the [examples/community/README.md](https://github.com/huggingface/diffusers/tree/main/examples/community/README.md) to include an example of the new pipeline.
+
+An example can be seen [here](https://github.com/huggingface/diffusers/pull/2400).
+
+Community pipeline PRs are only checked at a superficial level and ideally they should be maintained by their original authors.
+
+Contributing a community pipeline is a great way to understand how Diffusers models and schedulers work. Having contributed a community pipeline is usually the first stepping stone to contributing an official pipeline to the
+core package.
+
+### 7. Contribute to training examples
+
+Diffusers examples are a collection of training scripts that reside in [examples](https://github.com/huggingface/diffusers/tree/main/examples).
+
+We support two types of training examples:
+
+- Official training examples
+- Research training examples
+
+Research training examples are located in [examples/research_projects](https://github.com/huggingface/diffusers/tree/main/examples/research_projects) whereas official training examples include all folders under [examples](https://github.com/huggingface/diffusers/tree/main/examples) except the `research_projects` and `community` folders.
+The official training examples are maintained by the Diffusers' core maintainers whereas the research training examples are maintained by the community.
+This is because of the same reasons put forward in [6. Contribute a community pipeline](#contribute-a-community-pipeline) for official pipelines vs. community pipelines: It is not feasible for the core maintainers to maintain all possible training methods for diffusion models.
+If the Diffusers core maintainers and the community consider a certain training paradigm to be too experimental or not popular enough, the corresponding training code should be put in the `research_projects` folder and maintained by the author.
+
+Both official training and research examples consist of a directory that contains one or more training scripts, a requirements.txt file, and a README.md file. In order for the user to make use of the
+training examples, it is required to clone the repository:
+
+```
+git clone https://github.com/huggingface/diffusers
+```
+
+as well as to install all additional dependencies required for training:
+
+```
+pip install -r /examples/<your-example-folder>/requirements.txt
+```
+
+Therefore when adding an example, the `requirements.txt` file shall define all pip dependencies required for your training example so that once all those are installed, the user can run the example's training script. See, for example, the [DreamBooth `requirements.txt` file](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/requirements.txt).
+
+Training examples of the Diffusers library should adhere to the following philosophy:
+- All the code necessary to run the examples should be found in a single Python file
+- One should be able to run the example from the command line with `python <your-example>.py --args`
+- Examples should be kept simple and serve as **an example** on how to use Diffusers for training. The purpose of example scripts is **not** to create state-of-the-art diffusion models, but rather to reproduce known training schemes without adding too much custom logic. As a byproduct of this point, our examples also strive to serve as good educational materials.
+
+To contribute an example, it is highly recommended to look at already existing examples such as [dreambooth](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth.py) to get an idea of how they should look like.
+We strongly advise contributors to make use of the [Accelerate library](https://github.com/huggingface/accelerate) as it's tightly integrated
+with Diffusers.
+Once an example script works, please make sure to add a comprehensive `README.md` that states how to use the example exactly. This README should include:
+- An example command on how to run the example script as shown [here e.g.](https://github.com/huggingface/diffusers/tree/main/examples/dreambooth#running-locally-with-pytorch).
+- A link to some training results (logs, models, ...) that show what the user can expect as shown [here e.g.](https://api.wandb.ai/report/patrickvonplaten/xm6cd5q5).
+- If you are adding a non-official/research training example, **please don't forget** to add a sentence that you are maintaining this training example which includes your git handle as shown [here](https://github.com/huggingface/diffusers/tree/main/examples/research_projects/intel_opts#diffusers-examples-with-intel-optimizations).
+
+If you are contributing to the official training examples, please also make sure to add a test to [examples/test_examples.py](https://github.com/huggingface/diffusers/blob/main/examples/test_examples.py). This is not necessary for non-official training examples.
+
+### 8. Fixing a `Good second issue`
+
+*Good second issues* are marked by the [Good second issue](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22Good+second+issue%22) label. Good second issues are
+usually more complicated to solve than [Good first issues](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22good+first+issue%22).
+The issue description usually gives less guidance on how to fix the issue and requires
+a decent understanding of the library by the interested contributor.
+If you are interested in tackling a second good issue, feel free to open a PR to fix it and link the PR to the issue. If you see that a PR has already been opened for this issue but did not get merged, have a look to understand why it wasn't merged and try to open an improved PR.
+Good second issues are usually more difficult to get merged compared to good first issues, so don't hesitate to ask for help from the core maintainers. If your PR is almost finished the core maintainers can also jump into your PR and commit to it in order to get it merged.
+
+### 9. Adding pipelines, models, schedulers
+
+Pipelines, models, and schedulers are the most important pieces of the Diffusers library.
+They provide easy access to state-of-the-art diffusion technologies and thus allow the community to
+build powerful generative AI applications.
+
+By adding a new model, pipeline, or scheduler you might enable a new powerful use case for any of the user interfaces relying on Diffusers which can be of immense value for the whole generative AI ecosystem.
+
+Diffusers has a couple of open feature requests for all three components - feel free to gloss over them
+if you don't know yet what specific component you would like to add:
+- [Model or pipeline](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+pipeline%2Fmodel%22)
+- [Scheduler](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22New+scheduler%22)
+
+Before adding any of the three components, it is strongly recommended that you give the [Philosophy guide](https://github.com/huggingface/diffusers/issues?q=is%3Aopen+is%3Aissue+label%3A%22Good+second+issue%22) a read to better understand the design of any of the three components. Please be aware that
+we cannot merge model, scheduler, or pipeline additions that strongly diverge from our design philosophy
+as it will lead to API inconsistencies. If you fundamentally disagree with a design choice, please
+open a [Feedback issue](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=&template=feedback.md&title=) instead so that it can be discussed whether a certain design
+pattern/design choice shall be changed everywhere in the library and whether we shall update our design philosophy. Consistency across the library is very important for us.
+
+Please make sure to add links to the original codebase/paper to the PR and ideally also ping the
+original author directly on the PR so that they can follow the progress and potentially help with questions.
+
+If you are unsure or stuck in the PR, don't hesitate to leave a message to ask for a first review or help.
+
+## How to write a good issue
+
+**The better your issue is written, the higher the chances that it will be quickly resolved.**
+
+1. Make sure that you've used the correct template for your issue. You can pick between *Bug Report*, *Feature Request*, *Feedback about API Design*, *New model/pipeline/scheduler addition*, *Forum*, or a blank issue. Make sure to pick the correct one when opening [a new issue](https://github.com/huggingface/diffusers/issues/new/choose).
+2. **Be precise**: Give your issue a fitting title. Try to formulate your issue description as simple as possible. The more precise you are when submitting an issue, the less time it takes to understand the issue and potentially solve it. Make sure to open an issue for one issue only and not for multiple issues. If you found multiple issues, simply open multiple issues. If your issue is a bug, try to be as precise as possible about what bug it is - you should not just write "Error in diffusers".
+3. **Reproducibility**: No reproducible code snippet == no solution. If you encounter a bug, maintainers **have to be able to reproduce** it. Make sure that you include a code snippet that can be copy-pasted into a Python interpreter to reproduce the issue. Make sure that your code snippet works, *i.e.* that there are no missing imports or missing links to images, ... Your issue should contain an error message **and** a code snippet that can be copy-pasted without any changes to reproduce the exact same error message. If your issue is using local model weights or local data that cannot be accessed by the reader, the issue cannot be solved. If you cannot share your data or model, try to make a dummy model or dummy data.
+4. **Minimalistic**: Try to help the reader as much as you can to understand the issue as quickly as possible by staying as concise as possible. Remove all code / all information that is irrelevant to the issue. If you have found a bug, try to create the easiest code example you can to demonstrate your issue, do not just dump your whole workflow into the issue as soon as you have found a bug. E.g., if you train a model and get an error at some point during the training, you should first try to understand what part of the training code is responsible for the error and try to reproduce it with a couple of lines. Try to use dummy data instead of full datasets.
+5. Add links. If you are referring to a certain naming, method, or model make sure to provide a link so that the reader can better understand what you mean. If you are referring to a specific PR or issue, make sure to link it to your issue. Do not assume that the reader knows what you are talking about. The more links you add to your issue the better.
+6. Formatting. Make sure to nicely format your issue by formatting code into Python code syntax, and error messages into normal code syntax. See the [official GitHub formatting docs](https://docs.github.com/en/get-started/writing-on-github/getting-started-with-writing-and-formatting-on-github/basic-writing-and-formatting-syntax) for more information.
+7. Think of your issue not as a ticket to be solved, but rather as a beautiful entry to a well-written encyclopedia. Every added issue is a contribution to publicly available knowledge. By adding a nicely written issue you not only make it easier for maintainers to solve your issue, but you are helping the whole community to better understand a certain aspect of the library.
+
+## How to write a good PR
+
+1. Be a chameleon. Understand existing design patterns and syntax and make sure your code additions flow seamlessly into the existing code base. Pull requests that significantly diverge from existing design patterns or user interfaces will not be merged.
+2. Be laser focused. A pull request should solve one problem and one problem only. Make sure to not fall into the trap of "also fixing another problem while we're adding it". It is much more difficult to review pull requests that solve multiple, unrelated problems at once.
+3. If helpful, try to add a code snippet that displays an example of how your addition can be used.
+4. The title of your pull request should be a summary of its contribution.
+5. If your pull request addresses an issue, please mention the issue number in
+the pull request description to make sure they are linked (and people
+consulting the issue know you are working on it);
+6. To indicate a work in progress please prefix the title with `[WIP]`. These
+are useful to avoid duplicated work, and to differentiate it from PRs ready
+to be merged;
+7. Try to formulate and format your text as explained in [How to write a good issue](#how-to-write-a-good-issue).
+8. Make sure existing tests pass;
+9. Add high-coverage tests. No quality testing = no merge.
+- If you are adding new `@slow` tests, make sure they pass using
+`RUN_SLOW=1 python -m pytest tests/test_my_new_model.py`.
+CircleCI does not run the slow tests, but GitHub actions does every night!
+10. All public methods must have informative docstrings that work nicely with markdown. See `[pipeline_latent_diffusion.py](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/latent_diffusion/pipeline_latent_diffusion.py)` for an example.
+11. Due to the rapidly growing repository, it is important to make sure that no files that would significantly weigh down the repository are added. This includes images, videos, and other non-text files. We prefer to leverage a hf.co hosted `dataset` like
+[`hf-internal-testing`](https://huggingface.co/hf-internal-testing) or [huggingface/documentation-images](https://huggingface.co/datasets/huggingface/documentation-images) to place these files.
+If an external contribution, feel free to add the images to your PR and ask a Hugging Face member to migrate your images
+to this dataset.
+
+## How to open a PR

 Before writing code, we strongly advise you to search through the existing PRs or
 issues to make sure that nobody is already working on the same thing. If you are
@@ -99,144 +355,98 @@ You will need basic `git` proficiency to be able to contribute to
 manual. Type `git --help` in a shell and enjoy. If you prefer books, [Pro
 Git](https://git-scm.com/book/en/v2) is a very good reference.

-Follow these steps to start contributing ([supported Python versions](https://github.com/huggingface/diffusers/blob/main/setup.py#L212)):
+Follow these steps to start contributing ([supported Python versions](https://github.com/huggingface/diffusers/blob/main/setup.py#L244)):

 1. Fork the [repository](https://github.com/huggingface/diffusers) by
-   clicking on the 'Fork' button on the repository's page. This creates a copy of the code
-   under your GitHub user account.
+clicking on the 'Fork' button on the repository's page. This creates a copy of the code
+under your GitHub user account.

 2. Clone your fork to your local disk, and add the base repository as a remote:

-   ```bash
-   $ git clone git@github.com:<your Github handle>/diffusers.git
-   $ cd diffusers
-   $ git remote add upstream https://github.com/huggingface/diffusers.git
-   ```
+ ```bash
+ $ git clone git@github.com:<your Github handle>/diffusers.git
+ $ cd diffusers
+ $ git remote add upstream https://github.com/huggingface/diffusers.git
+ ```

 3. Create a new branch to hold your development changes:

-   ```bash
-   $ git checkout -b a-descriptive-name-for-my-changes
-   ```
+ ```bash
+ $ git checkout -b a-descriptive-name-for-my-changes
+ ```

-   **Do not** work on the `main` branch.
+**Do not** work on the `main` branch.

 4. Set up a development environment by running the following command in a virtual environment:

-   ```bash
-   $ pip install -e ".[dev]"
-   ```
+ ```bash
+ $ pip install -e ".[dev]"
+ ```

-   (If Diffusers was already installed in the virtual environment, remove
-   it with `pip uninstall diffusers` before reinstalling it in editable
-   mode with the `-e` flag.)
-
-   To run the full test suite, you might need the additional dependency on `transformers` and `datasets` which requires a separate source
-   install:
-
-   ```bash
-   $ git clone https://github.com/huggingface/transformers
-   $ cd transformers
-   $ pip install -e .
-   ```
-
-   ```bash
-   $ git clone https://github.com/huggingface/datasets
-   $ cd datasets
-   $ pip install -e .
-   ```
-
-   If you have already cloned that repo, you might need to `git pull` to get the most recent changes in the `datasets`
-   library.
+If you have already cloned the repo, you might need to `git pull` to get the most recent changes in the
+library.

 5. Develop the features on your branch.

-   As you work on the features, you should make sure that the test suite
-   passes. You should run the tests impacted by your changes like this:
+As you work on the features, you should make sure that the test suite
+passes. You should run the tests impacted by your changes like this:

-   ```bash
-   $ pytest tests/<TEST_TO_RUN>.py
-   ```
+ ```bash
+ $ pytest tests/<TEST_TO_RUN>.py
+ ```

-   You can also run the full suite with the following command, but it takes
-   a beefy machine to produce a result in a decent amount of time now that
-   Diffusers has grown a lot. Here is the command for it:
+You can also run the full suite with the following command, but it takes
+a beefy machine to produce a result in a decent amount of time now that
+Diffusers has grown a lot. Here is the command for it:

-   ```bash
-   $ make test
-   ```
+ ```bash
+ $ make test
+ ```

-   For more information about tests, check out the
-   [dedicated documentation](https://huggingface.co/docs/diffusers/testing)
+🧨 Diffusers relies on `black` and `isort` to format its source code
+consistently. After you make changes, apply automatic style corrections and code verifications
+that can't be automated in one go with:

-   🧨 Diffusers relies on `black` and `isort` to format its source code
-   consistently. After you make changes, apply automatic style corrections and code verifications
-   that can't be automated in one go with:
+ ```bash
+ $ make style
+ ```

-   ```bash
-   $ make style
-   ```
+🧨 Diffusers also uses `ruff` and a few custom scripts to check for coding mistakes. Quality
+control runs in CI, however, you can also run the same checks with:

-   🧨 Diffusers also uses `ruff` and a few custom scripts to check for coding mistakes. Quality
-   control runs in CI, however you can also run the same checks with:
+ ```bash
+ $ make quality
+ ```

-   ```bash
-   $ make quality
-   ```
+Once you're happy with your changes, add changed files using `git add` and
+make a commit with `git commit` to record your changes locally:

-   Once you're happy with your changes, add changed files using `git add` and
-   make a commit with `git commit` to record your changes locally:
+ ```bash
+ $ git add modified_file.py
+ $ git commit
+ ```

-   ```bash
-   $ git add modified_file.py
-   $ git commit
-   ```
+It is a good idea to sync your copy of the code with the original
+repository regularly. This way you can quickly account for changes:

-   It is a good idea to sync your copy of the code with the original
-   repository regularly. This way you can quickly account for changes:
+ ```bash
+ $ git pull upstream main
+ ```

-   ```bash
-   $ git fetch upstream
-   $ git rebase upstream/main
-   ```
+Push the changes to your account using:

-   Push the changes to your account using:
+ ```bash
+ $ git push -u origin a-descriptive-name-for-my-changes
+ ```

-   ```bash
-   $ git push -u origin a-descriptive-name-for-my-changes
-   ```
-
-6. Once you are satisfied (**and the checklist below is happy too**), go to the
-   webpage of your fork on GitHub. Click on 'Pull request' to send your changes
-   to the project maintainers for review.
+6. Once you are satisfied, go to the
+webpage of your fork on GitHub. Click on 'Pull request' to send your changes
+to the project maintainers for review.

 7. It's ok if maintainers ask you for changes. It happens to core contributors
-   too! So everyone can see the changes in the Pull request, work in your local
-   branch and push the changes to your fork. They will automatically appear in
-   the pull request.
-
-
-### Checklist
-
-1. The title of your pull request should be a summary of its contribution;
-2. If your pull request addresses an issue, please mention the issue number in
-   the pull request description to make sure they are linked (and people
-   consulting the issue know you are working on it);
-3. To indicate a work in progress please prefix the title with `[WIP]`. These
-   are useful to avoid duplicated work, and to differentiate it from PRs ready
-   to be merged;
-4. Make sure existing tests pass;
-5. Add high-coverage tests. No quality testing = no merge.
-   - If you are adding new `@slow` tests, make sure they pass using
-     `RUN_SLOW=1 python -m pytest tests/test_my_new_model.py`.
-   - If you are adding a new tokenizer, write tests, and make sure
-     `RUN_SLOW=1 python -m pytest tests/test_tokenization_{your_model_name}.py` passes.
-   CircleCI does not run the slow tests, but GitHub actions does every night!
-6. All public methods must have informative docstrings that work nicely with sphinx. See `[pipeline_latent_diffusion.py](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/latent_diffusion/pipeline_latent_diffusion.py)` for an example.
-7. Due to the rapidly growing repository, it is important to make sure that no files that would significantly weigh down the repository are added. This includes images, videos and other non-text files. We prefer to leverage a hf.co hosted `dataset` like
-   the ones hosted on [`hf-internal-testing`](https://huggingface.co/hf-internal-testing) in which to place these files and reference or [huggingface/documentation-images](https://huggingface.co/datasets/huggingface/documentation-images).
-   If an external contribution, feel free to add the images to your PR and ask a Hugging Face member to migrate your images
-   to this dataset.
+too! So everyone can see the changes in the Pull request, work in your local
+branch and push the changes to your fork. They will automatically appear in
+the pull request.

 ### Tests

@@ -286,6 +496,3 @@ $ git push --set-upstream origin your-branch-for-syncing
 ### Style guide

 For documentation strings, 🧨 Diffusers follows the [google style](https://google.github.io/styleguide/pyguide.html).
-
-
-**This guide was heavily inspired by the awesome [scikit-learn guide to contributing](https://github.com/scikit-learn/scikit-learn/blob/main/CONTRIBUTING.md).**
@@ -44,6 +44,8 @@ The team works daily to make the technical and non-technical tools available to

  - [**Safe Stable Diffusion**](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion_safe): It mitigates the well-known issue that models, like Stable Diffusion, that are trained on unfiltered, web-crawled datasets tend to suffer from inappropriate degeneration. Related paper: [Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models](https://arxiv.org/abs/2211.05105).

+  - [**Safety Checker**](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/safety_checker.py): It checks and compares the class probability of a set of hard-coded harmful concepts in the embedding space against an image after it has been generated. The harmful concepts are intentionally hidden to prevent reverse engineering of the checker.
+
 - **Staged released on the Hub**: in particularly sensitive situations, access to some repositories should be restricted. This staged release is an intermediary step that allows the repository’s authors to have more control over its use.

 - **Licensing**: [OpenRAILs](https://huggingface.co/blog/open_rail), a new type of licensing, allow us to ensure free access while having a set of restrictions that ensure more responsible use. 
@@ -310,7 +310,7 @@ for idx in range(len(dataset)):
    edited_images.append(edited_image)
 ```

-To measure the directional similarity, we first load CLIP's image and text encoders.
+To measure the directional similarity, we first load CLIP's image and text encoders:

 ```python
 from transformers import (
@@ -329,7 +329,7 @@ image_encoder = CLIPVisionModelWithProjection.from_pretrained(clip_id).to(device

 Notice that we are using a particular CLIP checkpoint, i.e., `openai/clip-vit-large-patch14`. This is because the Stable Diffusion pre-training was performed with this CLIP variant. For more details, refer to the [documentation](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/pix2pix#diffusers.StableDiffusionInstructPix2PixPipeline.text_encoder).

-Next, we prepare a PyTorch `nn.module` to compute directional similarity:
+Next, we prepare a PyTorch `nn.Module` to compute directional similarity:

 ```python
 import torch.nn as nn
@@ -410,7 +410,7 @@ It should be noted that the `StableDiffusionInstructPix2PixPipeline` exposes t

 We can extend the idea of this metric to measure how similar the original image and edited version are. To do that, we can just do `F.cosine_similarity(img_feat_two, img_feat_one)`. For these kinds of edits, we would still want the primary semantics of the images to be preserved as much as possible, i.e., a high similarity score.

-We can use these metrics for similar pipelines such as the[`StableDiffusionPix2PixZeroPipeline`](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/pix2pix_zero#diffusers.StableDiffusionPix2PixZeroPipeline)`.
+We can use these metrics for similar pipelines such as the [`StableDiffusionPix2PixZeroPipeline`](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/pix2pix_zero#diffusers.StableDiffusionPix2PixZeroPipeline).

 <Tip>

@@ -550,7 +550,7 @@ FID results tend to be fragile as they depend on a lot of factors:
 * The image format (not the same if we start from PNGs vs JPGs).

 Keeping that in mind, FID is often most useful when comparing similar runs, but it is 
-hard to to reproduce paper results unless the authors carefully disclose the FID 
+hard to reproduce paper results unless the authors carefully disclose the FID 
 measurement code.

 These points apply to other related metrics too, such as KID and IS. 
@@ -60,17 +60,17 @@ Let's walk through more in-detail design decisions for each class.

 ### Pipelines

-Pipelines are designed to be easy to use (therefore do not follow [*Simple over easy*](#simple-over-easy) 100%)), are not feature complete, and should loosely be seen as examples of how to use [models](#models) and [schedulers](#schedulers) for inference.
+Pipelines are designed to be easy to use (therefore do not follow [*Simple over easy*](#simple-over-easy) 100%), are not feature complete, and should loosely be seen as examples of how to use [models](#models) and [schedulers](#schedulers) for inference.

 The following design principles are followed:
 - Pipelines follow the single-file policy. All pipelines can be found in individual directories under src/diffusers/pipelines. One pipeline folder corresponds to one diffusion paper/project/release. Multiple pipeline files can be gathered in one pipeline folder, as it’s done for [`src/diffusers/pipelines/stable-diffusion`](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines/stable_diffusion). If pipelines share similar functionality, one can make use of the [#Copied from mechanism](https://github.com/huggingface/diffusers/blob/125d783076e5bd9785beb05367a2d2566843a271/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py#L251).
- Pipelines all inherit from [`DiffusionPipeline`]
+- Pipelines all inherit from [`DiffusionPipeline`].
 - Every pipeline consists of different model and scheduler components, that are documented in the [`model_index.json` file](https://huggingface.co/runwayml/stable-diffusion-v1-5/blob/main/model_index.json), are accessible under the same name as attributes of the pipeline and can be shared between pipelines with [`DiffusionPipeline.components`](https://huggingface.co/docs/diffusers/main/en/api/diffusion_pipeline#diffusers.DiffusionPipeline.components) function.
 - Every pipeline should be loadable via the [`DiffusionPipeline.from_pretrained`](https://huggingface.co/docs/diffusers/main/en/api/diffusion_pipeline#diffusers.DiffusionPipeline.from_pretrained) function.
 - Pipelines should be used **only** for inference.
 - Pipelines should be very readable, self-explanatory, and easy to tweak.
 - Pipelines should be designed to build on top of each other and be easy to integrate into higher-level APIs.
- Pipelines are **not** intended to be feature-complete user interfaces. For future complete user interfaces one should rather have a look at [InvokeAI](https://github.com/invoke-ai/InvokeAI), [Diffuzers](https://github.com/abhishekkrthakur/diffuzers), and [lama-cleaner](https://github.com/Sanster/lama-cleaner)
+- Pipelines are **not** intended to be feature-complete user interfaces. For future complete user interfaces one should rather have a look at [InvokeAI](https://github.com/invoke-ai/InvokeAI), [Diffuzers](https://github.com/abhishekkrthakur/diffuzers), and [lama-cleaner](https://github.com/Sanster/lama-cleaner).
 - Every pipeline should have one and only one way to run it via a `__call__` method. The naming of the `__call__` arguments should be shared across all pipelines.
 - Pipelines should be named after the task they are intended to solve.
 - In almost all cases, novel diffusion pipelines shall be implemented in a new pipeline folder/file.
@@ -104,7 +104,7 @@ The following design principles are followed:
 - Schedulers all inherit from `SchedulerMixin` and `ConfigMixin`.
 - Schedulers can be easily swapped out with the [`ConfigMixin.from_config`](https://huggingface.co/docs/diffusers/main/en/api/configuration#diffusers.ConfigMixin.from_config) method as explained in detail [here](./using-diffusers/schedulers.mdx).
 - Every scheduler has to have a `set_num_inference_steps`, and a `step` function. `set_num_inference_steps(...)` has to be called before every denoising process, *i.e.* before `step(...)` is called.
- Every scheduler exposes the timesteps to be "looped over" via a `timesteps` attribute, which is an array of timesteps the model will be called upon
+- Every scheduler exposes the timesteps to be "looped over" via a `timesteps` attribute, which is an array of timesteps the model will be called upon.
 - The `step(...)` function takes a predicted model output and the "current" sample (x_t) and returns the "previous", slightly more denoised sample (x_t-1).
 - Given the complexity of diffusion schedulers, the `step` function does not expose all the complexity and can be a bit of a "black box".
 - In almost all cases, novel schedulers shall be implemented in a new scheduling file.
@@ -73,9 +73,10 @@ The library has three main components:
 | [stable_diffusion_pix2pix](./api/pipelines/stable_diffusion/pix2pix) | [InstructPix2Pix: Learning to Follow Image Editing Instructions](https://arxiv.org/abs/2211.09800)  | Text-Guided Image Editing|
 | [stable_diffusion_pix2pix_zero](./api/pipelines/stable_diffusion/pix2pix_zero) | [Zero-shot Image-to-Image Translation](https://pix2pixzero.github.io/) | Text-Guided Image Editing |
 | [stable_diffusion_attend_and_excite](./api/pipelines/stable_diffusion/attend_and_excite) | [Attend-and-Excite: Attention-Based Semantic Guidance for Text-to-Image Diffusion Models](https://arxiv.org/abs/2301.13826) | Text-to-Image Generation |
-| [stable_diffusion_self_attention_guidance](./api/pipelines/stable_diffusion/self_attention_guidance) | [Improving Sample Quality of Diffusion Models Using Self-Attention Guidance](https://arxiv.org/abs/2210.00939) | Text-to-Image Generation |
+| [stable_diffusion_self_attention_guidance](./api/pipelines/stable_diffusion/self_attention_guidance) | [Improving Sample Quality of Diffusion Models Using Self-Attention Guidance](https://arxiv.org/abs/2210.00939) | Text-to-Image Generation Unconditional Image Generation |
 | [stable_diffusion_image_variation](./stable_diffusion/image_variation) | [Stable Diffusion Image Variations](https://github.com/LambdaLabsML/lambda-diffusers#stable-diffusion-image-variations) | Image-to-Image Generation |
 | [stable_diffusion_latent_upscale](./stable_diffusion/latent_upscale) | [Stable Diffusion Latent Upscaler](https://twitter.com/StabilityAI/status/1590531958815064065) | Text-Guided Super Resolution Image-to-Image |
+| [stable_diffusion_model_editing](./api/pipelines/stable_diffusion/model_editing) | [Editing Implicit Assumptions in Text-to-Image Diffusion Models](https://time-diffusion.github.io/) | Text-to-Image Model Editing |
 | [stable_diffusion_2](./api/pipelines/stable_diffusion_2) | [Stable Diffusion 2](https://stability.ai/blog/stable-diffusion-v2-release) | Text-to-Image Generation |
 | [stable_diffusion_2](./api/pipelines/stable_diffusion_2) | [Stable Diffusion 2](https://stability.ai/blog/stable-diffusion-v2-release) | Text-Guided Image Inpainting |
 | [stable_diffusion_2](./api/pipelines/stable_diffusion_2) | [Depth-Conditional Stable Diffusion](https://github.com/Stability-AI/stablediffusion#depth-conditional-stable-diffusion) | Depth-to-Image Generation |
@@ -84,8 +85,9 @@ The library has three main components:
 | [stable_unclip](./stable_unclip) | Stable unCLIP | Text-to-Image Generation |
 | [stable_unclip](./stable_unclip) | Stable unCLIP | Image-to-Image Text-Guided Generation |
 | [stochastic_karras_ve](./api/pipelines/stochastic_karras_ve) | [Elucidating the Design Space of Diffusion-Based Generative Models](https://arxiv.org/abs/2206.00364) | Unconditional Image Generation |
+| [text_to_video_sd](./api/pipelines/text_to_video) | [Modelscope's Text-to-video-synthesis Model in Open Domain](https://modelscope.cn/models/damo/text-to-video-synthesis/summary) | Text-to-Video Generation |
 | [unclip](./api/pipelines/unclip) | [Hierarchical Text-Conditional Image Generation with CLIP Latents](https://arxiv.org/abs/2204.06125)(implementation by [kakaobrain](https://github.com/kakaobrain/karlo)) | Text-to-Image Generation |
 | [versatile_diffusion](./api/pipelines/versatile_diffusion) | [Versatile Diffusion: Text, Images and Variations All in One Diffusion Model](https://arxiv.org/abs/2211.08332) | Text-to-Image Generation |
 | [versatile_diffusion](./api/pipelines/versatile_diffusion) | [Versatile Diffusion: Text, Images and Variations All in One Diffusion Model](https://arxiv.org/abs/2211.08332) | Image Variations Generation |
 | [versatile_diffusion](./api/pipelines/versatile_diffusion) | [Versatile Diffusion: Text, Images and Variations All in One Diffusion Model](https://arxiv.org/abs/2211.08332) | Dual Image and Text Guided Generation |
-| [vq_diffusion](./api/pipelines/vq_diffusion) | [Vector Quantized Diffusion Model for Text-to-Image Synthesis](https://arxiv.org/abs/2111.14822) | Text-to-Image Generation |
+| [vq_diffusion](./api/pipelines/vq_diffusion) | [Vector Quantized Diffusion Model for Text-to-Image Synthesis](https://arxiv.org/abs/2111.14822) | Text-to-Image Generation |
@@ -0,0 +1,167 @@
+<!--Copyright 2023 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
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# How to run Stable Diffusion with Core ML
+
+[Core ML](https://developer.apple.com/documentation/coreml) is the model format and machine learning library supported by Apple frameworks. If you are interested in running Stable Diffusion models inside your macOS or iOS/iPadOS apps, this guide will show you how to convert existing PyTorch checkpoints into the Core ML format and use them for inference with Python or Swift.
+
+Core ML models can leverage all the compute engines available in Apple devices: the CPU, the GPU, and the Apple Neural Engine (or ANE, a tensor-optimized accelerator available in Apple Silicon Macs and modern iPhones/iPads). Depending on the model and the device it's running on, Core ML can mix and match compute engines too, so some portions of the model may run on the CPU while others run on GPU, for example.
+
+<Tip>
+
+You can also run the `diffusers` Python codebase on Apple Silicon Macs using the `mps` accelerator built into PyTorch. This approach is explained in depth in [the mps guide](mps), but it is not compatible with native apps.
+
+</Tip>
+
+## Stable Diffusion Core ML Checkpoints
+
+Stable Diffusion weights (or checkpoints) are stored in the PyTorch format, so you need to convert them to the Core ML format before we can use them inside native apps.
+
+Thankfully, Apple engineers developed [a conversion tool](https://github.com/apple/ml-stable-diffusion#-converting-models-to-core-ml) based on `diffusers` to convert the PyTorch checkpoints to Core ML.
+
+Before you convert a model, though, take a moment to explore the Hugging Face Hub – chances are the model you're interested in is already available in Core ML format:
+
+- the [Apple](https://huggingface.co/apple) organization includes Stable Diffusion versions 1.4, 1.5, 2.0 base, and 2.1 base
+- [coreml](https://huggingface.co/coreml) organization includes custom DreamBoothed and finetuned models
+- use this [filter](https://huggingface.co/models?pipeline_tag=text-to-image&library=coreml&p=2&sort=likes) to return all available Core ML checkpoints
+
+If you can't find the model you're interested in, we recommend you follow the instructions for [Converting Models to Core ML](https://github.com/apple/ml-stable-diffusion#-converting-models-to-core-ml) by Apple.
+
+## Selecting the Core ML Variant to Use
+
+Stable Diffusion models can be converted to different Core ML variants intended for different purposes:
+
+- The type of attention blocks used. The attention operation is used to "pay attention" to the relationship between different areas in the image representations and to understand how the image and text representations are related. Attention is compute- and memory-intensive, so different implementations exist that consider the hardware characteristics of different devices. For Core ML Stable Diffusion models, there are two attention variants:
+    * `split_einsum` ([introduced by Apple](https://machinelearning.apple.com/research/neural-engine-transformers)) is optimized for ANE devices, which is available in modern iPhones, iPads and M-series computers.
+    * The "original" attention (the base implementation used in `diffusers`) is only compatible with CPU/GPU and not ANE. It can be *faster* to run your model on CPU + GPU using `original` attention than ANE. See [this performance benchmark](https://huggingface.co/blog/fast-mac-diffusers#performance-benchmarks) as well as some [additional measures provided by the community](https://github.com/huggingface/swift-coreml-diffusers/issues/31) for additional details.
+
+- The supported inference framework.
+    * `packages` are suitable for Python inference. This can be used to test converted Core ML models before attempting to integrate them inside native apps, or if you want to explore Core ML performance but don't need to support native apps. For example, an application with a web UI could perfectly use a Python Core ML backend.
+    * `compiled` models are required for Swift code. The `compiled` models in the Hub split the large UNet model weights into several files for compatibility with iOS and iPadOS devices. This corresponds to the [`--chunk-unet` conversion option](https://github.com/apple/ml-stable-diffusion#-converting-models-to-core-ml). If you want to support native apps, then you need to select the `compiled` variant.
+
+The official Core ML Stable Diffusion [models](https://huggingface.co/apple/coreml-stable-diffusion-v1-4/tree/main) include these variants, but the community ones may vary:
+
+```
+coreml-stable-diffusion-v1-4
+├── README.md
+├── original
+│   ├── compiled
+│   └── packages
+└── split_einsum
+    ├── compiled
+    └── packages
+```
+
+You can download and use the variant you need as shown below.
+
+## Core ML Inference in Python
+
+Install the following libraries to run Core ML inference in Python:
+
+```bash
+pip install huggingface_hub
+pip install git+https://github.com/apple/ml-stable-diffusion
+```
+
+### Download the Model Checkpoints
+
+To run inference in Python, use one of the versions stored in the `packages` folders because the `compiled` ones are only compatible with Swift. You may choose whether you want to use `original` or `split_einsum` attention.
+
+This is how you'd download the `original` attention variant from the Hub to a directory called `models`:
+
+```Python
+from huggingface_hub import snapshot_download
+from pathlib import Path
+
+repo_id = "apple/coreml-stable-diffusion-v1-4"
+variant = "original/packages"
+
+model_path = Path("./models") / (repo_id.split("/")[-1] + "_" + variant.replace("/", "_"))
+snapshot_download(repo_id, allow_patterns=f"{variant}/*", local_dir=model_path, local_dir_use_symlinks=False)
+print(f"Model downloaded at {model_path}")
+```
+
+
+### Inference[[python-inference]]
+
+Once you have downloaded a snapshot of the model, you can test it using Apple's Python script.
+
+```shell
+python -m python_coreml_stable_diffusion.pipeline --prompt "a photo of an astronaut riding a horse on mars" -i models/coreml-stable-diffusion-v1-4_original_packages -o </path/to/output/image> --compute-unit CPU_AND_GPU --seed 93
+```
+
+`<output-mlpackages-directory>` should point to the checkpoint you downloaded in the step above, and `--compute-unit` indicates the hardware you want to allow for inference. It must be one of the following options: `ALL`, `CPU_AND_GPU`, `CPU_ONLY`, `CPU_AND_NE`. You may also provide an optional output path, and a seed for reproducibility.
+
+The inference script assumes you're using the original version of the Stable Diffusion model, `CompVis/stable-diffusion-v1-4`. If you use another model, you *have* to specify its Hub id in the inference command line, using the `--model-version` option. This works for models already supported and custom models you trained or fine-tuned yourself.
+
+For example, if you want to use [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5):
+
+```shell
+python -m python_coreml_stable_diffusion.pipeline --prompt "a photo of an astronaut riding a horse on mars" --compute-unit ALL -o output --seed 93 -i models/coreml-stable-diffusion-v1-5_original_packages --model-version runwayml/stable-diffusion-v1-5
+```
+
+
+## Core ML inference in Swift
+
+Running inference in Swift is slightly faster than in Python because the models are already compiled in the `mlmodelc` format. This is noticeable on app startup when the model is loaded but shouldn’t be noticeable if you run several generations afterward.
+
+### Download
+
+To run inference in Swift on your Mac, you need one of the `compiled` checkpoint versions. We recommend you download them locally using Python code similar to the previous example, but with one of the `compiled` variants:
+
+```Python
+from huggingface_hub import snapshot_download
+from pathlib import Path
+
+repo_id = "apple/coreml-stable-diffusion-v1-4"
+variant = "original/compiled"
+
+model_path = Path("./models") / (repo_id.split("/")[-1] + "_" + variant.replace("/", "_"))
+snapshot_download(repo_id, allow_patterns=f"{variant}/*", local_dir=model_path, local_dir_use_symlinks=False)
+print(f"Model downloaded at {model_path}")
+```
+
+### Inference[[swift-inference]]
+
+To run inference, please clone Apple's repo:
+
+```bash
+git clone https://github.com/apple/ml-stable-diffusion
+cd ml-stable-diffusion
+```
+
+And then use Apple's command line tool, [Swift Package Manager](https://www.swift.org/package-manager/#):
+
+```bash
+swift run StableDiffusionSample --resource-path models/coreml-stable-diffusion-v1-4_original_compiled --compute-units all "a photo of an astronaut riding a horse on mars"
+```
+
+You have to specify in `--resource-path` one of the checkpoints downloaded in the previous step, so please make sure it contains compiled Core ML bundles with the extension `.mlmodelc`. The `--compute-units` has to be one of these values: `all`, `cpuOnly`, `cpuAndGPU`, `cpuAndNeuralEngine`.
+
+For more details, please refer to the [instructions in Apple's repo](https://github.com/apple/ml-stable-diffusion).
+
+
+## Supported Diffusers Features
+
+The Core ML models and inference code don't support many of the features, options, and flexibility of 🧨 Diffusers. These are some of the limitations to keep in mind:
+
+- Core ML models are only suitable for inference. They can't be used for training or fine-tuning.
+- Only two schedulers have been ported to Swift, the default one used by Stable Diffusion and `DPMSolverMultistepScheduler`, which we ported to Swift from our `diffusers` implementation. We recommend you use `DPMSolverMultistepScheduler`, since it produces the same quality in about half the steps.
+- Negative prompts, classifier-free guidance scale, and image-to-image tasks are available in the inference code. Advanced features such as depth guidance, ControlNet, and latent upscalers are not available yet.
+
+Apple's [conversion and inference repo](https://github.com/apple/ml-stable-diffusion) and our own [swift-coreml-diffusers](https://github.com/huggingface/swift-coreml-diffusers) repos are intended as technology demonstrators to enable other developers to build upon. 
+
+If you feel strongly about any missing features, please feel free to open a feature request or, better yet, a contribution PR :)
+
+## Native Diffusers Swift app
+
+One easy way to run Stable Diffusion on your own Apple hardware is to use [our open-source Swift repo](https://github.com/huggingface/swift-coreml-diffusers), based on `diffusers` and Apple's conversion and inference repo. You can study the code, compile it with [Xcode](https://developer.apple.com/xcode/) and adapt it for your own needs. For your convenience, there's also a [standalone Mac app in the App Store](https://apps.apple.com/app/diffusers/id1666309574), so you can play with it without having to deal with the code or IDE. If you are a developer and have determined that Core ML is the best solution to build your Stable Diffusion app, then you can use the rest of this guide to get started with your project. We can't wait to see what you'll build :)
@@ -19,7 +19,6 @@ We'll discuss how the following settings impact performance and memory.
 |                  | Latency | Speedup |
 | ---------------- | ------- | ------- |
 | original         | 9.50s   | x1      |
-| cuDNN auto-tuner | 9.37s   | x1.01   |
 | fp16             | 3.61s   | x2.63   |
 | channels last    | 3.30s   | x2.88   |
 | traced UNet      | 3.21s   | x2.96   |
@@ -31,18 +30,6 @@ We'll discuss how the following settings impact performance and memory.
  steps.
 </em>

-## Enable cuDNN auto-tuner
-
-[NVIDIA cuDNN](https://developer.nvidia.com/cudnn) supports many algorithms to compute a convolution. Autotuner runs a short benchmark and selects the kernel with the best performance on a given hardware for a given input size.
-
-Since we’re using **convolutional networks** (other types currently not supported), we can enable cuDNN autotuner before launching the inference by setting:
-
-```python
-import torch
-
-torch.backends.cudnn.benchmark = True
-```
-
 ### Use tf32 instead of fp32 (on Ampere and later CUDA devices)

 On Ampere and later CUDA devices matrix multiplications and convolutions can use the TensorFloat32 (TF32) mode for faster but slightly less accurate computations. By default PyTorch enables TF32 mode for convolutions but not matrix multiplications, and unless a network requires full float32 precision we recommend enabling this setting for matrix multiplications, too. It can significantly speed up computations with typically negligible loss of numerical accuracy. You can read more about it [here](https://huggingface.co/docs/transformers/v4.18.0/en/performance#tf32). All you need to do is to add this before your inference:
@@ -58,7 +45,10 @@ torch.backends.cuda.matmul.allow_tf32 = True
 To save more GPU memory and get more speed, you can load and run the model weights directly in half precision. This involves loading the float16 version of the weights, which was saved to a branch named `fp16`, and telling PyTorch to use the `float16` type when loading them:

 ```Python
-pipe = StableDiffusionPipeline.from_pretrained(
+import torch
+from diffusers import DiffusionPipeline
+
+pipe = DiffusionPipeline.from_pretrained(
    "runwayml/stable-diffusion-v1-5",
    
    torch_dtype=torch.float16,
@@ -85,13 +75,13 @@ For even additional memory savings, you can use a sliced version of attention th
  each head which can save a significant amount of memory.
 </Tip>

-To perform the attention computation sequentially over each head, you only need to invoke [`~StableDiffusionPipeline.enable_attention_slicing`] in your pipeline before inference, like here:
+To perform the attention computation sequentially over each head, you only need to invoke [`~DiffusionPipeline.enable_attention_slicing`] in your pipeline before inference, like here:

 ```Python
 import torch
-from diffusers import StableDiffusionPipeline
+from diffusers import DiffusionPipeline

-pipe = StableDiffusionPipeline.from_pretrained(
+pipe = DiffusionPipeline.from_pretrained(
    "runwayml/stable-diffusion-v1-5",
    
    torch_dtype=torch.float16,
@@ -221,7 +211,7 @@ image = pipe(prompt).images[0]
 Full-model offloading is an alternative that moves whole models to the GPU, instead of handling each model's constituent _modules_. This results in a negligible impact on inference time (compared with moving the pipeline to `cuda`), while still providing some memory savings.

 In this scenario, only one of the main components of the pipeline (typically: text encoder, unet and vae)
-will be in the GPU while the others wait in the CPU. Compoments like the UNet that run for multiple iterations will stay on GPU until they are no longer needed.
+will be in the GPU while the others wait in the CPU. Components like the UNet that run for multiple iterations will stay on GPU until they are no longer needed.

 This feature can be enabled by invoking `enable_model_cpu_offload()` on the pipeline, as shown below.

@@ -415,10 +405,10 @@ To leverage it just make sure you have:
 - Cuda available
 - [Installed the xformers library](xformers).
 ```python
-from diffusers import StableDiffusionPipeline
+from diffusers import DiffusionPipeline
 import torch

-pipe = StableDiffusionPipeline.from_pretrained(
+pipe = DiffusionPipeline.from_pretrained(
    "runwayml/stable-diffusion-v1-5",
    torch_dtype=torch.float16,
 ).to("cuda")
@@ -16,8 +16,8 @@ specific language governing permissions and limitations under the License.

 ## Requirements

- Optimum Habana 1.3 or later, [here](https://huggingface.co/docs/optimum/habana/installation) is how to install it.
- SynapseAI 1.7.
+- Optimum Habana 1.4 or later, [here](https://huggingface.co/docs/optimum/habana/installation) is how to install it.
+- SynapseAI 1.8.


 ## Inference Pipeline
@@ -62,9 +62,9 @@ For more information, check out Optimum Habana's [documentation](https://hugging

 ## Benchmark

-Here are the latencies for Habana Gaudi 1 and Gaudi 2 with the [Habana/stable-diffusion](https://huggingface.co/Habana/stable-diffusion) Gaudi configuration (mixed precision bf16/fp32):
+Here are the latencies for Habana first-generation Gaudi and Gaudi2 with the [Habana/stable-diffusion](https://huggingface.co/Habana/stable-diffusion) Gaudi configuration (mixed precision bf16/fp32):

-|         | Latency | Batch size |
-| ------- |:-------:|:----------:|
-| Gaudi 1 | 4.37s   | 4/8        |
-| Gaudi 2 | 1.19s   | 4/8        |
+|                        | Latency (batch size = 1) | Throughput (batch size = 8) |
+| ---------------------- |:------------------------:|:---------------------------:|
+| first-generation Gaudi | 4.29s                    | 0.283 images/s              |
+| Gaudi2                 | 1.54s                    | 0.904 images/s              |
@@ -19,20 +19,25 @@ specific language governing permissions and limitations under the License.
 - Mac computer with Apple silicon (M1/M2) hardware.
 - macOS 12.6 or later (13.0 or later recommended).
 - arm64 version of Python.
- PyTorch 1.13. You can install it with `pip` or `conda` using the instructions in https://pytorch.org/get-started/locally/.
+- PyTorch 2.0 (recommended) or 1.13 (minimum version supported for `mps`). You can install it with `pip` or `conda` using the instructions in https://pytorch.org/get-started/locally/.


 ## Inference Pipeline

 The snippet below demonstrates how to use the `mps` backend using the familiar `to()` interface to move the Stable Diffusion pipeline to your M1 or M2 device.

-We recommend to "prime" the pipeline using an additional one-time pass through it. This is a temporary workaround for a weird issue we have detected: the first inference pass produces slightly different results than subsequent ones. You only need to do this pass once, and it's ok to use just one inference step and discard the result.
+<Tip warning={true}>
+
+**If you are using PyTorch 1.13** you need to "prime" the pipeline using an additional one-time pass through it. This is a temporary workaround for a weird issue we detected: the first inference pass produces slightly different results than subsequent ones. You only need to do this pass once, and it's ok to use just one inference step and discard the result.
+
+</Tip>
+
+We strongly recommend you use PyTorch 2 or better, as it solves a number of problems like the one described in the previous tip.

 ```python
-# make sure you're logged in with `huggingface-cli login`
-from diffusers import StableDiffusionPipeline
+from diffusers import DiffusionPipeline

-pipe = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5")
+pipe = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5")
 pipe = pipe.to("mps")

 # Recommended if your computer has < 64 GB of RAM
@@ -40,7 +45,7 @@ pipe.enable_attention_slicing()

 prompt = "a photo of an astronaut riding a horse on mars"

-# First-time "warmup" pass (see explanation above)
+# First-time "warmup" pass if PyTorch version is 1.13 (see explanation above)
 _ = pipe(prompt, num_inference_steps=1)

 # Results match those from the CPU device after the warmup pass.
@@ -51,7 +56,7 @@ image = pipe(prompt).images[0]

 M1/M2 performance is very sensitive to memory pressure. The system will automatically swap if it needs to, but performance will degrade significantly when it does.

-We recommend you use _attention slicing_ to reduce memory pressure during inference and prevent swapping, particularly if your computer has lass than 64 GB of system RAM, or if you generate images at non-standard resolutions larger than 512 × 512 pixels. Attention slicing performs the costly attention operation in multiple steps instead of all at once. It usually has a performance impact of ~20% in computers without universal memory, but we have observed _better performance_ in most Apple Silicon computers, unless you have 64 GB or more.
+We recommend you use _attention slicing_ to reduce memory pressure during inference and prevent swapping, particularly if your computer has less than 64 GB of system RAM, or if you generate images at non-standard resolutions larger than 512 × 512 pixels. Attention slicing performs the costly attention operation in multiple steps instead of all at once. It usually has a performance impact of ~20% in computers without universal memory, but we have observed _better performance_ in most Apple Silicon computers, unless you have 64 GB or more.

 ```python
 pipeline.enable_attention_slicing()
@@ -59,5 +64,4 @@ pipeline.enable_attention_slicing()

 ## Known Issues

- As mentioned above, we are investigating a strange [first-time inference issue](https://github.com/huggingface/diffusers/issues/372).
 - Generating multiple prompts in a batch [crashes or doesn't work reliably](https://github.com/huggingface/diffusers/issues/363). We believe this is related to the [`mps` backend in PyTorch](https://github.com/pytorch/pytorch/issues/84039). This is being resolved, but for now we recommend to iterate instead of batching.
@@ -13,61 +13,53 @@ specific language governing permissions and limitations under the License.

 # How to use the ONNX Runtime for inference

-🤗 Diffusers provides a Stable Diffusion pipeline compatible with the ONNX Runtime. This allows you to run Stable Diffusion on any hardware that supports ONNX (including CPUs), and where an accelerated version of PyTorch is not available.
+🤗 [Optimum](https://github.com/huggingface/optimum) provides a Stable Diffusion pipeline compatible with ONNX Runtime. 

 ## Installation

- TODO
+Install 🤗 Optimum with the following command for ONNX Runtime support:
+
+```
+pip install optimum["onnxruntime"]
+```

 ## Stable Diffusion Inference

-The snippet below demonstrates how to use the ONNX runtime. You need to use `OnnxStableDiffusionPipeline` instead of `StableDiffusionPipeline`. You also need to download the weights from the `onnx` branch of the repository, and indicate the runtime provider you want to use.
+To load an ONNX model and run inference with the ONNX Runtime, you need to replace [`StableDiffusionPipeline`] with `ORTStableDiffusionPipeline`. In case you want to load
+a PyTorch model and convert it to the ONNX format on-the-fly, you can set `export=True`.

 ```python
-# make sure you're logged in with `huggingface-cli login`
-from diffusers import OnnxStableDiffusionPipeline
-
-pipe = OnnxStableDiffusionPipeline.from_pretrained(
-    "runwayml/stable-diffusion-v1-5",
-    revision="onnx",
-    provider="CUDAExecutionProvider",
-)
+from optimum.onnxruntime import ORTStableDiffusionPipeline

+model_id = "runwayml/stable-diffusion-v1-5"
+pipe = ORTStableDiffusionPipeline.from_pretrained(model_id, export=True)
 prompt = "a photo of an astronaut riding a horse on mars"
-image = pipe(prompt).images[0]
+images = pipe(prompt).images[0]
+pipe.save_pretrained("./onnx-stable-diffusion-v1-5")
 ```

-The snippet below demonstrates how to use the ONNX runtime with the Stable Diffusion upscaling pipeline.
+If you want to export the pipeline in the ONNX format offline and later use it for inference,
+you can use the [`optimum-cli export`](https://huggingface.co/docs/optimum/main/en/exporters/onnx/usage_guides/export_a_model#exporting-a-model-to-onnx-using-the-cli) command: 

-```python
-from diffusers import OnnxStableDiffusionPipeline, OnnxStableDiffusionUpscalePipeline
-
-prompt = "a photo of an astronaut riding a horse on mars"
-steps = 50
-
-txt2img = OnnxStableDiffusionPipeline.from_pretrained(
-    "runwayml/stable-diffusion-v1-5",
-    revision="onnx",
-    provider="CUDAExecutionProvider",
-)
-small_image = txt2img(
-    prompt,
-    num_inference_steps=steps,
-).images[0]
-
-generator = torch.manual_seed(0)
-upscale = OnnxStableDiffusionUpscalePipeline.from_pretrained(
-    "ssube/stable-diffusion-x4-upscaler-onnx",
-    provider="CUDAExecutionProvider",
-)
-large_image = upscale(
-    prompt,
-    small_image,
-    generator=generator,
-    num_inference_steps=steps,
-).images[0]
+```bash
+optimum-cli export onnx --model runwayml/stable-diffusion-v1-5 sd_v15_onnx/
 ```

+Then perform inference:
+
+```python 
+from optimum.onnxruntime import ORTStableDiffusionPipeline
+
+model_id = "sd_v15_onnx"
+pipe = ORTStableDiffusionPipeline.from_pretrained(model_id)
+prompt = "a photo of an astronaut riding a horse on mars"
+images = pipe(prompt).images[0]
+```
+
+Notice that we didn't have to specify `export=True` above.
+
+You can find more examples in [optimum documentation](https://huggingface.co/docs/optimum/).
+
 ## Known Issues

 - Generating multiple prompts in a batch seems to take too much memory. While we look into it, you may need to iterate instead of batching.
@@ -36,4 +36,4 @@ prompt = "a photo of an astronaut riding a horse on mars"
 images = pipe(prompt).images[0]
 ```

-You can find more examples in [optimum documentation](https://huggingface.co/docs/optimum/intel/inference#export-and-inference-of-stable-diffusion-models).
+You can find more examples (such as static reshaping and model compilation) in [optimum documentation](https://huggingface.co/docs/optimum/intel/inference#export-and-inference-of-stable-diffusion-models).
@@ -18,11 +18,10 @@ Starting from version `0.13.0`, Diffusers supports the latest optimization from


 ## Installation
-To benefit from the accelerated transformers implementation and `torch.compile`, we will need to install the nightly version of PyTorch, as the stable version is yet to be released. The first step is to install CUDA 11.7 or CUDA 11.8, 
-as PyTorch 2.0 does not support the previous versions. Once CUDA is installed, torch nightly can be installed using:
+To benefit from the accelerated attention implementation and `torch.compile`, you just need to install the latest versions of PyTorch 2.0 from `pip`, and make sure you are on diffusers 0.13.0 or later. As explained below, `diffusers` automatically uses the attention optimizations (but not `torch.compile`) when available.

 ```bash
-pip install --pre torch torchvision --index-url https://download.pytorch.org/whl/nightly/cu117
+pip install --upgrade torch torchvision diffusers
 ```

 ## Using accelerated transformers and torch.compile.
@@ -36,9 +35,9 @@ pip install --pre torch torchvision --index-url https://download.pytorch.org/whl

    ```Python
    import torch
-    from diffusers import StableDiffusionPipeline
+    from diffusers import DiffusionPipeline

-    pipe = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16)
+    pipe = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16)
    pipe = pipe.to("cuda")

    prompt = "a photo of an astronaut riding a horse on mars"
@@ -49,10 +48,10 @@ pip install --pre torch torchvision --index-url https://download.pytorch.org/whl

    ```Python
    import torch
-    from diffusers import StableDiffusionPipeline
+    from diffusers import DiffusionPipeline
    from diffusers.models.attention_processor import AttnProcessor2_0

-    pipe = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16).to("cuda")
+    pipe = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16).to("cuda")
    pipe.unet.set_attn_processor(AttnProcessor2_0())

    prompt = "a photo of an astronaut riding a horse on mars"
@@ -69,11 +68,9 @@ pip install --pre torch torchvision --index-url https://download.pytorch.org/whl

    ```python
    import torch
-    from diffusers import StableDiffusionPipeline
+    from diffusers import DiffusionPipeline

-    pipe = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16).to(
-        "cuda"
-    )
+    pipe = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16).to("cuda")
    pipe.unet = torch.compile(pipe.unet)

    batch_size = 10
@@ -89,10 +86,9 @@ pip install --pre torch torchvision --index-url https://download.pytorch.org/whl
 ## Benchmark

 We conducted a simple benchmark on different GPUs to compare vanilla attention, xFormers, `torch.nn.functional.scaled_dot_product_attention` and `torch.compile+torch.nn.functional.scaled_dot_product_attention`.
-For the benchmark we used the the [stable-diffusion-v1-4](https://huggingface.co/CompVis/stable-diffusion-v1-4) model with 50 steps. The `xFormers` benchmark is done using the `torch==1.13.1` version, while the accelerated transformers optimizations are tested using nightly versions of PyTorch 2.0. The tables below summarize the results we got.
-
-The `Speed over xformers` columns denote the speed-up gained over `xFormers` using the `torch.compile+torch.nn.functional.scaled_dot_product_attention`.
+For the benchmark we used the [stable-diffusion-v1-4](https://huggingface.co/CompVis/stable-diffusion-v1-4) model with 50 steps. The `xFormers` benchmark is done using the `torch==1.13.1` version, while the accelerated transformers optimizations are tested using nightly versions of PyTorch 2.0. The tables below summarize the results we got.

+Please refer to [our featured blog post in the PyTorch site](https://pytorch.org/blog/accelerated-diffusers-pt-20/) for more details.

 ### FP16 benchmark

@@ -103,10 +99,14 @@ ___The time reported is in seconds.___

 | GPU | Batch Size | Vanilla Attention | xFormers | PyTorch2.0 SDPA | SDPA + torch.compile | Speed over xformers (%) |
 | --- | --- | --- | --- | --- | --- | --- |
-| A100 | 10 | 12.02 | 8.7 | 8.79 | 7.89 | 9.31 |
-| A100 | 16 | 18.95 | 13.57 | 13.67 | 12.25 | 9.73 |
-| A100 | 32 (1) | OOM | 26.56 | 26.68 | 24.08 | 9.34 |
-| A100 | 64 | | 52.51 | 53.03 | 47.81 | 8.95 |
+| A100 | 1 | 2.69 | 2.7 | 1.98 | 2.47 | 8.52 |
+| A100 | 2 | 3.21 | 3.04 | 2.38 | 2.78 | 8.55 |
+| A100 | 4 | 5.27 | 3.91 | 3.89 | 3.53 | 9.72 |
+| A100 | 8 | 9.74 | 7.03 | 7.04 | 6.62 | 5.83 |
+| A100 | 10 | 12.02 | 8.7 | 8.67 | 8.45 | 2.87 |
+| A100 | 16 | 18.95 | 13.57 | 13.55 | 13.20 | 2.73 |
+| A100 | 32 (1) | OOM | 26.56 | 26.68 | 25.85 | 2.67 |
+| A100 | 64 | | 52.51 | 53.03 | 50.93 | 3.01 |
 | | | | | | | |
 | A10 | 4 | 13.94 | 9.81 | 10.01 | 9.35 | 4.69 |
 | A10 | 8 | 27.09 | 19 | 19.53 | 18.33 | 3.53 |
@@ -125,25 +125,28 @@ ___The time reported is in seconds.___
 | V100 | 10 | OOM | 19.52 | 19.28 | 18.18 | 6.86 |
 | V100 | 16 | OOM | 30.29 | 29.84 | 28.22 | 6.83 |
 | | | | | | | |
-| 3090 | 4 | 10.04 | 7.82 | 7.89 | 7.47 | 4.48 |
-| 3090 | 8 | 19.27 | 14.97 | 15.04 | 14.22 | 5.01 |
-| 3090 | 10| 24.08 | 18.7 | 18.7 | 17.69 | 5.40 |
-| 3090 | 16 | OOM | 29.06 | 29.06 | 28.2 | 2.96 |
-| 3090 | 32 (1) | | 58.05 | 58 | 54.88 | 5.46 |
-| 3090 | 64 (1) | | 126.54 | 126.03 | 117.33 | 7.28 |
+| 3090 | 1 | 2.94 | 2.5 | 2.42 | 2.33 | 6.80 |
+| 3090 | 4 | 10.04 | 7.82 | 7.72 | 7.38 | 5.63 |
+| 3090 | 8 | 19.27 | 14.97 | 14.88 | 14.15 | 5.48 |
+| 3090 | 10| 24.08 | 18.7 | 18.62 | 18.12 | 3.10 |
+| 3090 | 16 | OOM | 29.06 | 28.88 | 28.2 | 2.96 |
+| 3090 | 32 (1) | | 58.05 | 57.42 | 56.28 | 3.05 |
+| 3090 | 64 (1) | | 126.54 | 114.27 | 112.21 | 11.32 |
 | | | | | | | |
-| 3090 Ti | 4 | 9.07 | 7.14 | 7.15 | 6.81 | 4.62 |
-| 3090 Ti | 8 | 17.51 | 13.65 | 13.72 | 12.99 | 4.84 |
-| 3090 Ti | 10 (2) | 21.79 | 16.85 | 16.93 | 16.02 | 4.93 |
-| 3090 Ti | 16 | OOM | 26.1 | 26.28 | 25.46 | 2.45 |
-| 3090 Ti | 32 (1) | | 51.78 | 52.04 | 49.15 | 5.08 |
-| 3090 Ti | 64 (1) | | 112.02 | 112.33 | 103.91 | 7.24 |
+| 3090 Ti | 1 | 2.7 | 2.26 | 2.19 | 2.12 | 6.19 |
+| 3090 Ti | 4 | 9.07 | 7.14 | 7.00 | 6.71 | 6.02 |
+| 3090 Ti | 8 | 17.51 | 13.65 | 13.53 | 12.94 | 5.20 |
+| 3090 Ti | 10 (2) | 21.79 | 16.85 | 16.77 | 16.44 | 2.43 |
+| 3090 Ti | 16 | OOM | 26.1 | 26.04 | 25.53 | 2.18 |
+| 3090 Ti | 32 (1) | | 51.78 | 51.71 | 50.91 | 1.68 |
+| 3090 Ti | 64 (1) | | 112.02 | 102.78 | 100.89 | 9.94 |
 | | | | | | | |
-| 4090 | 4 | 10.48 | 8.37 | 8.32 | 8.01 | 4.30 |
-| 4090 | 8 | 14.33 | 10.22 | 10.42 | 9.78 | 4.31 |
-| 4090 | 16 | | 17.07 | 17.46 | 17.15 | -0.47 |
-| 4090 | 32 (1) | | 39.03 | 39.86 | 37.97 | 2.72 |
-| 4090 | 64 (1) | | 77.29 | 79.44 | 77.67 | -0.49 |
+| 4090 | 1 | 4.47 | 3.98 | 1.28 | 1.21 | 69.60 |
+| 4090 | 4 | 10.48 | 8.37 | 3.76 | 3.56 | 57.47 |
+| 4090 | 8 | 14.33 | 10.22 | 7.43 | 6.99 | 31.60 |
+| 4090 | 16 | | 17.07 | 14.98 | 14.58 | 14.59 |
+| 4090 | 32 (1) | | 39.03 | 30.18 | 29.49 | 24.44 |
+| 4090 | 64 (1) | | 77.29 | 61.34 | 59.96 | 22.42 |


 				
@@ -155,11 +158,13 @@ Using `torch.compile` in addition to the accelerated transformers implementation

 | GPU | Batch Size | Vanilla Attention | xFormers | PyTorch2.0 SDPA | SDPA + torch.compile | Speed over xformers (%) | Speed over vanilla (%) |
 | --- | --- | --- | --- | --- | --- | --- | --- |
-| A100 | 4 | 16.56 | 12.42 | 12.2 | 11.84 | 4.67 | 28.50 |
-| A100 | 10 | OOM | 29.93 | 29.44 | 28.5 | 4.78 | |
-| A100 | 16 | | 47.08 | 46.27 | 44.8 | 4.84 | |
-| A100 | 32 | | 92.89 | 91.34 | 88.35 | 4.89 | |
-| A100 | 64 | | 185.3 | 182.71 | 176.48 | 4.76 | |
+| A100 | 1 | 4.97 | 3.86 | 2.6 | 2.86 | 25.91 | 42.45 |
+| A100 | 2 | 9.03 | 6.76 | 4.41 | 4.21 | 37.72 | 53.38 |
+| A100 | 4 | 16.70 | 12.42 | 7.94 | 7.54 | 39.29 | 54.85 |
+| A100 | 10 | OOM | 29.93 | 18.70 | 18.46 | 38.32 | |
+| A100 | 16 | | 47.08 | 29.41 | 29.04 | 38.32 | |
+| A100 | 32 | | 92.89 | 57.55 | 56.67 | 38.99 | |
+| A100 | 64 | | 185.3 | 114.8 | 112.98 | 39.03 | |
 | | | | | | | |
 | A10 | 1 | 10.59 | 8.81 | 7.51 | 7.35 | 16.57 | 30.59 |
 | A10 | 4 | 34.77 | 27.63 | 22.77 | 22.07 | 20.12 | 36.53 |
@@ -179,30 +184,27 @@ Using `torch.compile` in addition to the accelerated transformers implementation
 | V100 | 8 | | 43.95 | 43.37 | 42.25 | 3.87 | |
 | V100 | 16 | | 84.99 | 84.73 | 82.55 | 2.87 | |
 | | | | | | | |
-| 3090 | 1 | 7.09 | 6.78 | 6.11 | 6.03 | 11.06 | 14.95 |
-| 3090 | 4 | 22.69 | 21.45 | 18.67 | 18.09 | 15.66 | 20.27 |
-| 3090 | 8 | | 42.59 | 36.75 | 35.59 | 16.44 | |
-| 3090 | 16 | | 85.35 | 72.37 | 70.25 | 17.69 | |
-| 3090 | 32 (1) | | 162.05 | 138.99 | 134.53 | 16.98 | |
-| 3090 | 48 | | 241.91 | 207.75 | | 14.12 | |
+| 3090 | 1 | 7.09 | 6.78 | 5.34 | 5.35 | 21.09 | 24.54 |
+| 3090 | 4 | 22.69 | 21.45 | 18.56 | 18.18 | 15.24 | 19.88 |
+| 3090 | 8 | | 42.59 | 36.68 | 35.61 | 16.39 | |
+| 3090 | 16 | | 85.35 | 72.93 | 70.18 | 17.77 | |
+| 3090 | 32 (1) | | 162.05 | 143.46 | 138.67 | 14.43 | |
 | | | | | | | |
-| 3090 Ti | 1 | 6.45 | 6.19 | 5.64 | 5.49 | 11.31 | 14.88 |
-| 3090 Ti | 4 | 20.32 | 19.31 | 16.9 | 16.37 | 15.23 | 19.44 |
-| 3090 Ti | 8 (2) | | 37.93 | 33.05 | 31.99 | 15.66 | |
-| 3090 Ti | 16 | | 75.37 | 65.25 | 64.32 | 14.66 | |
-| 3090 Ti | 32 (1) | | 142.55 | 124.44 | 120.74 | 15.30 | |
-| 3090 Ti | 48 | | 213.19 | 186.55 | | 12.50 | |
+| 3090 Ti | 1 | 6.45 | 6.19 | 4.99 | 4.89 | 21.00 | 24.19 |
+| 3090 Ti | 4 | 20.32 | 19.31 | 17.02 | 16.48 | 14.66 | 18.90 |
+| 3090 Ti | 8 | | 37.93 | 33.21 | 32.24 | 15.00 | |
+| 3090 Ti | 16 | | 75.37 | 66.63 | 64.5 | 14.42 | |
+| 3090 Ti | 32 (1) | | 142.55 | 128.89 | 124.92 | 12.37 | |
 | | | | | | | |
-| 4090 | 1 | 5.54 | 4.99 | 4.51 | 4.44 | 11.02 | 19.86 |
-| 4090 | 4 | 13.67 | 11.4 | 10.3 | 9.84 | 13.68 | 28.02 |
-| 4090 | 8 | | 19.79 | 17.13 | 16.19 | 18.19 | |
-| 4090 | 16 | | 38.62 | 33.14 | 32.31 | 16.34 | |
-| 4090 | 32 (1) | | 76.57 | 65.96 | 62.05 | 18.96 | |
-| 4090 | 48 | | 114.44 | 98.78 | | 13.68 | |
+| 4090 | 1 | 5.54 | 4.99 | 2.66 | 2.58 | 48.30 | 53.43 |
+| 4090 | 4 | 13.67 | 11.4 | 8.81 | 8.46 | 25.79 | 38.11 |
+| 4090 | 8 |  | 19.79 | 17.55 | 16.62 | 16.02 |  |
+| 4090 | 16 |  | 38.62 | 35.65 | 34.07 | 11.78 |  |
+| 4090 | 32 (1) |  | 76.57 | 69.48 | 65.35 | 14.65 |  |
+| 4090 | 48 |  | 114.44 | 106.3 |  | 7.11 |  |


+(1) Batch Size >= 32 requires enable_vae_slicing() because of https://github.com/pytorch/pytorch/issues/81665.
+This is required for PyTorch 1.13.1, and also for PyTorch 2.0 and large batch sizes.

-(1) Batch Size >= 32 requires enable_vae_slicing() because of https://github.com/pytorch/pytorch/issues/81665																										
-This is required for PyTorch 1.13.1, and also for PyTorch 2.0 and batch size of 64
-
-For more details about how this benchmark was run, please refer to [this PR](https://github.com/huggingface/diffusers/pull/2303).	
+For more details about how this benchmark was run, please refer to [this PR](https://github.com/huggingface/diffusers/pull/2303) and to [the blog post](https://pytorch.org/blog/accelerated-diffusers-pt-20/).
@@ -141,7 +141,7 @@ Different schedulers come with different denoising speeds and quality trade-offs
 ```py
 >>> from diffusers import EulerDiscreteScheduler

->>> pipeline = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5")
+>>> pipeline = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5")
 >>> pipeline.scheduler = EulerDiscreteScheduler.from_config(pipeline.scheduler.config)
 ```

@@ -1,333 +1,271 @@
-<!--Copyright 2023 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                                                                                                                                                                                          
-                                                                                                                                                                                                                                              
-http://www.apache.org/licenses/LICENSE-2.0                                                                                                                                                                                                    
-                                                                                                                                                                                                                                              
-Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on                                                                                                                           
-an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the                                                                                                                            
-specific language governing permissions and limitations under the License.                                                                                                                                                                    
-->                                                                                                                                                                                                                                           
-                                                                                                                                                                                                                                              
-# The Stable Diffusion Guide 🎨                                                                                                                                                                                                           
-<a target="_blank" href="https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_101_guide.ipynb">                                                                                                                                                                                                                                                                                                                                                            
-    <img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"/>                                                                                                                                                 
-</a>                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    
-
-## Intro                                                                                                                                                                                                                                                                                                                                                                                                                                                                                
-
-Stable Diffusion is a [Latent Diffusion model](https://github.com/CompVis/latent-diffusion) developed by researchers from the Machine Vision and Learning group at LMU Munich, *a.k.a* CompVis.                                                                                                                                                                                                                                                                                             
-Model checkpoints were publicly released at the end of August 2022 by a collaboration of Stability AI, CompVis, and Runway with support from EleutherAI and LAION. For more information, you can check out [the official blog post](https://stability.ai/blog/stable-diffusion-public-release).                                                                                                                                                                                                  
-                                                                                                                                                                                                                                              
-Since its public release the community has done an incredible job at working together to make the stable diffusion checkpoints **faster**, **more memory efficient**, and **more performant**.                                                                                                                                                                                                                                                                                              
-                                                                                                                                                                                                                                              
-🧨 Diffusers offers a simple API to run stable diffusion with all memory, computing, and quality improvements.                                                                                                                                   
-                                                                                                                                                                                                                                              
-This notebook walks you through the improvements one-by-one so you can best leverage [`StableDiffusionPipeline`] for **inference**.                                                                                                          
-                                                                                                                                                                                                                                              
-## Prompt Engineering 🎨                                                                                                                                                                                                                      
-                                                                                                                                                                                                                                              
-When running *Stable Diffusion* in inference, we usually want to generate a certain type, or style of image and then improve upon it. Improving upon a previously generated image means running inference over and over again with a different prompt and potentially a different seed until we are happy with our generation.                                                                                                                                                                 
-
-So to begin with, it is most important to speed up stable diffusion as much as possible to generate as many pictures as possible in a given amount of time.                                                                                   
-
-This can be done by both improving the **computational efficiency** (speed) and the **memory efficiency** (GPU RAM).                                                                                                                          
-
-Let's start by looking into computational efficiency first.                                                                                                                                                                               
-
-Throughout the notebook, we will focus on [runwayml/stable-diffusion-v1-5](https://huggingface.co/runwayml/stable-diffusion-v1-5):                                                                                                           
-
-``` python                                                                                                                                                                                                                                    
-model_id = "runwayml/stable-diffusion-v1-5"                                                                                                                                                                                                   
-```                                                                                                                                                                                                                                           
-
-Let's load the pipeline.                                                                                                                                                                                                                      
-
-## Speed Optimization                                                                                                                                                                                                                                                                                                                                                                                                                                                                       
-
-``` python                                                                                                                                                                                                                                    
-from diffusers import StableDiffusionPipeline                                                                                                                                                                                                 
-                                                                                                                                                                                                                                              
-pipe = StableDiffusionPipeline.from_pretrained(model_id)                                                                                                                                                                                      
-```                                                                                                                                                                                                                                           
-                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            
-We aim at generating a beautiful photograph of an *old warrior chief* and will later try to find the best prompt to generate such a photograph. For now, let's keep the prompt simple:                                                                                                                                                                                                                                                                                                      
-                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            
-``` python                                                                                                                                                                                                                                    
-prompt = "portrait photo of a old warrior chief"                                                                                                                                                                                                                                                                                                                                                                                                                                            
-```                                                                                                                                                                                                                                           
-                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            
-To begin with, we should make sure we run inference on GPU, so let's move the pipeline to GPU, just like you would with any PyTorch module.                                                                                                   
-                                                                                                                                                                                                                                              
-``` python                                                                                                                                                                                                                                    
-pipe = pipe.to("cuda")                                                                                                                                                                                                                                                                                                                                                                                                                                                                      
-```                                                                                                                                                                                                                                           
-                                                                                                                                                                                                                                              
-To generate an image, you should use the [~`StableDiffusionPipeline.__call__`] method.                                                                                                                                                        
-                                                                                                                                                                                                                                              
-To make sure we can reproduce more or less the same image in every call, let's make use of the generator. See the documentation on reproducibility [here](./conceptual/reproducibility) for more information.                                                                                                                                                                                                                                                                                   
-
-``` python                                                                                                                                                                                                                                    
-generator = torch.Generator("cuda").manual_seed(0)                                                                                                                                                                                            
-```                                                                                                                                                                                                                                           
-                                                                                                                                                                                                                                              
-Now, let's take a spin on it.                                                                                                                                                                                                                 
-                                                                                                                                                                                                                                              
-``` python                                                                                                                                                                                                                                    
-image = pipe(prompt, generator=generator).images[0]                                                                                                                                                                                           
-image                                                                                                                                                                                                                                         
-```                                                                                                                                                                                                                                           
-
-![img](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_1.png)                                                                                                                                  
-                                                                                                                                                                                                                                              
-Cool, this now took roughly 30 seconds on a T4 GPU (you might see faster inference if your allocated GPU is better than a T4).                                                                                                              
-                                                                                                                                                                                                                                              
-The default run we did above used full float32 precision and ran the default number of inference steps (50). The easiest speed-ups come from switching to float16 (or half) precision and simply running fewer inference steps. Let's load the model now in float16 instead.                                                                                                                                                                                                                            
-                                                                                                                                                                                                                                              
-``` python                                                                                                                                                                                                                                    
-import torch                                                                                                                                                                                                                                  
-
-pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)                                                                                                                                                           
-pipe = pipe.to("cuda")                                                                                                                                                                                                                        
-```                                                                                                                                                                                                                                           
-
-And we can again call the pipeline to generate an image.                                                                                                                                                                                      
-
-``` python                                                                                                                                                                                                                                    
-generator = torch.Generator("cuda").manual_seed(0)                                                                                                                                                                                            
-
-image = pipe(prompt, generator=generator).images[0]                                                                                                                                                                                           
-image                                                                                                                                                                                                                                         
-```                                                                                                                                                                                                                                           
-![img](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_2.png)                                                                                                                                  
-
-Cool, this is almost three times as fast for arguably the same image quality.                                                                                                                                                           
-                                                                                                                                                                                                                                              
-We strongly suggest always running your pipelines in float16 as so far we have very rarely seen degradations in quality because of it.                                                                                                         
-
-Next, let's see if we need to use 50 inference steps or whether we could use significantly fewer. The number of inference steps is associated with the denoising scheduler we use. Choosing a more efficient scheduler could help us decrease the number of steps.
-
-Let's have a look at all the schedulers the stable diffusion pipeline is compatible with.                                                                                                                                                        
-                                                                                                                                                                                                                                              
-``` python                                                                                                                                                                                                                                    
-pipe.scheduler.compatibles                                                                                                                                                                                                                    
-```                                                                                                                                                                                                                                           
-                                                                                                                                                                                                                                              
-```                                                                                                                                                                                                                                           
-    [diffusers.schedulers.scheduling_dpmsolver_singlestep.DPMSolverSinglestepScheduler,                                                                                                                                                       
-     diffusers.schedulers.scheduling_lms_discrete.LMSDiscreteScheduler,                                                                                                                                                                       
-     diffusers.schedulers.scheduling_heun_discrete.HeunDiscreteScheduler,                                                                                                                                                                     
-     diffusers.schedulers.scheduling_pndm.PNDMScheduler,                                                                                                                                                                                      
-     diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler,                                                                                                                                                                   
-     diffusers.schedulers.scheduling_euler_ancestral_discrete.EulerAncestralDiscreteScheduler,                                                                                                                                                
-     diffusers.schedulers.scheduling_dpmsolver_multistep.DPMSolverMultistepScheduler,                                                                                                                                                         
-     diffusers.schedulers.scheduling_ddpm.DDPMScheduler,                                                                                                                                                                                      
-     diffusers.schedulers.scheduling_ddim.DDIMScheduler]                                                                                                                                                                                      
-```                                                                                                                                                                                                                                           
-
-Cool, that's a lot of schedulers.                                                                                                                                                                                                                                                                                                                                                                                                                                                           
-
-🧨 Diffusers is constantly adding a bunch of novel schedulers/samplers that can be used with Stable Diffusion. For more information, we recommend taking a look at the official documentation [here](https://huggingface.co/docs/diffusers/main/en/api/schedulers/overview).                                                                                                                                                                                                              
-                                                                                                                                                                                                                                              
-Alright, right now Stable Diffusion is using the `PNDMScheduler` which usually requires around 50 inference steps. However, other schedulers such as `DPMSolverMultistepScheduler` or `DPMSolverSinglestepScheduler` seem to get away with just 20 to 25 inference steps. Let's try them out.                                                                                                                                                                                               
-                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            
-You can set a new scheduler by making use of the [from_config](https://huggingface.co/docs/diffusers/main/en/api/configuration#diffusers.ConfigMixin.from_config) function.                                                                                                                                                                                                                                                                                                                 
-                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            
-``` python                                                                                                                                                                                                                                    
-from diffusers import DPMSolverMultistepScheduler                                                                                                                                                                                             
-                                                                                                                                                                                                                                              
-pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)                                                                                                                                                               
-```                                                                                                                                                                                                                                           
-                                                                                                                                                                                                                                              
-Now, let's try to reduce the number of inference steps to just 20.                                                                                                                                                                            
-
-``` python                                                                                                                                                                                                                                    
-generator = torch.Generator("cuda").manual_seed(0)                                                                                                                                                                                            
-                                                                                                                                                                                                                                              
-image = pipe(prompt, generator=generator, num_inference_steps=20).images[0]                                                                                                                                                                   
-image                                                                                                                                                                                                                                         
-```                                                                                                                                                                                                                                           
-                                                                                                                                                                                                                                              
-![img](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_3.png)                                                                                                                                                                                                                                                                                                                                                                                
-                                                                                                                                                                                                                                              
-The image now does look a little different, but it's arguably still of equally high quality. We now cut inference time to just 4 seconds though 😍. 
-
-## Memory Optimization                                                                                                                                                                                                                        
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.

-Less memory used in generation indirectly implies more speed, since we're often trying to maximize how many images we can generate per second. Usually, the more images per inference run, the more images per second too.                                                                                                                                              
+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

-The easiest way to see how many images we can generate at once is to simply try it out, and see when we get a *"Out-of-memory (OOM)"* error.                                                                                                          
+http://www.apache.org/licenses/LICENSE-2.0

-We can run batched inference by simply passing a list of prompts and generators. Let's define a quick function that generates a batch for us.                                                                                                
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+                                                               
+# Effective and efficient diffusion

-``` python                                                                                                                                                                                                                                    
-def get_inputs(batch_size=1):                                                                                                                                                                                                                 
-  generator = [torch.Generator("cuda").manual_seed(i) for i in range(batch_size)]                                                                                                                                                             
-  prompts = batch_size * [prompt]                                                                                                                                                                                                             
-  num_inference_steps = 20                                                                                                                                                                                                                    
+[[open-in-colab]]

-  return {"prompt": prompts, "generator": generator, "num_inference_steps": num_inference_steps}                                                                                                                                              
-```                                                                                                                                                                                                                                           
-This function returns a list of prompts and a list of generators, so we can reuse the generator that produced a result we like.
+Getting the [`DiffusionPipeline`] to generate images in a certain style or include what you want can be tricky. Often times, you have to run the [`DiffusionPipeline`] several times before you end up with an image you're happy with. But generating something out of nothing is a computationally intensive process, especially if you're running inference over and over again. 

-We also need a method that allows us to easily display a batch of images.                                                                                                                                                                     
+This is why it's important to get the most *computational* (speed) and *memory* (GPU RAM) efficiency from the pipeline to reduce the time between inference cycles so you can iterate faster.

-``` python                                                                                                                                                                                                                                    
-from PIL import Image                                                                                                                                                                                                                         
+This tutorial walks you through how to generate faster and better with the [`DiffusionPipeline`].

-def image_grid(imgs, rows=2, cols=2):                                                                                                                                                                                                         
-    w, h = imgs[0].size                                                                                                                                                                                                                       
-    grid = Image.new('RGB', size=(cols*w, rows*h))                                                                                                                                                                                            
-                                                                                                                                                                                                                                              
-    for i, img in enumerate(imgs):                                                                                                                                                                                                            
-        grid.paste(img, box=(i%cols*w, i//cols*h))                                                                                                                                                                                            
-    return grid                                                                                                                                                                                                                               
-```                                                                                                                                                                                                                                           
+Begin by loading the [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5) model:

-Cool, let's see how much memory we can use starting with `batch_size=4`.                                                                                                                                                                      
+```python
+from diffusers import DiffusionPipeline

-``` python                                                                                                                                                                                                                                    
-images = pipe(**get_inputs(batch_size=4)).images                                                                                                                                                                                              
-image_grid(images)                                                                                                                                                                                                                            
-```                                                                                                                                                                                                                                           
+model_id = "runwayml/stable-diffusion-v1-5"
+pipeline = DiffusionPipeline.from_pretrained(model_id)
+```

-![img](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_4.png)                                                                                                                                  
+The example prompt you'll use is a portrait of an old warrior chief, but feel free to use your own prompt:

-Going over a batch_size of 4 will error out in this notebook (assuming we are running it on a T4 GPU). Also, we can see we only generate slightly more images per second (3.75s/image) compared to 4s/image previously.                                                                                                                                                                                                                                                                                                                   
+```python
+prompt = "portrait photo of a old warrior chief"
+```

-However, the community has found some nice tricks to improve the memory constraints further. After stable diffusion was released, the community found improvements within days and shared them freely over GitHub - open-source at its finest! I believe the original idea came from [this](https://github.com/basujindal/stable-diffusion/pull/117) GitHub thread.                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    
+## Speed

-By far most of the memory is taken up by the cross-attention layers. Instead of running this operation in batch, one can run it sequentially to save a significant amount of memory.                                                                                                                                                                                                                                                                                                         
+<Tip>

-It can easily be enabled by calling `enable_attention_slicing` as is documented [here](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/text2img#diffusers.StableDiffusionPipeline.enable_attention_slicing).                                                                                                                                                                                                                                                   
+💡 If you don't have access to a GPU, you can use one for free from a GPU provider like [Colab](https://colab.research.google.com/)!

-``` python                                                                                                                                                                                                                                    
-pipe.enable_attention_slicing()                                                                                                                                                                                                               
-```                                                                                                                                                                                                                                           
+</Tip>

-Great, now that attention slicing is enabled, let's try to double the batch size again, going for `batch_size=8`.                                                                                                                              
+One of the simplest ways to speed up inference is to place the pipeline on a GPU the same way you would with any PyTorch module:

-``` python                                                                                                                                                                                                                                    
-images = pipe(**get_inputs(batch_size=8)).images                                                                                                                                                                                              
-image_grid(images, rows=2, cols=4)                                                                                                                                                                                                            
-```                                                                                                                                                                                                                                           
+```python
+pipeline = pipeline.to("cuda")
+```

-![img](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_5.png)                                                                                                                                  
+To make sure you can use the same image and improve on it, use a [`Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) and set a seed for [reproducibility](./using-diffusers/reproducibility):

-Nice, it works. However, the speed gain is again not very big (it might however be much more significant on other GPUs).                                                                                                                      
+```python
+generator = torch.Generator("cuda").manual_seed(0)
+```

-We're at roughly 3.5 seconds per image 🔥 which is probably the fastest we can be with a simple T4 without sacrificing quality.                                                                                                               
+Now you can generate an image:

-Next, let's look into how to improve the quality!                                                                                                                                                                                             
+```python
+image = pipeline(prompt, generator=generator).images[0]
+image
+```

-## Quality Improvements                                                                                                                                                                                                                       
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_1.png">
+</div>

-Now that our image generation pipeline is blazing fast, let's try to get maximum image quality.                                                                                                                                               
+This process took ~30 seconds on a T4 GPU (it might be faster if your allocated GPU is better than a T4). By default, the [`DiffusionPipeline`] runs inference with full `float32` precision for 50 inference steps. You can speed this up by switching to a lower precision like `float16` or running fewer inference steps. 

-First of all, image quality is extremely subjective, so it's difficult to make general claims here.                                                                                                                                           
+Let's start by loading the model in `float16` and generate an image:

-The most obvious step to take to improve quality is to use *better checkpoints*. Since the release of Stable Diffusion, many improved versions have been released, which are summarized here:                                                                                                                                                                                                                                                                                                
+```python
+import torch

-   *Official Release - 22 Aug 2022*: [Stable-Diffusion 1.4](https://huggingface.co/CompVis/stable-diffusion-v1-4)                                                                                                                            
-   *20 October 2022*: [Stable-Diffusion 1.5](https://huggingface.co/runwayml/stable-diffusion-v1-5)                                                                                                                                          
-   *24 Nov 2022*: [Stable-Diffusion 2.0](https://huggingface.co/stabilityai/stable-diffusion-2-0)                                                                                                                                            
-   *7 Dec 2022*: [Stable-Diffusion 2.1](https://huggingface.co/stabilityai/stable-diffusion-2-1)                                                                                                                                             
+pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
+pipeline = pipeline.to("cuda")
+generator = torch.Generator("cuda").manual_seed(0)
+image = pipeline(prompt, generator=generator).images[0]
+image
+```

-Newer versions don't necessarily mean better image quality with the same parameters. People mentioned that *2.0* is slightly worse than *1.5* for certain prompts, but given the right prompt engineering *2.0* and *2.1* seem to be better.                                                                                                                                                                                                                                                 
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_2.png">
+</div>

-Overall, we strongly recommend just trying the models out and reading up on advice online (e.g. it has been shown that using negative prompts is very important for 2.0 and 2.1 to get the highest possible quality. See for example [this nice blog post](https://minimaxir.com/2022/11/stable-diffusion-negative-prompt/).                                                                                                                                                                   
+This time, it only took ~11 seconds to generate the image, which is almost 3x faster than before!

-Additionally, the community has started fine-tuning many of the above versions on certain styles with some of them having an extremely high quality and gaining a lot of traction.                                                                                                                                                                                                                                                                                                          
+<Tip>

-We recommend having a look at all [diffusers checkpoints sorted by downloads and trying out the different checkpoints](https://huggingface.co/models?library=diffusers).                                                                                                                                                                                                                                                                                                               
+💡 We strongly suggest always running your pipelines in `float16`, and so far, we've rarely seen any degradation in output quality.

-For the following, we will stick to v1.5 for simplicity.                                                                                                                                                                                      
+</Tip>

-Next, we can also try to optimize single components of the pipeline, e.g. switching out the latent decoder. For more details on how the whole Stable Diffusion pipeline works, please have a look at [this blog post](https://huggingface.co/blog/stable_diffusion).                                                                                                                                                                                                                        
+Another option is to reduce the number of inference steps. Choosing a more efficient scheduler could help decrease the number of steps without sacrificing output quality. You can find which schedulers are compatible with the current model in the [`DiffusionPipeline`] by calling the `compatibles` method:

-Let's load [stabilityai's newest auto-decoder](https://huggingface.co/stabilityai/stable-diffusion-2-1).                                                                                                                                      
+```python
+pipeline.scheduler.compatibles
+[
+    diffusers.schedulers.scheduling_lms_discrete.LMSDiscreteScheduler,
+    diffusers.schedulers.scheduling_unipc_multistep.UniPCMultistepScheduler,
+    diffusers.schedulers.scheduling_k_dpm_2_discrete.KDPM2DiscreteScheduler,
+    diffusers.schedulers.scheduling_deis_multistep.DEISMultistepScheduler,
+    diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler,
+    diffusers.schedulers.scheduling_dpmsolver_multistep.DPMSolverMultistepScheduler,
+    diffusers.schedulers.scheduling_ddpm.DDPMScheduler,
+    diffusers.schedulers.scheduling_dpmsolver_singlestep.DPMSolverSinglestepScheduler,
+    diffusers.schedulers.scheduling_k_dpm_2_ancestral_discrete.KDPM2AncestralDiscreteScheduler,
+    diffusers.schedulers.scheduling_heun_discrete.HeunDiscreteScheduler,
+    diffusers.schedulers.scheduling_pndm.PNDMScheduler,
+    diffusers.schedulers.scheduling_euler_ancestral_discrete.EulerAncestralDiscreteScheduler,
+    diffusers.schedulers.scheduling_ddim.DDIMScheduler,
+]
+```

-``` python                                                                                                                                                                                                                                    
-from diffusers import AutoencoderKL                                                                                                                                                                                                           
+The Stable Diffusion model uses the [`PNDMScheduler`] by default which usually requires ~50 inference steps, but more performant schedulers like [`DPMSolverMultistepScheduler`], require only ~20 or 25 inference steps. Use the [`ConfigMixin.from_config`] method to load a new scheduler:

-vae = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse", torch_dtype=torch.float16).to("cuda")                                                                                                                                        
-```                                                                                                                                                                                                                                           
+```python
+from diffusers import DPMSolverMultistepScheduler

-Now we can set it to the vae of the pipeline to use it.                                                                                                                                                                                       
+pipeline.scheduler = DPMSolverMultistepScheduler.from_config(pipeline.scheduler.config)
+```

-``` python                                                                                                                                                                                                                                    
-pipe.vae = vae                                                                                                                                                                                                                                
-```                                                                                                                                                                                                                                           
+Now set the `num_inference_steps` to 20:

-Let's run the same prompt as before to compare quality.                                                                                                                                                                                       
+```python
+generator = torch.Generator("cuda").manual_seed(0)
+image = pipeline(prompt, generator=generator, num_inference_steps=20).images[0]
+image
+```

-``` python                                                                                                                                                                                                                                    
-images = pipe(**get_inputs(batch_size=8)).images                                                                                                                                                                                              
-image_grid(images, rows=2, cols=4)                                                                                                                                                                                                            
-```                                                                                                                                                                                                                                           
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_3.png">
+</div>

-![img](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_6.png)                                                                                                                                  
+Great, you've managed to cut the inference time to just 4 seconds! ⚡️

-Seems like the difference is only very minor, but the new generations are arguably a bit *sharper*.                                                                                                                                           
+## Memory

-Cool, finally, let's look a bit into prompt engineering.                                                                                                                                                                                      
+The other key to improving pipeline performance is consuming less memory, which indirectly implies more speed, since you're often trying to maximize the number of images generated per second. The easiest way to see how many images you can generate at once is to try out different batch sizes until you get an `OutOfMemoryError` (OOM).

-Our goal was to generate a photo of an old warrior chief. Let's now try to bring a bit more color into the photos and make the look more impressive.                                                                                          
+Create a function that'll generate a batch of images from a list of prompts and `Generators`. Make sure to assign each `Generator` a seed so you can reuse it if it produces a good result.

-Originally our prompt was "*portrait photo of an old warrior chief*".                                                                                                                                                                          
+```python
+def get_inputs(batch_size=1):
+    generator = [torch.Generator("cuda").manual_seed(i) for i in range(batch_size)]
+    prompts = batch_size * [prompt]
+    num_inference_steps = 20

-To improve the prompt, it often helps to add cues that could have been used online to save high-quality photos, as well as add more details.                                                                                         
-Essentially, when doing prompt engineering, one has to think:                                                                                                                                                                                 
+    return {"prompt": prompts, "generator": generator, "num_inference_steps": num_inference_steps}
+```

-   How was the photo or similar photos of the one I want probably stored on the internet?                                                                                                                                                    
-   What additional detail can I give that steers the models into the style that I want?                                                                                                                                                      
+You'll also need a function that'll display each batch of images:

-Cool, let's add more details.                                                                                                                                                                                                                 
+```python
+from PIL import image

-``` python                                                                                                                                                                                                                                    
-prompt += ", tribal panther make up, blue on red, side profile, looking away, serious eyes"                                                                                                                                                   
-```                                                                                                                                                                                                                                           

-and let's also add some cues that usually help to generate higher quality images.                                                                                                                                                             
+def image_grid(imgs, rows=2, cols=2):
+    w, h = imgs[0].size
+    grid = Image.new("RGB", size=(cols * w, rows * h))

-``` python                                                                                                                                                                                                                                    
-prompt += " 50mm portrait photography, hard rim lighting photography--beta --ar 2:3  --beta --upbeta"                                                                                                                                         
-prompt                                                                                                                                                                                                                                        
-```                                                                                                                                                                                                                                                                                                                                                                                                                                                                           
+    for i, img in enumerate(imgs):
+        grid.paste(img, box=(i % cols * w, i // cols * h))
+    return grid
+```

-Cool, let's now try this prompt.                                                                                                                                                                                                              
+Start with `batch_size=4` and see how much memory you've consumed:

-``` python                                                                                                                                                                                                                                    
-images = pipe(**get_inputs(batch_size=8)).images                                                                                                                                                                                              
-image_grid(images, rows=2, cols=4)                                                                                                                                                                                                            
-```                                                                                                                                                                                                                                           
+```python
+images = pipeline(**get_inputs(batch_size=4)).images
+image_grid(images)
+```

-![img](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_7.png)                                                                                                                                  
+Unless you have a GPU with more RAM, the code above probably returned an `OOM` error! Most of the memory is taken up by the cross-attention layers. Instead of running this operation in a batch, you can run it sequentially to save a significant amount of memory. All you have to do is configure the pipeline to use the [`~DiffusionPipeline.enable_attention_slicing`] function:

-Pretty impressive! We got some very high-quality image generations there. The 2nd image is my personal favorite, so I'll re-use this seed and see whether I can tweak the prompts slightly by using "oldest warrior", "old", "", and "young" instead of "old".                                                                                                                                                                                                                    
+```python
+pipeline.enable_attention_slicing()
+```

-``` python                                                                                                                                                                                                                                    
-prompts = [                                                                                                                                                                                                                                   
-    "portrait photo of the oldest warrior chief, tribal panther make up, blue on red, side profile, looking away, serious eyes 50mm portrait photography, hard rim lighting photography--beta --ar 2:3  --beta --upbeta",                                                                                                                                                                                                                                                                   
-    "portrait photo of a old warrior chief, tribal panther make up, blue on red, side profile, looking away, serious eyes 50mm portrait photography, hard rim lighting photography--beta --ar 2:3  --beta --upbeta",                                                                                                                                                                                                                                                                        
-    "portrait photo of a warrior chief, tribal panther make up, blue on red, side profile, looking away, serious eyes 50mm portrait photography, hard rim lighting photography--beta --ar 2:3  --beta --upbeta",                                                                                                                                                                                                                                                                            
-    "portrait photo of a young warrior chief, tribal panther make up, blue on red, side profile, looking away, serious eyes 50mm portrait photography, hard rim lighting photography--beta --ar 2:3  --beta --upbeta",                                                                                                                                                                                                                                                                      
-]                                                                                                                                                                                                                                             
+Now try increasing the `batch_size` to 8!

-generator = [torch.Generator("cuda").manual_seed(1) for _ in range(len(prompts))]  # 1 because we want the 2nd image                                                                                                                          
+```python
+images = pipeline(**get_inputs(batch_size=8)).images
+image_grid(images, rows=2, cols=4)
+```

-images = pipe(prompt=prompts, generator=generator, num_inference_steps=25).images                                                                                                                                                             
-image_grid(images)                                                                                                                                                                                                                            
-```                                                                                                                                                                                                                                           
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_5.png">
+</div>

-![img](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_8.png)                                                                                                                                  
+Whereas before you couldn't even generate a batch of 4 images, now you can generate a batch of 8 images at ~3.5 seconds per image! This is probably the fastest you can go on a T4 GPU without sacrificing quality.

-The first picture looks nice! The eye movement slightly changed and looks nice. This finished up our 101-guide on how to use Stable Diffusion 🤗.                                                                                              
+## Quality

-For more information on optimization or other guides, I recommend taking a look at the following:                                                                                                                                            
+In the last two sections, you learned how to optimize the speed of your pipeline by using `fp16`, reducing the number of inference steps by using a more performant scheduler, and enabling attention slicing to reduce memory consumption. Now you're going to focus on how to improve the quality of generated images.

-   [Blog post about Stable Diffusion](https://huggingface.co/blog/stable_diffusion): In-detail blog post explaining Stable Diffusion.                                                                                                        
-   [FlashAttention](https://huggingface.co/docs/diffusers/optimization/xformers): XFormers flash attention can optimize your model even further with more speed and memory improvements.                                                                                                                                                                                                                                                                                                   
-   [Dreambooth](https://huggingface.co/docs/diffusers/training/dreambooth) - Quickly customize the model by fine-tuning it.                                                                                                                  
-   [General info on Stable Diffusion](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/overview) - Info on other tasks that are powered by Stable Diffusion.
+### Better checkpoints
+
+The most obvious step is to use better checkpoints. The Stable Diffusion model is a good starting point, and since its official launch, several improved versions have also been released. However, using a newer version doesn't automatically mean you'll get better results. You'll still have to experiment with different checkpoints yourself, and do a little research (such as using [negative prompts](https://minimaxir.com/2022/11/stable-diffusion-negative-prompt/)) to get the best results.
+
+As the field grows, there are more and more high-quality checkpoints finetuned to produce certain styles. Try exploring the [Hub](https://huggingface.co/models?library=diffusers&sort=downloads) and [Diffusers Gallery](https://huggingface.co/spaces/huggingface-projects/diffusers-gallery) to find one you're interested in!
+
+### Better pipeline components
+
+You can also try replacing the current pipeline components with a newer version. Let's try loading the latest [autodecoder](https://huggingface.co/stabilityai/stable-diffusion-2-1/tree/main/vae) from Stability AI into the pipeline, and generate some images:
+
+```python
+from diffusers import AutoencoderKL
+
+vae = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse", torch_dtype=torch.float16).to("cuda")
+pipeline.vae = vae
+images = pipeline(**get_inputs(batch_size=8)).images
+image_grid(images, rows=2, cols=4)
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_6.png">
+</div>
+
+### Better prompt engineering
+
+The text prompt you use to generate an image is super important, so much so that it is called *prompt engineering*. Some considerations to keep during prompt engineering are:
+
+- How is the image or similar images of the one I want to generate stored on the internet?
+- What additional detail can I give that steers the model towards the style I want?
+
+With this in mind, let's improve the prompt to include color and higher quality details:
+
+```python
+prompt += ", tribal panther make up, blue on red, side profile, looking away, serious eyes"
+prompt += " 50mm portrait photography, hard rim lighting photography--beta --ar 2:3  --beta --upbeta"
+```
+
+Generate a batch of images with the new prompt:
+
+```python
+images = pipeline(**get_inputs(batch_size=8)).images
+image_grid(images, rows=2, cols=4)
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_7.png">
+</div>
+
+Pretty impressive! Let's tweak the second image - corresponding to the `Generator` with a seed of `1` - a bit more by adding some text about the age of the subject:
+
+```python
+prommpts = [
+    "portrait photo of the oldest warrior chief, tribal panther make up, blue on red, side profile, looking away, serious eyes 50mm portrait photography, hard rim lighting photography--beta --ar 2:3  --beta --upbeta",
+    "portrait photo of a old warrior chief, tribal panther make up, blue on red, side profile, looking away, serious eyes 50mm portrait photography, hard rim lighting photography--beta --ar 2:3  --beta --upbeta",
+    "portrait photo of a warrior chief, tribal panther make up, blue on red, side profile, looking away, serious eyes 50mm portrait photography, hard rim lighting photography--beta --ar 2:3  --beta --upbeta",
+    "portrait photo of a young warrior chief, tribal panther make up, blue on red, side profile, looking away, serious eyes 50mm portrait photography, hard rim lighting photography--beta --ar 2:3  --beta --upbeta",
+]
+
+generator = [torch.Generator("cuda").manual_seed(1) for _ in range(len(prompts))]
+images = pipeline(prompt=prompts, generator=generator, num_inference_steps=25).images
+image_grid(images)
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/diffusers/docs-images/resolve/main/stable_diffusion_101/sd_101_8.png">
+</div>
+
+## Next steps
+
+In this tutorial, you learned how to optimize a [`DiffusionPipeline`] for computational and memory efficiency as well as improving the quality of generated outputs. If you're interested in making your pipeline even faster, take a look at the following resources:
+
+- Enable [xFormers](./optimization/xformers) memory efficient attention mechanism for faster speed and reduced memory consumption.
+- Learn how in [PyTorch 2.0](./optimization/torch2.0), [`torch.compile`](https://pytorch.org/docs/stable/generated/torch.compile.html) can yield 2-9% faster inference speed.
+- Many optimization techniques for inference are also included in this memory and speed [guide](./optimization/fp16), such as memory offloading.
@@ -0,0 +1,313 @@
+<!--Copyright 2023 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
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# ControlNet
+
+[Adding Conditional Control to Text-to-Image Diffusion Models](https://arxiv.org/abs/2302.05543) (ControlNet) by Lvmin Zhang and Maneesh Agrawala.
+
+This example is based on the [training example in the original ControlNet repository](https://github.com/lllyasviel/ControlNet/blob/main/docs/train.md). It trains a ControlNet to fill circles using a [small synthetic dataset](https://huggingface.co/datasets/fusing/fill50k).
+
+## Installing the dependencies
+
+Before running the scripts, make sure to install the library's training dependencies.
+
+<Tip warning={true}>
+
+To successfully run the latest versions of the example scripts, we highly recommend **installing from source** and keeping the installation up to date. We update the example scripts frequently and install example-specific requirements.
+
+</Tip>
+
+To do this, execute the following steps in a new virtual environment:
+```bash
+git clone https://github.com/huggingface/diffusers
+cd diffusers
+pip install -e .
+```
+
+Then navigate into the example folder and run:
+```bash
+pip install -r requirements.txt
+```
+
+And initialize an [🤗Accelerate](https://github.com/huggingface/accelerate/) environment with:
+
+```bash
+accelerate config
+```
+
+Or for a default 🤗Accelerate configuration without answering questions about your environment:
+
+```bash
+accelerate config default
+```
+
+Or if your environment doesn't support an interactive shell like a notebook:
+
+```python
+from accelerate.utils import write_basic_config
+
+write_basic_config()
+```
+
+## Circle filling dataset
+
+The original dataset is hosted in the ControlNet [repo](https://huggingface.co/lllyasviel/ControlNet/blob/main/training/fill50k.zip), but we re-uploaded it [here](https://huggingface.co/datasets/fusing/fill50k) to be compatible with 🤗 Datasets so that it can handle the data loading within the training script.
+
+Our training examples use [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5) because that is what the original set of ControlNet models was trained on. However, ControlNet can be trained to augment any compatible Stable Diffusion model (such as [`CompVis/stable-diffusion-v1-4`](https://huggingface.co/CompVis/stable-diffusion-v1-4)) or [`stabilityai/stable-diffusion-2-1`](https://huggingface.co/stabilityai/stable-diffusion-2-1).
+
+## Training
+
+Download the following images to condition our training with:
+
+```sh
+wget https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/conditioning_image_1.png
+
+wget https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/conditioning_image_2.png
+```
+
+
+```bash
+export MODEL_DIR="runwayml/stable-diffusion-v1-5"
+export OUTPUT_DIR="path to save model"
+
+accelerate launch train_controlnet.py \
+ --pretrained_model_name_or_path=$MODEL_DIR \
+ --output_dir=$OUTPUT_DIR \
+ --dataset_name=fusing/fill50k \
+ --resolution=512 \
+ --learning_rate=1e-5 \
+ --validation_image "./conditioning_image_1.png" "./conditioning_image_2.png" \
+ --validation_prompt "red circle with blue background" "cyan circle with brown floral background" \
+ --train_batch_size=4
+```
+
+This default configuration requires ~38GB VRAM.
+
+By default, the training script logs outputs to tensorboard. Pass `--report_to wandb` to use Weights &
+Biases.
+
+Gradient accumulation with a smaller batch size can be used to reduce training requirements to ~20 GB VRAM.
+
+```bash
+export MODEL_DIR="runwayml/stable-diffusion-v1-5"
+export OUTPUT_DIR="path to save model"
+
+accelerate launch train_controlnet.py \
+ --pretrained_model_name_or_path=$MODEL_DIR \
+ --output_dir=$OUTPUT_DIR \
+ --dataset_name=fusing/fill50k \
+ --resolution=512 \
+ --learning_rate=1e-5 \
+ --validation_image "./conditioning_image_1.png" "./conditioning_image_2.png" \
+ --validation_prompt "red circle with blue background" "cyan circle with brown floral background" \
+ --train_batch_size=1 \
+ --gradient_accumulation_steps=4
+```
+
+## Training with multiple GPUs
+
+`accelerate` allows for seamless multi-GPU training. Follow the instructions [here](https://huggingface.co/docs/accelerate/basic_tutorials/launch)
+for running distributed training with `accelerate`. Here is an example command:
+
+```bash 
+export MODEL_DIR="runwayml/stable-diffusion-v1-5"
+export OUTPUT_DIR="path to save model"
+
+accelerate launch --mixed_precision="fp16" --multi_gpu train_controlnet.py \
+ --pretrained_model_name_or_path=$MODEL_DIR \
+ --output_dir=$OUTPUT_DIR \
+ --dataset_name=fusing/fill50k \
+ --resolution=512 \
+ --learning_rate=1e-5 \
+ --validation_image "./conditioning_image_1.png" "./conditioning_image_2.png" \
+ --validation_prompt "red circle with blue background" "cyan circle with brown floral background" \
+ --train_batch_size=4 \
+ --mixed_precision="fp16" \
+ --tracker_project_name="controlnet-demo" \
+ --report_to=wandb
+```
+
+## Example results
+
+#### After 300 steps with batch size 8
+
+| |  | 
+|-------------------|:-------------------------:|
+| | red circle with blue background  | 
+![conditioning image](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/conditioning_image_1.png) | ![red circle with blue background](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/red_circle_with_blue_background_300_steps.png) |
+| | cyan circle with brown floral background | 
+![conditioning image](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/conditioning_image_2.png) | ![cyan circle with brown floral background](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/cyan_circle_with_brown_floral_background_300_steps.png) |
+
+
+#### After 6000 steps with batch size 8:
+
+| |  | 
+|-------------------|:-------------------------:|
+| | red circle with blue background  | 
+![conditioning image](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/conditioning_image_1.png) | ![red circle with blue background](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/red_circle_with_blue_background_6000_steps.png) |
+| | cyan circle with brown floral background | 
+![conditioning image](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/conditioning_image_2.png) | ![cyan circle with brown floral background](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_training/cyan_circle_with_brown_floral_background_6000_steps.png) |
+
+## Training on a 16 GB GPU
+
+Enable the following optimizations to train on a 16GB GPU:
+
+- Gradient checkpointing
+- bitsandbyte's 8-bit optimizer (take a look at the [installation]((https://github.com/TimDettmers/bitsandbytes#requirements--installation) instructions if you don't already have it installed)
+
+Now you can launch the training script:
+
+```bash
+export MODEL_DIR="runwayml/stable-diffusion-v1-5"
+export OUTPUT_DIR="path to save model"
+
+accelerate launch train_controlnet.py \
+ --pretrained_model_name_or_path=$MODEL_DIR \
+ --output_dir=$OUTPUT_DIR \
+ --dataset_name=fusing/fill50k \
+ --resolution=512 \
+ --learning_rate=1e-5 \
+ --validation_image "./conditioning_image_1.png" "./conditioning_image_2.png" \
+ --validation_prompt "red circle with blue background" "cyan circle with brown floral background" \
+ --train_batch_size=1 \
+ --gradient_accumulation_steps=4 \
+ --gradient_checkpointing \
+ --use_8bit_adam
+```
+
+## Training on a 12 GB GPU
+
+Enable the following optimizations to train on a 12GB GPU:
+- Gradient checkpointing
+- bitsandbyte's 8-bit optimizer (take a look at the [installation]((https://github.com/TimDettmers/bitsandbytes#requirements--installation) instructions if you don't already have it installed)
+- xFormers (take a look at the [installation](https://huggingface.co/docs/diffusers/training/optimization/xformers) instructions if you don't already have it installed)
+- set gradients to `None`
+
+```bash
+export MODEL_DIR="runwayml/stable-diffusion-v1-5"
+export OUTPUT_DIR="path to save model"
+
+accelerate launch train_controlnet.py \
+ --pretrained_model_name_or_path=$MODEL_DIR \
+ --output_dir=$OUTPUT_DIR \
+ --dataset_name=fusing/fill50k \
+ --resolution=512 \
+ --learning_rate=1e-5 \
+ --validation_image "./conditioning_image_1.png" "./conditioning_image_2.png" \
+ --validation_prompt "red circle with blue background" "cyan circle with brown floral background" \
+ --train_batch_size=1 \
+ --gradient_accumulation_steps=4 \
+ --gradient_checkpointing \
+ --use_8bit_adam \
+ --enable_xformers_memory_efficient_attention \
+ --set_grads_to_none
+```
+
+When using `enable_xformers_memory_efficient_attention`, please make sure to install `xformers` by `pip install xformers`. 
+
+## Training on an 8 GB GPU
+
+We have not exhaustively tested DeepSpeed support for ControlNet. While the configuration does
+save memory, we have not confirmed whether the configuration trains successfully. You will very likely
+have to make changes to the config to have a successful training run.
+
+Enable the following optimizations to train on a 8GB GPU:
+- Gradient checkpointing
+- bitsandbyte's 8-bit optimizer (take a look at the [installation]((https://github.com/TimDettmers/bitsandbytes#requirements--installation) instructions if you don't already have it installed)
+- xFormers (take a look at the [installation](https://huggingface.co/docs/diffusers/training/optimization/xformers) instructions if you don't already have it installed)
+- set gradients to `None`
+- DeepSpeed stage 2 with parameter and optimizer offloading
+- fp16 mixed precision
+
+[DeepSpeed](https://www.deepspeed.ai/) can offload tensors from VRAM to either 
+CPU or NVME. This requires significantly more RAM (about 25 GB).
+
+You'll have to configure your environment with `accelerate config` to enable DeepSpeed stage 2.
+
+The configuration file should look like this:
+
+```yaml
+compute_environment: LOCAL_MACHINE
+deepspeed_config:
+  gradient_accumulation_steps: 4
+  offload_optimizer_device: cpu
+  offload_param_device: cpu
+  zero3_init_flag: false
+  zero_stage: 2
+distributed_type: DEEPSPEED
+```
+
+<Tip>
+
+See [documentation](https://huggingface.co/docs/accelerate/usage_guides/deepspeed) for more DeepSpeed configuration options.
+
+<Tip>
+
+Changing the default Adam optimizer to DeepSpeed's Adam
+`deepspeed.ops.adam.DeepSpeedCPUAdam` gives a substantial speedup but
+it requires a CUDA toolchain with the same version as PyTorch. 8-bit optimizer
+does not seem to be compatible with DeepSpeed at the moment.
+
+```bash
+export MODEL_DIR="runwayml/stable-diffusion-v1-5"
+export OUTPUT_DIR="path to save model"
+
+accelerate launch train_controlnet.py \
+ --pretrained_model_name_or_path=$MODEL_DIR \
+ --output_dir=$OUTPUT_DIR \
+ --dataset_name=fusing/fill50k \
+ --resolution=512 \
+ --validation_image "./conditioning_image_1.png" "./conditioning_image_2.png" \
+ --validation_prompt "red circle with blue background" "cyan circle with brown floral background" \
+ --train_batch_size=1 \
+ --gradient_accumulation_steps=4 \
+ --gradient_checkpointing \
+ --enable_xformers_memory_efficient_attention \
+ --set_grads_to_none \
+ --mixed_precision fp16
+```
+
+## Inference
+
+The trained model can be run with the [`StableDiffusionControlNetPipeline`].
+Set `base_model_path` and `controlnet_path` to the values `--pretrained_model_name_or_path` and 
+`--output_dir` were respectively set to in the training script.
+
+```py
+from diffusers import StableDiffusionControlNetPipeline, ControlNetModel, UniPCMultistepScheduler
+from diffusers.utils import load_image
+import torch
+
+base_model_path = "path to model"
+controlnet_path = "path to controlnet"
+
+controlnet = ControlNetModel.from_pretrained(controlnet_path, torch_dtype=torch.float16)
+pipe = StableDiffusionControlNetPipeline.from_pretrained(
+    base_model_path, controlnet=controlnet, torch_dtype=torch.float16
+)
+
+# speed up diffusion process with faster scheduler and memory optimization
+pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config)
+# remove following line if xformers is not installed
+pipe.enable_xformers_memory_efficient_attention()
+
+pipe.enable_model_cpu_offload()
+
+control_image = load_image("./conditioning_image_1.png")
+prompt = "pale golden rod circle with old lace background"
+
+# generate image
+generator = torch.manual_seed(0)
+image = pipe(prompt, num_inference_steps=20, generator=generator, image=control_image).images[0]
+
+image.save("./output.png")
+```
@@ -0,0 +1,289 @@
+<!--Copyright 2023 Custom Diffusion authors 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
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Custom Diffusion training example 
+
+[Custom Diffusion](https://arxiv.org/abs/2212.04488) is a method to customize text-to-image models like Stable Diffusion given just a few (4~5) images of a subject.
+The `train_custom_diffusion.py` script shows how to implement the training procedure and adapt it for stable diffusion.
+
+## Running locally with PyTorch
+
+### Installing the dependencies
+
+Before running the scripts, make sure to install the library's training dependencies:
+
+**Important**
+
+To make sure you can successfully run the latest versions of the example scripts, we highly recommend **installing from source** and keeping the install up to date as we update the example scripts frequently and install some example-specific requirements. To do this, execute the following steps in a new virtual environment:
+
+```bash
+git clone https://github.com/huggingface/diffusers
+cd diffusers
+pip install -e .
+```
+
+Then cd in the example folder and run
+
+```bash
+pip install -r requirements.txt
+pip install clip-retrieval 
+```
+
+And initialize an [🤗Accelerate](https://github.com/huggingface/accelerate/) environment with:
+
+```bash
+accelerate config
+```
+
+Or for a default accelerate configuration without answering questions about your environment
+
+```bash
+accelerate config default
+```
+
+Or if your environment doesn't support an interactive shell e.g. a notebook
+
+```python
+from accelerate.utils import write_basic_config
+
+write_basic_config()
+```
+### Cat example 😺
+
+Now let's get our dataset. Download dataset from [here](https://www.cs.cmu.edu/~custom-diffusion/assets/data.zip) and unzip it. 
+
+We also collect 200 real images using `clip-retrieval` which are combined with the target images in the training dataset as a regularization. This prevents overfitting to the the given target image. The following flags enable the regularization `with_prior_preservation`, `real_prior` with `prior_loss_weight=1.`. 
+The `class_prompt` should be the category name same as target image. The collected real images are with text captions similar to the `class_prompt`. The retrieved image are saved in `class_data_dir`. You can disable `real_prior` to use generated images as regularization. To collect the real images use this command first before training. 
+
+```bash
+pip install clip-retrieval
+python retrieve.py --class_prompt cat --class_data_dir real_reg/samples_cat --num_class_images 200
+```
+
+**___Note: Change the `resolution` to 768 if you are using the [stable-diffusion-2](https://huggingface.co/stabilityai/stable-diffusion-2) 768x768 model.___**
+
+```bash
+export MODEL_NAME="CompVis/stable-diffusion-v1-4"
+export OUTPUT_DIR="path-to-save-model"
+export INSTANCE_DIR="./data/cat"
+
+accelerate launch train_custom_diffusion.py \
+  --pretrained_model_name_or_path=$MODEL_NAME  \
+  --instance_data_dir=$INSTANCE_DIR \
+  --output_dir=$OUTPUT_DIR \
+  --class_data_dir=./real_reg/samples_cat/ \
+  --with_prior_preservation --real_prior --prior_loss_weight=1.0 \
+  --class_prompt="cat" --num_class_images=200 \
+  --instance_prompt="photo of a <new1> cat"  \
+  --resolution=512  \
+  --train_batch_size=2  \
+  --learning_rate=1e-5  \
+  --lr_warmup_steps=0 \
+  --max_train_steps=250 \
+  --scale_lr --hflip  \
+  --modifier_token "<new1>" 
+```
+
+**Use `--enable_xformers_memory_efficient_attention` for faster training with lower VRAM requirement (16GB per GPU). Follow [this guide](https://github.com/facebookresearch/xformers) for installation instructions.**
+
+To track your experiments using Weights and Biases (`wandb`) and to save intermediate results (whcih we HIGHLY recommend), follow these steps:
+
+* Install `wandb`: `pip install wandb`.
+* Authorize: `wandb login`. 
+* Then specify a `validation_prompt` and set `report_to` to `wandb` while launching training. You can also configure the following related arguments:
+    * `num_validation_images`
+    * `validation_steps`
+
+Here is an example command:
+
+```bash
+accelerate launch train_custom_diffusion.py \
+  --pretrained_model_name_or_path=$MODEL_NAME  \
+  --instance_data_dir=$INSTANCE_DIR \
+  --output_dir=$OUTPUT_DIR \
+  --class_data_dir=./real_reg/samples_cat/ \
+  --with_prior_preservation --real_prior --prior_loss_weight=1.0 \
+  --class_prompt="cat" --num_class_images=200 \
+  --instance_prompt="photo of a <new1> cat"  \
+  --resolution=512  \
+  --train_batch_size=2  \
+  --learning_rate=1e-5  \
+  --lr_warmup_steps=0 \
+  --max_train_steps=250 \
+  --scale_lr --hflip  \
+  --modifier_token "<new1>" \
+  --validation_prompt="<new1> cat sitting in a bucket" \
+  --report_to="wandb"
+```
+
+Here is an example [Weights and Biases page](https://wandb.ai/sayakpaul/custom-diffusion/runs/26ghrcau) where you can check out the intermediate results along with other training details.  
+
+If you specify `--push_to_hub`, the learned parameters will be pushed to a repository on the Hugging Face Hub. Here is an [example repository](https://huggingface.co/sayakpaul/custom-diffusion-cat).
+
+### Training on multiple concepts 🐱🪵
+
+Provide a [json](https://github.com/adobe-research/custom-diffusion/blob/main/assets/concept_list.json) file with the info about each concept, similar to [this](https://github.com/ShivamShrirao/diffusers/blob/main/examples/dreambooth/train_dreambooth.py).
+
+To collect the real images run this command for each concept in the json file. 
+
+```bash
+pip install clip-retrieval
+python retrieve.py --class_prompt {} --class_data_dir {} --num_class_images 200
+```
+
+And then we're ready to start training!
+
+```bash
+export MODEL_NAME="CompVis/stable-diffusion-v1-4"
+export OUTPUT_DIR="path-to-save-model"
+
+accelerate launch train_custom_diffusion.py \
+  --pretrained_model_name_or_path=$MODEL_NAME  \
+  --output_dir=$OUTPUT_DIR \
+  --concepts_list=./concept_list.json \
+  --with_prior_preservation --real_prior --prior_loss_weight=1.0 \
+  --resolution=512  \
+  --train_batch_size=2  \
+  --learning_rate=1e-5  \
+  --lr_warmup_steps=0 \
+  --max_train_steps=500 \
+  --num_class_images=200 \
+  --scale_lr --hflip  \
+  --modifier_token "<new1>+<new2>" 
+```
+
+Here is an example [Weights and Biases page](https://wandb.ai/sayakpaul/custom-diffusion/runs/3990tzkg) where you can check out the intermediate results along with other training details.  
+
+### Training on human faces
+
+For fine-tuning on human faces we found the following configuration to work better: `learning_rate=5e-6`, `max_train_steps=1000 to 2000`, and `freeze_model=crossattn` with at least 15-20 images. 
+
+To collect the real images use this command first before training. 
+
+```bash
+pip install clip-retrieval
+python retrieve.py --class_prompt person --class_data_dir real_reg/samples_person --num_class_images 200
+```
+
+Then start training!
+
+```bash
+export MODEL_NAME="CompVis/stable-diffusion-v1-4"
+export OUTPUT_DIR="path-to-save-model"
+export INSTANCE_DIR="path-to-images"
+
+accelerate launch train_custom_diffusion.py \
+  --pretrained_model_name_or_path=$MODEL_NAME  \
+  --instance_data_dir=$INSTANCE_DIR \
+  --output_dir=$OUTPUT_DIR \
+  --class_data_dir=./real_reg/samples_person/ \
+  --with_prior_preservation --real_prior --prior_loss_weight=1.0 \
+  --class_prompt="person" --num_class_images=200 \
+  --instance_prompt="photo of a <new1> person"  \
+  --resolution=512  \
+  --train_batch_size=2  \
+  --learning_rate=5e-6  \
+  --lr_warmup_steps=0 \
+  --max_train_steps=1000 \
+  --scale_lr --hflip --noaug \
+  --freeze_model crossattn \
+  --modifier_token "<new1>" \
+  --enable_xformers_memory_efficient_attention 
+```
+
+## Inference
+
+Once you have trained a model using the above command, you can run inference using the below command. Make sure to include the `modifier token` (e.g. \<new1\> in above example) in your prompt.
+
+```python
+import torch
+from diffusers import DiffusionPipeline
+
+pipe = DiffusionPipeline.from_pretrained("CompVis/stable-diffusion-v1-4", torch_dtype=torch.float16).to("cuda")
+pipe.unet.load_attn_procs("path-to-save-model", weight_name="pytorch_custom_diffusion_weights.bin")
+pipe.load_textual_inversion("path-to-save-model", weight_name="<new1>.bin")
+
+image = pipe(
+    "<new1> cat sitting in a bucket",
+    num_inference_steps=100,
+    guidance_scale=6.0,
+    eta=1.0,
+).images[0]
+image.save("cat.png")
+```
+
+It's possible to directly load these parameters from a Hub repository:
+
+```python
+import torch
+from huggingface_hub.repocard import RepoCard
+from diffusers import DiffusionPipeline
+
+model_id = "sayakpaul/custom-diffusion-cat"
+card = RepoCard.load(model_id)
+base_model_id = card.data.to_dict()["base_model"]
+
+pipe = DiffusionPipeline.from_pretrained(base_model_id, torch_dtype=torch.float16).to("cuda")
+pipe.unet.load_attn_procs(model_id, weight_name="pytorch_custom_diffusion_weights.bin")
+pipe.load_textual_inversion(model_id, weight_name="<new1>.bin")
+
+image = pipe(
+    "<new1> cat sitting in a bucket",
+    num_inference_steps=100,
+    guidance_scale=6.0,
+    eta=1.0,
+).images[0]
+image.save("cat.png")
+```
+
+Here is an example of performing inference with multiple concepts:
+
+```python
+import torch
+from huggingface_hub.repocard import RepoCard
+from diffusers import DiffusionPipeline
+
+model_id = "sayakpaul/custom-diffusion-cat-wooden-pot"
+card = RepoCard.load(model_id)
+base_model_id = card.data.to_dict()["base_model"]
+
+pipe = DiffusionPipeline.from_pretrained(base_model_id, torch_dtype=torch.float16).to("cuda")
+pipe.unet.load_attn_procs(model_id, weight_name="pytorch_custom_diffusion_weights.bin")
+pipe.load_textual_inversion(model_id, weight_name="<new1>.bin")
+pipe.load_textual_inversion(model_id, weight_name="<new2>.bin")
+
+image = pipe(
+    "the <new1> cat sculpture in the style of a <new2> wooden pot",
+    num_inference_steps=100,
+    guidance_scale=6.0,
+    eta=1.0,
+).images[0]
+image.save("multi-subject.png")
+```
+
+Here, `cat` and `wooden pot` refer to the multiple concepts.
+
+### Inference from a training checkpoint
+
+You can also perform inference from one of the complete checkpoint saved during the training process, if you used the `--checkpointing_steps` argument. 
+
+TODO.
+
+## Set grads to none
+
+To save even more memory, pass the `--set_grads_to_none` argument to the script. This will set grads to None instead of zero. However, be aware that it changes certain behaviors, so if you start experiencing any problems, remove this argument.
+
+More info: https://pytorch.org/docs/stable/generated/torch.optim.Optimizer.zero_grad.html
+
+## Experimental results
+
+You can refer to [our webpage](https://www.cs.cmu.edu/~custom-diffusion/) that discusses our experiments in detail. 
@@ -60,7 +60,18 @@ DreamBooth finetuning is very sensitive to hyperparameters and easy to overfit.

 <frameworkcontent>
 <pt>
-Let's try DreamBooth with a [few images of a dog](https://drive.google.com/drive/folders/1BO_dyz-p65qhBRRMRA4TbZ8qW4rB99JZ); download and save them to a directory and then set the `INSTANCE_DIR` environment variable to that path:
+Let's try DreamBooth with a
+[few images of a dog](https://huggingface.co/datasets/diffusers/dog-example);
+download and save them to a directory and then set the `INSTANCE_DIR` environment variable to that path:
+
+```python 
+local_dir = "./path_to_training_images"
+snapshot_download(
+    "diffusers/dog-example",
+    local_dir=local_dir, repo_type="dataset",
+    ignore_patterns=".gitattributes",
+)
+```

 ```bash
 export MODEL_NAME="CompVis/stable-diffusion-v1-4"
@@ -118,7 +129,7 @@ python train_dreambooth_flax.py \

 Prior preservation is used to avoid overfitting and language-drift (check out the [paper](https://arxiv.org/abs/2208.12242) to learn more if you're interested). For prior preservation, you use other images of the same class as part of the training process. The nice thing is that you can generate those images using the Stable Diffusion model itself! The training script will save the generated images to a local path you specify.

-The author's recommend generating `num_epochs * num_samples` images for prior preservation. In most cases, 200-300 images work well.
+The authors recommend generating `num_epochs * num_samples` images for prior preservation. In most cases, 200-300 images work well.

 <frameworkcontent>
 <pt>
@@ -237,7 +248,7 @@ python train_dreambooth_flax.py \

 ## Finetuning with LoRA

-You can also use Low-Rank Adaptation of Large Language Models (LoRA), a fine-tuning technique for accelerating training large models, on DreamBooth. For more details, take a look at the [LoRA training](training/lora#dreambooth) guide.
+You can also use Low-Rank Adaptation of Large Language Models (LoRA), a fine-tuning technique for accelerating training large models, on DreamBooth. For more details, take a look at the [LoRA training](./lora#dreambooth) guide.

 ## Saving checkpoints while training

@@ -321,7 +332,7 @@ Depending on your hardware, there are a few different ways to optimize DreamBoot

 ### xFormers

-[xFormers](https://github.com/facebookresearch/xformers) is a toolbox for optimizing Transformers, and it include a [memory-efficient attention](https://facebookresearch.github.io/xformers/components/ops.html#module-xformers.ops) mechanism that is used in 🧨 Diffusers. You'll need to [install xFormers](./optimization/xformers) and then add the following argument to your training script:
+[xFormers](https://github.com/facebookresearch/xformers) is a toolbox for optimizing Transformers, and it includes a [memory-efficient attention](https://facebookresearch.github.io/xformers/components/ops.html#module-xformers.ops) mechanism that is used in 🧨 Diffusers. You'll need to [install xFormers](./optimization/xformers) and then add the following argument to your training script:

 ```bash
  --enable_xformers_memory_efficient_attention
@@ -457,11 +468,11 @@ If you have **`"accelerate>=0.16.0"`** installed, you can use the following code
 inference from an intermediate checkpoint:

 ```python
-from diffusers import StableDiffusionPipeline
+from diffusers import DiffusionPipeline
 import torch

 model_id = "path_to_saved_model"
-pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16).to("cuda")
+pipe = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16).to("cuda")

 prompt = "A photo of sks dog in a bucket"
 image = pipe(prompt, num_inference_steps=50, guidance_scale=7.5).images[0]
@@ -469,4 +480,4 @@ image = pipe(prompt, num_inference_steps=50, guidance_scale=7.5).images[0]
 image.save("dog-bucket.png")
 ```

-You may also run inference from any of the [saved training checkpoints](#inference-from-a-saved-checkpoint).
+You may also run inference from any of the [saved training checkpoints](#inference-from-a-saved-checkpoint).
@@ -0,0 +1,202 @@
+<!--Copyright 2023 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
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# InstructPix2Pix 
+
+[InstructPix2Pix](https://arxiv.org/abs/2211.09800) is a method to fine-tune text-conditioned diffusion models such that they can follow an edit instruction for an input image. Models fine-tuned using this method take the following as inputs:
+
+<p align="center">
+    <img src="https://huggingface.co/datasets/diffusers/docs-images/resolve/main/evaluation_diffusion_models/edit-instruction.png" alt="instructpix2pix-inputs" width=600/>
+</p>
+
+The output is an "edited" image that reflects the edit instruction applied on the input image:
+
+<p align="center">
+    <img src="https://huggingface.co/datasets/diffusers/docs-images/resolve/main/output-gs%407-igs%401-steps%4050.png" alt="instructpix2pix-output" width=600/>
+</p>
+
+The `train_instruct_pix2pix.py` script shows how to implement the training procedure and adapt it for Stable Diffusion.
+
+***Disclaimer: Even though `train_instruct_pix2pix.py` implements the InstructPix2Pix
+training procedure while being faithful to the [original implementation](https://github.com/timothybrooks/instruct-pix2pix) we have only tested it on a [small-scale dataset](https://huggingface.co/datasets/fusing/instructpix2pix-1000-samples). This can impact the end results. For better results, we recommend longer training runs with a larger dataset. [Here](https://huggingface.co/datasets/timbrooks/instructpix2pix-clip-filtered) you can find a large dataset for InstructPix2Pix training.***
+
+## Running locally with PyTorch
+
+### Installing the dependencies
+
+Before running the scripts, make sure to install the library's training dependencies:
+
+**Important**
+
+To make sure you can successfully run the latest versions of the example scripts, we highly recommend **installing from source** and keeping the install up to date as we update the example scripts frequently and install some example-specific requirements. To do this, execute the following steps in a new virtual environment:
+```bash
+git clone https://github.com/huggingface/diffusers
+cd diffusers
+pip install -e .
+```
+
+Then cd in the example folder and run
+```bash
+pip install -r requirements.txt
+```
+
+And initialize an [🤗Accelerate](https://github.com/huggingface/accelerate/) environment with:
+
+```bash
+accelerate config
+```
+
+Or for a default accelerate configuration without answering questions about your environment
+
+```bash
+accelerate config default
+```
+
+Or if your environment doesn't support an interactive shell e.g. a notebook
+
+```python
+from accelerate.utils import write_basic_config
+
+write_basic_config()
+```
+
+### Toy example
+
+As mentioned before, we'll use a [small toy dataset](https://huggingface.co/datasets/fusing/instructpix2pix-1000-samples) for training. The dataset 
+is a smaller version of the [original dataset](https://huggingface.co/datasets/timbrooks/instructpix2pix-clip-filtered) used in the InstructPix2Pix paper.
+
+Configure environment variables such as the dataset identifier and the Stable Diffusion
+checkpoint:
+
+```bash
+export MODEL_NAME="runwayml/stable-diffusion-v1-5"
+export DATASET_ID="fusing/instructpix2pix-1000-samples"
+```
+
+Now, we can launch training:
+
+```bash
+accelerate launch --mixed_precision="fp16" train_instruct_pix2pix.py \
+    --pretrained_model_name_or_path=$MODEL_NAME \
+    --dataset_name=$DATASET_ID \
+    --enable_xformers_memory_efficient_attention \
+    --resolution=256 --random_flip \
+    --train_batch_size=4 --gradient_accumulation_steps=4 --gradient_checkpointing \
+    --max_train_steps=15000 \
+    --checkpointing_steps=5000 --checkpoints_total_limit=1 \
+    --learning_rate=5e-05 --max_grad_norm=1 --lr_warmup_steps=0 \
+    --conditioning_dropout_prob=0.05 \
+    --mixed_precision=fp16 \
+    --seed=42 
+```
+
+Additionally, we support performing validation inference to monitor training progress
+with Weights and Biases. You can enable this feature with `report_to="wandb"`:
+
+```bash
+accelerate launch --mixed_precision="fp16" train_instruct_pix2pix.py \
+    --pretrained_model_name_or_path=$MODEL_NAME \
+    --dataset_name=$DATASET_ID \
+    --enable_xformers_memory_efficient_attention \
+    --resolution=256 --random_flip \
+    --train_batch_size=4 --gradient_accumulation_steps=4 --gradient_checkpointing \
+    --max_train_steps=15000 \
+    --checkpointing_steps=5000 --checkpoints_total_limit=1 \
+    --learning_rate=5e-05 --max_grad_norm=1 --lr_warmup_steps=0 \
+    --conditioning_dropout_prob=0.05 \
+    --mixed_precision=fp16 \
+    --val_image_url="https://hf.co/datasets/diffusers/diffusers-images-docs/resolve/main/mountain.png" \
+    --validation_prompt="make the mountains snowy" \
+    --seed=42 \
+    --report_to=wandb 
+ ```
+
+ We recommend this type of validation as it can be useful for model debugging. Note that you need `wandb` installed to use this. You can install `wandb` by running `pip install wandb`. 
+
+ [Here](https://wandb.ai/sayakpaul/instruct-pix2pix/runs/ctr3kovq), you can find an example training run that includes some validation samples and the training hyperparameters.
+
+ ***Note: In the original paper, the authors observed that even when the model is trained with an image resolution of 256x256, it generalizes well to bigger resolutions such as 512x512. This is likely because of the larger dataset they used during training.***
+
+ ## Training with multiple GPUs
+
+`accelerate` allows for seamless multi-GPU training. Follow the instructions [here](https://huggingface.co/docs/accelerate/basic_tutorials/launch)
+for running distributed training with `accelerate`. Here is an example command:
+
+```bash 
+accelerate launch --mixed_precision="fp16" --multi_gpu train_instruct_pix2pix.py \
+ --pretrained_model_name_or_path=runwayml/stable-diffusion-v1-5 \
+ --dataset_name=sayakpaul/instructpix2pix-1000-samples \
+ --use_ema \
+ --enable_xformers_memory_efficient_attention \
+ --resolution=512 --random_flip \
+ --train_batch_size=4 --gradient_accumulation_steps=4 --gradient_checkpointing \
+ --max_train_steps=15000 \
+ --checkpointing_steps=5000 --checkpoints_total_limit=1 \
+ --learning_rate=5e-05 --lr_warmup_steps=0 \
+ --conditioning_dropout_prob=0.05 \
+ --mixed_precision=fp16 \
+ --seed=42 
+```
+
+ ## Inference
+
+ Once training is complete, we can perform inference:
+
+ ```python
+import PIL
+import requests
+import torch
+from diffusers import StableDiffusionInstructPix2PixPipeline
+
+model_id = "your_model_id"  # <- replace this
+pipe = StableDiffusionInstructPix2PixPipeline.from_pretrained(model_id, torch_dtype=torch.float16).to("cuda")
+generator = torch.Generator("cuda").manual_seed(0)
+
+url = "https://huggingface.co/datasets/sayakpaul/sample-datasets/resolve/main/test_pix2pix_4.png"
+
+
+def download_image(url):
+    image = PIL.Image.open(requests.get(url, stream=True).raw)
+    image = PIL.ImageOps.exif_transpose(image)
+    image = image.convert("RGB")
+    return image
+
+
+image = download_image(url)
+prompt = "wipe out the lake"
+num_inference_steps = 20
+image_guidance_scale = 1.5
+guidance_scale = 10
+
+edited_image = pipe(
+    prompt,
+    image=image,
+    num_inference_steps=num_inference_steps,
+    image_guidance_scale=image_guidance_scale,
+    guidance_scale=guidance_scale,
+    generator=generator,
+).images[0]
+edited_image.save("edited_image.png")
+```
+
+An example model repo obtained using this training script can be found
+here - [sayakpaul/instruct-pix2pix](https://huggingface.co/sayakpaul/instruct-pix2pix).
+
+We encourage you to play with the following three parameters to control
+speed and quality during performance:
+
+* `num_inference_steps`
+* `image_guidance_scale`
+* `guidance_scale`
+
+Particularly, `image_guidance_scale` and `guidance_scale` can have a profound impact
+on the generated ("edited") image (see [here](https://twitter.com/RisingSayak/status/1628392199196151808?s=20) for an example).
@@ -16,7 +16,9 @@ specific language governing permissions and limitations under the License.

 <Tip warning={true}>

-Currently, LoRA is only supported for the attention layers of the [`UNet2DConditionalModel`].
+Currently, LoRA is only supported for the attention layers of the [`UNet2DConditionalModel`]. We also 
+support LoRA fine-tuning of the text encoder for DreamBooth in a limited capacity. For more details on how we support 
+LoRA fine-tuning of the text encoder, refer to the discussion on [this PR](https://github.com/huggingface/diffusers/pull/2918). 

 </Tip>

@@ -175,6 +177,11 @@ accelerate launch train_dreambooth_lora.py \
  --push_to_hub
 ```

+It's also possible to additionally fine-tune the text encoder with LoRA. This, in most cases, leads
+to better results with a slight increase in the compute. To allow fine-tuning the text encoder with LoRA,
+specify the `--train_text_encoder` while launching the `train_dreambooth_lora.py` script. 
+
+
 ### Inference[[dreambooth-inference]]

 Now you can use the model for inference by loading the base model in the [`StableDiffusionPipeline`]:
@@ -38,6 +38,9 @@ Training examples show how to pretrain or fine-tune diffusion models for a varie
 - [Text Inversion](./text_inversion)
 - [Dreambooth](./dreambooth)
 - [LoRA Support](./lora)
+- [ControlNet](./controlnet)
+- [InstructPix2Pix](./instructpix2pix)
+- [Custom Diffusion](./custom_diffusion)

 If possible, please [install xFormers](../optimization/xformers) for memory efficient attention. This could help make your training faster and less memory intensive.

@@ -47,6 +50,10 @@ If possible, please [install xFormers](../optimization/xformers) for memory effi
 | [**Text-to-Image fine-tuning**](./text2image) | ✅ | ✅ | 
 | [**Textual Inversion**](./text_inversion) | ✅ | - | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_textual_inversion_training.ipynb)
 | [**Dreambooth**](./dreambooth) | ✅ | - | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_dreambooth_training.ipynb)
+| [**Training with LoRA**](./lora) | ✅ | - | - |
+| [**ControlNet**](./controlnet) | ✅ | ✅ | - |
+| [**InstructPix2Pix**](./instructpix2pix) | ✅ | ✅ | - |
+| [**Custom Diffusion**](./custom_diffusion) | ✅ | ✅ | - |

 ## Community

@@ -74,25 +74,13 @@ To load a checkpoint to resume training, pass the argument `--resume_from_checkp
 <pt>
 Launch the [PyTorch training script](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py) for a fine-tuning run on the [Pokémon BLIP captions](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions) dataset like this:

-```bash
-export MODEL_NAME="CompVis/stable-diffusion-v1-4"
-export dataset_name="lambdalabs/pokemon-blip-captions"
-
-accelerate launch train_text_to_image.py \
-  --pretrained_model_name_or_path=$MODEL_NAME \
-  --dataset_name=$dataset_name \
-  --use_ema \
-  --resolution=512 --center_crop --random_flip \
-  --train_batch_size=1 \
-  --gradient_accumulation_steps=4 \
-  --gradient_checkpointing \
-  --mixed_precision="fp16" \
-  --max_train_steps=15000 \
-  --learning_rate=1e-05 \
-  --max_grad_norm=1 \
-  --lr_scheduler="constant" --lr_warmup_steps=0 \
-  --output_dir="sd-pokemon-model" 
-```
+<literalinclude>
+{"path": "../../../../examples/text_to_image/README.md",
+"language": "bash",
+"start-after": "accelerate_snippet_start",
+"end-before": "accelerate_snippet_end",
+"dedent": 0}
+</literalinclude>

 To finetune on your own dataset, prepare the dataset according to the format required by 🤗 [Datasets](https://huggingface.co/docs/datasets/index). You can [upload your dataset to the Hub](https://huggingface.co/docs/datasets/image_dataset#upload-dataset-to-the-hub), or you can [prepare a local folder with your files](https://huggingface.co/docs/datasets/image_dataset#imagefolder).

@@ -118,6 +106,31 @@ accelerate launch train_text_to_image.py \
  --lr_scheduler="constant" --lr_warmup_steps=0 \
  --output_dir=${OUTPUT_DIR}
 ```
+
+#### Training with multiple GPUs
+
+`accelerate` allows for seamless multi-GPU training. Follow the instructions [here](https://huggingface.co/docs/accelerate/basic_tutorials/launch)
+for running distributed training with `accelerate`. Here is an example command:
+
+```bash
+export MODEL_NAME="CompVis/stable-diffusion-v1-4"
+export dataset_name="lambdalabs/pokemon-blip-captions"
+
+accelerate launch --mixed_precision="fp16" --multi_gpu  train_text_to_image.py \
+  --pretrained_model_name_or_path=$MODEL_NAME \
+  --dataset_name=$dataset_name \
+  --use_ema \
+  --resolution=512 --center_crop --random_flip \
+  --train_batch_size=1 \
+  --gradient_accumulation_steps=4 \
+  --gradient_checkpointing \
+  --max_train_steps=15000 \ 
+  --learning_rate=1e-05 \
+  --max_grad_norm=1 \
+  --lr_scheduler="constant" --lr_warmup_steps=0 \
+  --output_dir="sd-pokemon-model" 
+```
+
 </pt>
 <jax>
 With Flax, it's possible to train a Stable Diffusion model faster on TPUs and GPUs thanks to [@duongna211](https://github.com/duongna21). This is very efficient on TPU hardware but works great on GPUs too. The Flax training script doesn't support features like gradient checkpointing or gradient accumulation yet, so you'll need a GPU with at least 30GB of memory or a TPU v3.
@@ -167,6 +180,28 @@ python train_text_to_image_flax.py \
 </jax>
 </frameworkcontent>

+## Training with Min-SNR weighting
+
+We support training with the Min-SNR weighting strategy proposed in [Efficient Diffusion Training via Min-SNR Weighting Strategy](https://arxiv.org/abs/2303.09556) which helps to achieve faster convergence
+by rebalancing the loss. In order to use it, one needs to set the `--snr_gamma` argument. The recommended
+value when using it is 5.0. 
+
+You can find [this project on Weights and Biases](https://wandb.ai/sayakpaul/text2image-finetune-minsnr) that compares the loss surfaces of the following setups:
+
+* Training without the Min-SNR weighting strategy
+* Training with the Min-SNR weighting strategy (`snr_gamma` set to 5.0)
+* Training with the Min-SNR weighting strategy (`snr_gamma` set to 1.0)
+
+For our small Pokemons dataset, the effects of Min-SNR weighting strategy might not appear to be pronounced, but for larger datasets, we believe the effects will be more pronounced.
+
+Also, note that in this example, we either predict `epsilon` (i.e., the noise) or the `v_prediction`. For both of these cases, the formulation of the Min-SNR weighting strategy that we have used holds. 
+
+<Tip warning={true}>
+
+Training with Min-SNR weighting strategy is only supported in PyTorch.
+
+</Tip>
+
 ## LoRA

 You can also use Low-Rank Adaptation of Large Language Models (LoRA), a fine-tuning technique for accelerating training large models, for fine-tuning text-to-image models. For more details, take a look at the [LoRA training](lora#text-to-image) guide.
@@ -19,7 +19,7 @@ specific language governing permissions and limitations under the License.
 [Textual Inversion](https://arxiv.org/abs/2208.01618) is a technique for capturing novel concepts from a small number of example images. While the technique was originally demonstrated with a [latent diffusion model](https://github.com/CompVis/latent-diffusion), it has since been applied to other model variants like [Stable Diffusion](https://huggingface.co/docs/diffusers/main/en/conceptual/stable_diffusion). The learned concepts can be used to better control the images generated from text-to-image pipelines. It learns new "words" in the text encoder's embedding space, which are used within text prompts for personalized image generation.

 ![Textual Inversion example](https://textual-inversion.github.io/static/images/editing/colorful_teapot.JPG)
-<small>By using just 3-5 images you can teach new concepts to a model such as Stable Diffusion for personalized image generation <a href="https://github.com/rinongal/textual_inversion">(image source)</a></small>
+<small>By using just 3-5 images you can teach new concepts to a model such as Stable Diffusion for personalized image generation <a href="https://github.com/rinongal/textual_inversion">(image source)</a>.</small>

 This guide will show you how to train a [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5) model with Textual Inversion. All the training scripts for Textual Inversion used in this guide can be found [here](https://github.com/huggingface/diffusers/tree/main/examples/textual_inversion) if you're interested in taking a closer look at how things work under the hood.

@@ -157,24 +157,61 @@ If you're interested in following along with your model training progress, you c

 ## Inference

-Once you have trained a model, you can use it for inference with the [`StableDiffusionPipeline]. Make sure you include the `placeholder_token` in your prompt, in this case, it is `<cat-toy>`.
+Once you have trained a model, you can use it for inference with the [`StableDiffusionPipeline`].
+
+The textual inversion script will by default only save the textual inversion embedding vector(s) that have 
+been added to the text encoder embedding matrix and consequently been trained.

 <frameworkcontent>
 <pt>
+<Tip>
+
+💡 The community has created a large library of different textual inversion embedding vectors, called [sd-concepts-library](https://huggingface.co/sd-concepts-library).
+Instead of training textual inversion embeddings from scratch you can also see whether a fitting textual inversion embedding has already been added to the libary.
+
+</Tip>
+
+To load the textual inversion embeddings you first need to load the base model that was used when training 
+your textual inversion embedding vectors. Here we assume that [`runwayml/stable-diffusion-v1-5`](runwayml/stable-diffusion-v1-5)
+was used as a base model so we load it first:
 ```python
 from diffusers import StableDiffusionPipeline
+import torch

-model_id = "path-to-your-trained-model"
+model_id = "runwayml/stable-diffusion-v1-5"
 pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16).to("cuda")
+```

+Next, we need to load the textual inversion embedding vector which can be done via the [`TextualInversionLoaderMixin.load_textual_inversion`]
+function. Here we'll load the embeddings of the "<cat-toy>" example from before.
+```python
+pipe.load_textual_inversion("sd-concepts-library/cat-toy")
+```
+
+Now we can run the pipeline making sure that the placeholder token `<cat-toy>` is used in our prompt.
+
+```python
 prompt = "A <cat-toy> backpack"

-image = pipe(prompt, num_inference_steps=50, guidance_scale=7.5).images[0]
-
+image = pipe(prompt, num_inference_steps=50).images[0]
 image.save("cat-backpack.png")
 ```
+
+The function [`TextualInversionLoaderMixin.load_textual_inversion`] can not only 
+load textual embedding vectors saved in Diffusers' format, but also embedding vectors
+saved in [Automatic1111](https://github.com/AUTOMATIC1111/stable-diffusion-webui) format.
+To do so, you can first download an embedding vector from [civitAI](https://civitai.com/models/3036?modelVersionId=8387)
+and then load it locally:
+```python
+pipe.load_textual_inversion("./charturnerv2.pt")
+```
 </pt>
 <jax>
+Currently there is no `load_textual_inversion` function for Flax so one has to make sure the textual inversion
+embedding vector is saved as part of the model after training.
+
+The model can then be run just like any other Flax model:
+
 ```python
 import jax
 import numpy as np
@@ -212,4 +249,4 @@ image.save("cat-backpack.png")

 Usually, text prompts are tokenized into an embedding before being passed to a model, which is often a transformer. Textual Inversion does something similar, but it learns a new token embedding, `v*`, from a special token `S*` in the diagram above. The model output is used to condition the diffusion model, which helps the diffusion model understand the prompt and new concepts from just a few example images.

-To do this, Textual Inversion uses a generator model and noisy versions of the training images. The generator tries to predict less noisy versions of the images, and the token embedding `v*` is optimized based on how well the generator does. If the token embedding successfully captures the new concept, it gives more useful information to the diffusion model and helps create clearer images with less noise. This optimization process typically occurs after several thousand steps of exposure to a variety of prompt and image variants.
+To do this, Textual Inversion uses a generator model and noisy versions of the training images. The generator tries to predict less noisy versions of the images, and the token embedding `v*` is optimized based on how well the generator does. If the token embedding successfully captures the new concept, it gives more useful information to the diffusion model and helps create clearer images with less noise. This optimization process typically occurs after several thousand steps of exposure to a variety of prompt and image variants.
@@ -122,6 +122,26 @@ accelerate launch train_unconditional.py \
    <img src="https://user-images.githubusercontent.com/26864830/180248200-928953b4-db38-48db-b0c6-8b740fe6786f.png"/>
 </div>

+### Training with multiple GPUs
+
+`accelerate` allows for seamless multi-GPU training. Follow the instructions [here](https://huggingface.co/docs/accelerate/basic_tutorials/launch)
+for running distributed training with `accelerate`. Here is an example command:
+
+```bash
+accelerate launch --mixed_precision="fp16" --multi_gpu train_unconditional.py \
+  --dataset_name="huggan/pokemon" \
+  --resolution=64 --center_crop --random_flip \
+  --output_dir="ddpm-ema-pokemon-64" \
+  --train_batch_size=16 \
+  --num_epochs=100 \
+  --gradient_accumulation_steps=1 \
+  --use_ema \
+  --learning_rate=1e-4 \
+  --lr_warmup_steps=500 \
+  --mixed_precision="fp16" \
+  --logger="wandb"
+```
+
 ## Finetuning with your own data

 There are two ways to finetune a model on your own dataset:
@@ -344,7 +344,7 @@ Now you can wrap all these components together in a training loop with 🤗 Acce

 ...             # Sample a random timestep for each image
 ...             timesteps = torch.randint(
-...                 0, noise_scheduler.num_train_timesteps, (bs,), device=clean_images.device
+...                 0, noise_scheduler.config.num_train_timesteps, (bs,), device=clean_images.device
 ...             ).long()

 ...             # Add noise to the clean images according to the noise magnitude at each timestep
@@ -10,22 +10,27 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Conditional Image Generation
+# Conditional image generation
+
+[[open-in-colab]]
+
+Conditional image generation allows you to generate images from a text prompt. The text is converted into embeddings which are used to condition the model to generate an image from noise.

 The [`DiffusionPipeline`] is the easiest way to use a pre-trained diffusion system for inference.

-Start by creating an instance of [`DiffusionPipeline`] and specify which pipeline checkpoint you would like to download.
-You can use the [`DiffusionPipeline`] for any [Diffusers' checkpoint](https://huggingface.co/models?library=diffusers&sort=downloads).
-In this guide though, you'll use [`DiffusionPipeline`] for text-to-image generation with [Latent Diffusion](https://huggingface.co/CompVis/ldm-text2im-large-256):
+Start by creating an instance of [`DiffusionPipeline`] and specify which pipeline [checkpoint](https://huggingface.co/models?library=diffusers&sort=downloads) you would like to download.
+
+In this guide, you'll use [`DiffusionPipeline`] for text-to-image generation with [Latent Diffusion](https://huggingface.co/CompVis/ldm-text2im-large-256):

 ```python
 >>> from diffusers import DiffusionPipeline

 >>> generator = DiffusionPipeline.from_pretrained("CompVis/ldm-text2im-large-256")
 ```
+
 The [`DiffusionPipeline`] downloads and caches all modeling, tokenization, and scheduling components. 
-Because the model consists of roughly 1.4 billion parameters, we strongly recommend running it on GPU.
-You can move the generator object to GPU, just like you would in PyTorch.
+Because the model consists of roughly 1.4 billion parameters, we strongly recommend running it on a GPU.
+You can move the generator object to a GPU, just like you would in PyTorch:

 ```python
 >>> generator.to("cuda")
@@ -37,10 +42,19 @@ Now you can use the `generator` on your text prompt:
 >>> image = generator("An image of a squirrel in Picasso style").images[0]
 ```

-The output is by default wrapped into a [PIL Image object](https://pillow.readthedocs.io/en/stable/reference/Image.html?highlight=image#the-image-class).
+The output is by default wrapped into a [`PIL.Image`](https://pillow.readthedocs.io/en/stable/reference/Image.html?highlight=image#the-image-class) object.

-You can save the image by simply calling:
+You can save the image by calling:

 ```python
 >>> image.save("image_of_squirrel_painting.png")
 ```
+
+Try out the Spaces below, and feel free to play around with the guidance scale parameter to see how it affects the image quality!
+
+<iframe
+	src="https://stabilityai-stable-diffusion.hf.space"
+	frameborder="0"
+	width="850"
+	height="500"
+></iframe>
@@ -10,17 +10,21 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# How to build a community pipeline
+# How to contribute a community pipeline

-*Note*: this page was built from the GitHub Issue on Community Pipelines [#841](https://github.com/huggingface/diffusers/issues/841).
+<Tip>

-Let's make an example!
-Say you want to define a pipeline that just does a single forward pass to a U-Net and then calls a scheduler only once (Note, this doesn't make any sense from a scientific point of view, but only represents an example of how things work under the hood).
+💡 Take a look at GitHub Issue [#841](https://github.com/huggingface/diffusers/issues/841) for more context about why we're adding community pipelines to help everyone easily share their work without being slowed down.

-Cool! So you open your favorite IDE and start creating your pipeline 💻.
-First, what model weights and configurations do we need?
-We have a U-Net and a scheduler, so our pipeline should take a U-Net and a scheduler as an argument.
-Also, as stated above, you'd like to be able to load weights and the scheduler config for Hub and share your code with others, so we'll inherit from `DiffusionPipeline`:
+</Tip>
+
+Community pipelines allow you to add any additional features you'd like on top of the [`DiffusionPipeline`]. The main benefit of building on top of the `DiffusionPipeline` is anyone can load and use your pipeline by only adding one more argument, making it super easy for the community to access.
+
+This guide will show you how to create a community pipeline and explain how they work. To keep things simple, you'll create a "one-step" pipeline where the `UNet` does a single forward pass and calls the scheduler once.
+
+## Initialize the pipeline
+
+You should start by creating a `one_step_unet.py` file for your community pipeline. In this file, create a pipeline class that inherits from the [`DiffusionPipeline`] to be able to load model weights and the scheduler configuration from the Hub. The one-step pipeline needs a `UNet` and a scheduler, so you'll need to add these as arguments to the `__init__` function:

 ```python
 from diffusers import DiffusionPipeline
@@ -32,50 +36,52 @@ class UnetSchedulerOneForwardPipeline(DiffusionPipeline):
        super().__init__()
 ```

-Now, we must save the `unet` and `scheduler` in a config file so that you can save your pipeline with `save_pretrained`.
-Therefore, make sure you add every component that is save-able to the `register_modules` function:
+To ensure your pipeline and its components (`unet` and `scheduler`) can be saved with [`~DiffusionPipeline.save_pretrained`], add them to the `register_modules` function:

-```python
-from diffusers import DiffusionPipeline
-import torch
+```diff
+  from diffusers import DiffusionPipeline
+  import torch

+  class UnetSchedulerOneForwardPipeline(DiffusionPipeline):
+      def __init__(self, unet, scheduler):
+          super().__init__()

-class UnetSchedulerOneForwardPipeline(DiffusionPipeline):
-    def __init__(self, unet, scheduler):
-        super().__init__()
-
-        self.register_modules(unet=unet, scheduler=scheduler)
+         self.register_modules(unet=unet, scheduler=scheduler)
 ```

-Cool, the init is done! 🔥 Now, let's go into the forward pass, which we recommend defining as `__call__` . Here you're given all the creative freedom there is. For our amazing "one-step" pipeline, we simply create a random image and call the unet once and the scheduler once:
+Cool, the `__init__` step is done and you can move to the forward pass now! 🔥 

-```python
-from diffusers import DiffusionPipeline
-import torch
+## Define the forward pass
+
+In the forward pass, which we recommend defining as `__call__`, you have complete creative freedom to add whatever feature you'd like. For our amazing one-step pipeline, create a random image and only call the `unet` and `scheduler` once by setting `timestep=1`:
+
+```diff
+  from diffusers import DiffusionPipeline
+  import torch


-class UnetSchedulerOneForwardPipeline(DiffusionPipeline):
-    def __init__(self, unet, scheduler):
-        super().__init__()
+  class UnetSchedulerOneForwardPipeline(DiffusionPipeline):
+      def __init__(self, unet, scheduler):
+          super().__init__()

-        self.register_modules(unet=unet, scheduler=scheduler)
+          self.register_modules(unet=unet, scheduler=scheduler)

-    def __call__(self):
-        image = torch.randn(
-            (1, self.unet.in_channels, self.unet.sample_size, self.unet.sample_size),
-        )
-        timestep = 1
+     def __call__(self):
+         image = torch.randn(
+             (1, self.unet.config.in_channels, self.unet.config.sample_size, self.unet.config.sample_size),
+         )
+         timestep = 1

-        model_output = self.unet(image, timestep).sample
-        scheduler_output = self.scheduler.step(model_output, timestep, image).prev_sample
+         model_output = self.unet(image, timestep).sample
+         scheduler_output = self.scheduler.step(model_output, timestep, image).prev_sample

-        return scheduler_output
+         return scheduler_output
 ```

-Cool, that's it! 🚀 You can now run this pipeline by passing a `unet` and a `scheduler` to the init:
+That's it! 🚀 You can now run this pipeline by passing a `unet` and `scheduler` to it:

 ```python
-from diffusers import DDPMScheduler, Unet2DModel
+from diffusers import DDPMScheduler, UNet2DModel

 scheduler = DDPMScheduler()
 unet = UNet2DModel()
@@ -85,7 +91,7 @@ pipeline = UnetSchedulerOneForwardPipeline(unet=unet, scheduler=scheduler)
 output = pipeline()
 ```

-But what's even better is that you can load pre-existing weights into the pipeline if they match exactly your pipeline structure. This is e.g. the case for [https://huggingface.co/google/ddpm-cifar10-32](https://huggingface.co/google/ddpm-cifar10-32) so that we can do the following:
+But what's even better is you can load pre-existing weights into the pipeline if the pipeline structure is identical. For example, you can load the [`google/ddpm-cifar10-32`](https://huggingface.co/google/ddpm-cifar10-32) weights into the one-step pipeline:

 ```python
 pipeline = UnetSchedulerOneForwardPipeline.from_pretrained("google/ddpm-cifar10-32")
@@ -93,33 +99,11 @@ pipeline = UnetSchedulerOneForwardPipeline.from_pretrained("google/ddpm-cifar10-
 output = pipeline()
 ```

-We want to share this amazing pipeline with the community, so we would open a PR request to add the following code under `one_step_unet.py` to [https://github.com/huggingface/diffusers/tree/main/examples/community](https://github.com/huggingface/diffusers/tree/main/examples/community) .
+## Share your pipeline

-```python
-from diffusers import DiffusionPipeline
-import torch
+Open a Pull Request on the 🧨 Diffusers [repository](https://github.com/huggingface/diffusers) to add your awesome pipeline in `one_step_unet.py` to the [examples/community](https://github.com/huggingface/diffusers/tree/main/examples/community) subfolder.

-
-class UnetSchedulerOneForwardPipeline(DiffusionPipeline):
-    def __init__(self, unet, scheduler):
-        super().__init__()
-
-        self.register_modules(unet=unet, scheduler=scheduler)
-
-    def __call__(self):
-        image = torch.randn(
-            (1, self.unet.in_channels, self.unet.sample_size, self.unet.sample_size),
-        )
-        timestep = 1
-
-        model_output = self.unet(image, timestep).sample
-        scheduler_output = self.scheduler.step(model_output, timestep, image).prev_sample
-
-        return scheduler_output
-```
-
-Our amazing pipeline got merged here: [#840](https://github.com/huggingface/diffusers/pull/840).
-Now everybody that has `diffusers >= 0.4.0` installed can use our pipeline magically 🪄 as follows:
+Once it is merged, anyone with `diffusers >= 0.4.0` installed can use this pipeline magically 🪄 by specifying it in the `custom_pipeline` argument:

 ```python
 from diffusers import DiffusionPipeline
@@ -128,28 +112,59 @@ pipe = DiffusionPipeline.from_pretrained("google/ddpm-cifar10-32", custom_pipeli
 pipe()
 ```

-Another way to upload your custom_pipeline, besides sending a PR, is uploading the code that contains it to the Hugging Face Hub, [as exemplified here](https://huggingface.co/docs/diffusers/using-diffusers/custom_pipeline_overview#loading-custom-pipelines-from-the-hub).
+Another way to share your community pipeline is to upload the `one_step_unet.py` file directly to your preferred [model repository](https://huggingface.co/docs/hub/models-uploading) on the Hub. Instead of specifying the `one_step_unet.py` file, pass the model repository id to the `custom_pipeline` argument:

-**Try it out now - it works!**
+```python
+from diffusers import DiffusionPipeline

-In general, you will want to create much more sophisticated pipelines, so we recommend looking at existing pipelines here: [https://github.com/huggingface/diffusers/tree/main/examples/community](https://github.com/huggingface/diffusers/tree/main/examples/community).
+pipeline = DiffusionPipeline.from_pretrained("google/ddpm-cifar10-32", custom_pipeline="stevhliu/one_step_unet")
+```

-IMPORTANT:
-You can use whatever package you want in your community pipeline file - as long as the user has it installed, everything will work fine. Make sure you have one and only one pipeline class that inherits from `DiffusionPipeline` as this will be automatically detected.
+Take a look at the following table to compare the two sharing workflows to help you decide the best option for you:
+
+|                | GitHub community pipeline                                                                                        | HF Hub community pipeline                                                                 |
+|----------------|------------------------------------------------------------------------------------------------------------------|-------------------------------------------------------------------------------------------|
+| usage          | same                                                                                                             | same                                                                                      |
+| review process | open a Pull Request on GitHub and undergo a review process from the Diffusers team before merging; may be slower | upload directly to a Hub repository without any review; this is the fastest workflow      |
+| visibility     | included in the official Diffusers repository and documentation                                                  | included on your HF Hub profile and relies on your own usage/promotion to gain visibility |
+
+<Tip>
+
+💡 You can use whatever package you want in your community pipeline file - as long as the user has it installed, everything will work fine. Make sure you have one and only one pipeline class that inherits from `DiffusionPipeline` because this is automatically detected.
+
+</Tip>

 ## How do community pipelines work?
-A community pipeline is a class that has to inherit from ['DiffusionPipeline']:
-and that has been added to `examples/community` [files](https://github.com/huggingface/diffusers/tree/main/examples/community).
-The community can load the pipeline code via the custom_pipeline argument from DiffusionPipeline. See docs [here](https://huggingface.co/docs/diffusers/api/diffusion_pipeline#diffusers.DiffusionPipeline.from_pretrained.custom_pipeline):

-This means:
-The model weights and configs of the pipeline should be loaded from the `pretrained_model_name_or_path` [argument](https://huggingface.co/docs/diffusers/api/diffusion_pipeline#diffusers.DiffusionPipeline.from_pretrained.pretrained_model_name_or_path):
-whereas the code that powers the community pipeline is defined in a file added in [`examples/community`](https://github.com/huggingface/diffusers/tree/main/examples/community).
+A community pipeline is a class that inherits from [`DiffusionPipeline`] which means:

-Now, it might very well be that only some of your pipeline components weights can be downloaded from an official repo.
-The other components should then be passed directly to init as is the case for the ClIP guidance notebook [here](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/CLIP_Guided_Stable_diffusion_with_diffusers.ipynb#scrollTo=z9Kglma6hjki).
+- It can be loaded with the [`custom_pipeline`] argument.
+- The model weights and scheduler configuration are loaded from [`pretrained_model_name_or_path`].
+- The code that implements a feature in the community pipeline is defined in a `pipeline.py` file.

-The magic behind all of this is that we load the code directly from GitHub. You can check it out in more detail if you follow the functionality defined here:
+Sometimes you can't load all the pipeline components weights from an official repository. In this case, the other components should be passed directly to the pipeline:
+
+```python
+from diffusers import DiffusionPipeline
+from transformers import CLIPFeatureExtractor, CLIPModel
+
+model_id = "CompVis/stable-diffusion-v1-4"
+clip_model_id = "laion/CLIP-ViT-B-32-laion2B-s34B-b79K"
+
+feature_extractor = CLIPFeatureExtractor.from_pretrained(clip_model_id)
+clip_model = CLIPModel.from_pretrained(clip_model_id, torch_dtype=torch.float16)
+
+pipeline = DiffusionPipeline.from_pretrained(
+    model_id,
+    custom_pipeline="clip_guided_stable_diffusion",
+    clip_model=clip_model,
+    feature_extractor=feature_extractor,
+    scheduler=scheduler,
+    torch_dtype=torch.float16,
+)
+```
+
+The magic behind community pipelines is contained in the following code. It allows the community pipeline to be loaded from GitHub or the Hub, and it'll be available to all 🧨 Diffusers packages.

 ```python
 # 2. Load the pipeline class, if using custom module then load it from the hub
@@ -164,6 +179,3 @@ else:
    diffusers_module = importlib.import_module(cls.__module__.split(".")[0])
    pipeline_class = getattr(diffusers_module, config_dict["_class_name"])
 ```
-
-This is why a community pipeline merged to GitHub will be directly available to all `diffusers` packages.
-
@@ -10,7 +10,7 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Custom Pipelines
+# Community pipelines

 > **For more information about community pipelines, please have a look at [this issue](https://github.com/huggingface/diffusers/issues/841).**

@@ -45,11 +45,11 @@ The following code requires roughly 12GB of GPU RAM.

 ```python
 from diffusers import DiffusionPipeline
-from transformers import CLIPFeatureExtractor, CLIPModel
+from transformers import CLIPImageProcessor, CLIPModel
 import torch


-feature_extractor = CLIPFeatureExtractor.from_pretrained("laion/CLIP-ViT-B-32-laion2B-s34B-b79K")
+feature_extractor = CLIPImageProcessor.from_pretrained("laion/CLIP-ViT-B-32-laion2B-s34B-b79K")
 clip_model = CLIPModel.from_pretrained("laion/CLIP-ViT-B-32-laion2B-s34B-b79K", torch_dtype=torch.float16)


@@ -10,19 +10,21 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Loading and Adding Custom Pipelines
+# Load community pipelines

-Diffusers allows you to conveniently load any custom pipeline from the Hugging Face Hub as well as any [official community pipeline](https://github.com/huggingface/diffusers/tree/main/examples/community) 
-via the [`DiffusionPipeline`] class.
+Community pipelines are any [`DiffusionPipeline`] class that are different from the original implementation as specified in their paper (for example, the [`StableDiffusionControlNetPipeline`] corresponds to the [Text-to-Image Generation with ControlNet Conditioning](https://arxiv.org/abs/2302.05543) paper). They provide additional functionality or extend the original implementation of a pipeline.

-## Loading custom pipelines from the Hub
+There are many cool community pipelines like [Speech to Image](https://github.com/huggingface/diffusers/tree/main/examples/community#speech-to-image) or [Composable Stable Diffusion](https://github.com/huggingface/diffusers/tree/main/examples/community#composable-stable-diffusion), and you can find all the official community pipelines [here](https://github.com/huggingface/diffusers/tree/main/examples/community).

-Custom pipelines can be easily loaded from any model repository on the Hub that defines a diffusion pipeline in a `pipeline.py` file.
-Let's load a dummy pipeline from [hf-internal-testing/diffusers-dummy-pipeline](https://huggingface.co/hf-internal-testing/diffusers-dummy-pipeline).
+To load any community pipeline on the Hub, pass the repository id of the community pipeline to the `custom_pipeline` argument and the model repository where you'd like to load the pipeline weights and components from. For example, the example below loads a dummy pipeline from [`hf-internal-testing/diffusers-dummy-pipeline`](https://huggingface.co/hf-internal-testing/diffusers-dummy-pipeline/blob/main/pipeline.py) and the pipeline weights and components from [`google/ddpm-cifar10-32`](https://huggingface.co/google/ddpm-cifar10-32):

-All you need to do is pass the custom pipeline repo id with the `custom_pipeline` argument alongside the repo from where you wish to load the pipeline modules.
+<Tip warning={true}>

-```python
+🔒 By loading a community pipeline from the Hugging Face Hub, you are trusting that the code you are loading is safe. Make sure to inspect the code online before loading and running it automatically!
+
+</Tip>
+
+```py
 from diffusers import DiffusionPipeline

 pipeline = DiffusionPipeline.from_pretrained(
@@ -30,31 +32,15 @@ pipeline = DiffusionPipeline.from_pretrained(
 )
 ```

-This will load the custom pipeline as defined in the [model repository](https://huggingface.co/hf-internal-testing/diffusers-dummy-pipeline/blob/main/pipeline.py).
+Loading an official community pipeline is similar, but you can mix loading weights from an official repository id and pass pipeline components directly. The example below loads the community [CLIP Guided Stable Diffusion](https://github.com/huggingface/diffusers/tree/main/examples/community#clip-guided-stable-diffusion) pipeline, and you can pass the CLIP model components directly to it:

-<Tip warning={true} >
-
-By loading a custom pipeline from the Hugging Face Hub, you are trusting that the code you are loading 
-is safe 🔒. Make sure to check out the code online before loading & running it automatically.
-
-</Tip>
-
-## Loading official community pipelines
-
-Community pipelines are summarized in the [community examples folder](https://github.com/huggingface/diffusers/tree/main/examples/community). 
-
-Similarly, you need to pass both the *repo id* from where you wish to load the weights as well as the `custom_pipeline` argument. Here the `custom_pipeline` argument should consist simply of the filename of the community pipeline excluding the `.py` suffix, *e.g.* `clip_guided_stable_diffusion`.
-
-Since community pipelines are often more complex, one can mix loading weights from an official *repo id*
-and passing pipeline modules directly.
-
-```python
+```py
 from diffusers import DiffusionPipeline
-from transformers import CLIPFeatureExtractor, CLIPModel
+from transformers import CLIPImageProcessor, CLIPModel

 clip_model_id = "laion/CLIP-ViT-B-32-laion2B-s34B-b79K"

-feature_extractor = CLIPFeatureExtractor.from_pretrained(clip_model_id)
+feature_extractor = CLIPImageProcessor.from_pretrained(clip_model_id)
 clip_model = CLIPModel.from_pretrained(clip_model_id)

 pipeline = DiffusionPipeline.from_pretrained(
@@ -65,57 +51,4 @@ pipeline = DiffusionPipeline.from_pretrained(
 )
 ```

-## Adding custom pipelines to the Hub
-
-To add a custom pipeline to the Hub, all you need to do is to define a pipeline class that inherits 
-from [`DiffusionPipeline`] in a `pipeline.py` file.
-Make sure that the whole pipeline is encapsulated within a single class and that the `pipeline.py` file
-has only one such class.
-
-Let's quickly define an example pipeline.
-
-
-```python
-import torch
-from diffusers import DiffusionPipeline
-
-
-class MyPipeline(DiffusionPipeline):
-    def __init__(self, unet, scheduler):
-        super().__init__()
-
-    self.register_modules(unet=unet, scheduler=scheduler)
-
-    @torch.no_grad()
-    def __call__(self, batch_size: int = 1, num_inference_steps: int = 50):
-        # Sample gaussian noise to begin loop
-        image = torch.randn((batch_size, self.unet.in_channels, self.unet.sample_size, self.unet.sample_size))
-
-        image = image.to(self.device)
-
-        # set step values
-        self.scheduler.set_timesteps(num_inference_steps)
-
-        for t in self.progress_bar(self.scheduler.timesteps):
-            # 1. predict noise model_output
-            model_output = self.unet(image, t).sample
-
-            # 2. predict previous mean of image x_t-1 and add variance depending on eta
-            # eta corresponds to η in paper and should be between [0, 1]
-            # do x_t -> x_t-1
-            image = self.scheduler.step(model_output, t, image, eta).prev_sample
-
-        image = (image / 2 + 0.5).clamp(0, 1)
-        image = image.cpu().permute(0, 2, 3, 1).numpy()
-
-        return image
-```
-
-Now you can upload this short file under the name `pipeline.py` in your preferred [model repository](https://huggingface.co/docs/hub/models-uploading). For Stable Diffusion pipelines, you may also [join the community organisation for shared pipelines](https://huggingface.co/organizations/sd-diffusers-pipelines-library/share/BUPyDUuHcciGTOKaExlqtfFcyCZsVFdrjr) to upload yours.
-Finally, we can load the custom pipeline by passing the model repository name, *e.g.* `sd-diffusers-pipelines-library/my_custom_pipeline` alongside the model repository from where we want to load the `unet` and `scheduler` components.
-
-```python
-my_pipeline = DiffusionPipeline.from_pretrained(
-    "google/ddpm-cifar10-32", custom_pipeline="patrickvonplaten/my_custom_pipeline"
-)
-```
+For more information about community pipelines, take a look at the [Community pipelines](custom_pipeline_examples) guide for how to use them and if you're interested in adding a community pipeline check out the [How to contribute a community pipeline](contribute_pipeline) guide!
@@ -10,9 +10,13 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Text-Guided Image-to-Image Generation
+# Text-guided depth-to-image generation

-The [`StableDiffusionDepth2ImgPipeline`] lets you pass a text prompt and an initial image to condition the generation of new images as well as a `depth_map` to preserve the images' structure. If no `depth_map` is provided, the pipeline will automatically predict the depth via an integrated depth-estimation model.
+[[open-in-colab]]
+
+The [`StableDiffusionDepth2ImgPipeline`] lets you pass a text prompt and an initial image to condition the generation of new images. In addition, you can also pass a `depth_map` to preserve the image structure. If no `depth_map` is provided, the pipeline automatically predicts the depth via an integrated [depth-estimation model](https://github.com/isl-org/MiDaS).
+
+Start by creating an instance of the [`StableDiffusionDepth2ImgPipeline`]:

 ```python
 import torch
@@ -25,11 +29,28 @@ pipe = StableDiffusionDepth2ImgPipeline.from_pretrained(
    "stabilityai/stable-diffusion-2-depth",
    torch_dtype=torch.float16,
 ).to("cuda")
+```

+Now pass your prompt to the pipeline. You can also pass a `negative_prompt` to prevent certain words from guiding how an image is generated:

+```python
 url = "http://images.cocodataset.org/val2017/000000039769.jpg"
 init_image = Image.open(requests.get(url, stream=True).raw)
 prompt = "two tigers"
 n_prompt = "bad, deformed, ugly, bad anatomy"
 image = pipe(prompt=prompt, image=init_image, negative_prompt=n_prompt, strength=0.7).images[0]
+image
 ```
+
+| Input                                                                           | Output                                                                                                                                |
+|---------------------------------------------------------------------------------|---------------------------------------------------------------------------------------------------------------------------------------|
+| <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/coco-cats.png" width="500"/> | <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/depth2img-tigers.png" width="500"/> |
+
+Play around with the Spaces below and see if you notice a difference between generated images with and without a depth map!
+
+<iframe
+	src="https://radames-stable-diffusion-depth2img.hf.space"
+	frameborder="0"
+	width="850"
+	height="500"
+></iframe>
@@ -10,11 +10,11 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Text-Guided Image-to-Image Generation
+# Text-guided image-to-image generation

 [[open-in-colab]]

-The [`StableDiffusionImg2ImgPipeline`] lets you pass a text prompt and an initial image to condition the generation of new images. This tutorial shows how to use it for text-guided image-to-image generation with Stable Diffusion model.
+The [`StableDiffusionImg2ImgPipeline`] lets you pass a text prompt and an initial image to condition the generation of new images.

 Before you begin, make sure you have all the necessary libraries installed:

@@ -22,27 +22,22 @@ Before you begin, make sure you have all the necessary libraries installed:
 !pip install diffusers transformers ftfy accelerate
 ```

-Get started by creating a [`StableDiffusionImg2ImgPipeline`] with a pretrained Stable Diffusion model.
+Get started by creating a [`StableDiffusionImg2ImgPipeline`] with a pretrained Stable Diffusion model like [`nitrosocke/Ghibli-Diffusion`](https://huggingface.co/nitrosocke/Ghibli-Diffusion).

 ```python
 import torch
 import requests
 from PIL import Image
 from io import BytesIO
-
 from diffusers import StableDiffusionImg2ImgPipeline
-```

-Load the pipeline:
-
-```python
 device = "cuda"
-pipe = StableDiffusionImg2ImgPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16).to(
+pipe = StableDiffusionImg2ImgPipeline.from_pretrained("nitrosocke/Ghibli-Diffusion", torch_dtype=torch.float16).to(
    device
 )
 ```

-Download an initial image and preprocess it so we can pass it to the pipeline:
+Download and preprocess an initial image so you can pass it to the pipeline:

 ```python
 url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg"
@@ -53,61 +48,52 @@ init_image.thumbnail((768, 768))
 init_image
 ```

-![img](https://huggingface.co/datasets/YiYiXu/test-doc-assets/resolve/main/image_2_image_using_diffusers_cell_8_output_0.jpeg)
-
-Define the prompt and run the pipeline:
-
-```python
-prompt = "A fantasy landscape, trending on artstation"
-```
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/YiYiXu/test-doc-assets/resolve/main/image_2_image_using_diffusers_cell_8_output_0.jpeg"/>
+</div>

 <Tip>

-`strength` is a value between 0.0 and 1.0, that controls the amount of noise that is added to the input image. Values that approach 1.0 allow for lots of variations but will also produce images that are not semantically consistent with the input.
+💡 `strength` is a value between 0.0 and 1.0 that controls the amount of noise added to the input image. Values that approach 1.0 allow for lots of variations but will also produce images that are not semantically consistent with the input.

 </Tip>

-Let's generate two images with same pipeline and seed, but with different values for `strength`:
+Define the prompt (for this checkpoint finetuned on Ghibli-style art, you need to prefix the prompt with the `ghibli style` tokens) and run the pipeline:

 ```python
+prompt = "ghibli style, a fantasy landscape with castles"
 generator = torch.Generator(device=device).manual_seed(1024)
 image = pipe(prompt=prompt, image=init_image, strength=0.75, guidance_scale=7.5, generator=generator).images[0]
-```
-
-```python
 image
 ```

-![img](https://huggingface.co/datasets/YiYiXu/test-doc-assets/resolve/main/image_2_image_using_diffusers_cell_13_output_0.jpeg)
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ghibli-castles.png"/>
+</div>

-
-```python
-image = pipe(prompt=prompt, image=init_image, strength=0.5, guidance_scale=7.5, generator=generator).images[0]
-image
-```
-
-![img](https://huggingface.co/datasets/YiYiXu/test-doc-assets/resolve/main/image_2_image_using_diffusers_cell_14_output_1.jpeg)
-
-
-As you can see, when using a lower value for `strength`, the generated image is more closer to the original `image`.
-
-Now let's use a different scheduler - [LMSDiscreteScheduler](https://huggingface.co/docs/diffusers/api/schedulers#diffusers.LMSDiscreteScheduler):
+You can also try experimenting with a different scheduler to see how that affects the output:

 ```python
 from diffusers import LMSDiscreteScheduler

 lms = LMSDiscreteScheduler.from_config(pipe.scheduler.config)
 pipe.scheduler = lms
-```
-
-```python
 generator = torch.Generator(device=device).manual_seed(1024)
 image = pipe(prompt=prompt, image=init_image, strength=0.75, guidance_scale=7.5, generator=generator).images[0]
-```
-
-```python
 image
 ```

-![img](https://huggingface.co/datasets/YiYiXu/test-doc-assets/resolve/main/image_2_image_using_diffusers_cell_19_output_0.jpeg)
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lms-ghibli.png"/>
+</div>

+Check out the Spaces below, and try generating images with different values for `strength`. You'll notice that using lower values for `strength` produces images that are more similar to the original image.
+
+Feel free to also switch the scheduler to the [`LMSDiscreteScheduler`] and see how that affects the output.
+
+<iframe
+	src="https://stevhliu-ghibli-img2img.hf.space"
+	frameborder="0"
+	width="850"
+	height="500"
+></iframe>
@@ -10,9 +10,13 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Text-Guided Image-Inpainting
+# Text-guided image-inpainting

-The [`StableDiffusionInpaintPipeline`] lets you edit specific parts of an image by providing a mask and a text prompt. It uses a version of Stable Diffusion specifically trained for in-painting tasks.
+[[open-in-colab]]
+
+The [`StableDiffusionInpaintPipeline`] allows you to edit specific parts of an image by providing a mask and a text prompt. It uses a version of Stable Diffusion, like [`runwayml/stable-diffusion-inpainting`](https://huggingface.co/runwayml/stable-diffusion-inpainting) specifically trained for inpainting tasks.
+
+Get started by loading an instance of the [`StableDiffusionInpaintPipeline`]:

 ```python
 import PIL
@@ -22,7 +26,16 @@ from io import BytesIO

 from diffusers import StableDiffusionInpaintPipeline

+pipeline = StableDiffusionInpaintPipeline.from_pretrained(
+    "runwayml/stable-diffusion-inpainting",
+    torch_dtype=torch.float16,
+)
+pipeline = pipeline.to("cuda")
+```

+Download an image and a mask of a dog which you'll eventually replace:
+
+```python
 def download_image(url):
    response = requests.get(url)
    return PIL.Image.open(BytesIO(response.content)).convert("RGB")
@@ -33,24 +46,31 @@ mask_url = "https://raw.githubusercontent.com/CompVis/latent-diffusion/main/data

 init_image = download_image(img_url).resize((512, 512))
 mask_image = download_image(mask_url).resize((512, 512))
+```

-pipe = StableDiffusionInpaintPipeline.from_pretrained(
-    "runwayml/stable-diffusion-inpainting",
-    torch_dtype=torch.float16,
-)
-pipe = pipe.to("cuda")
+Now you can create a prompt to replace the mask with something else:

+```python
 prompt = "Face of a yellow cat, high resolution, sitting on a park bench"
 image = pipe(prompt=prompt, image=init_image, mask_image=mask_image).images[0]
 ```

-`image`          | `mask_image` | `prompt` | **Output** |
+`image`          | `mask_image` | `prompt` | output |
 :-------------------------:|:-------------------------:|:-------------------------:|-------------------------:|
-<img src="https://raw.githubusercontent.com/CompVis/latent-diffusion/main/data/inpainting_examples/overture-creations-5sI6fQgYIuo.png" alt="drawing" width="250"/> | <img src="https://raw.githubusercontent.com/CompVis/latent-diffusion/main/data/inpainting_examples/overture-creations-5sI6fQgYIuo_mask.png" alt="drawing" width="250"/> | ***Face of a yellow cat, high resolution, sitting on a park bench*** | <img src="https://huggingface.co/datasets/patrickvonplaten/images/resolve/main/test.png" alt="drawing" width="250"/> |
+<img src="https://raw.githubusercontent.com/CompVis/latent-diffusion/main/data/inpainting_examples/overture-creations-5sI6fQgYIuo.png" alt="drawing" width="250"/> | <img src="https://raw.githubusercontent.com/CompVis/latent-diffusion/main/data/inpainting_examples/overture-creations-5sI6fQgYIuo_mask.png" alt="drawing" width="250"/> | ***Face of a yellow cat, high resolution, sitting on a park bench*** | <img src="https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/in_paint/yellow_cat_sitting_on_a_park_bench.png" alt="drawing" width="250"/> |


-You can also run this example on colab [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/in_painting_with_stable_diffusion_using_diffusers.ipynb)
-
 <Tip warning={true}>
-A previous experimental implementation of in-painting used a different, lower-quality process. To ensure backwards compatibility, loading a pretrained pipeline that doesn't contain the new model will still apply the old in-painting method.
+
+A previous experimental implementation of inpainting used a different, lower-quality process. To ensure backwards compatibility, loading a pretrained pipeline that doesn't contain the new model will still apply the old inpainting method.
+
 </Tip>
+
+Check out the Spaces below to try out image inpainting yourself!
+
+<iframe
+	src="https://runwayml-stable-diffusion-inpainting.hf.space"
+	frameborder="0"
+	width="850"
+	height="500"
+></iframe>
@@ -10,20 +10,28 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Loading
+# Load pipelines, models, and schedulers

-A core premise of the diffusers library is to make diffusion models **as accessible as possible**.
-Accessibility is therefore achieved by providing an API to load complete diffusion pipelines as well as individual components with a single line of code. 
+Having an easy way to use a diffusion system for inference is essential to 🧨 Diffusers. Diffusion systems often consist of multiple components like parameterized models, tokenizers, and schedulers that interact in complex ways. That is why we designed the [`DiffusionPipeline`] to wrap the complexity of the entire diffusion system into an easy-to-use API, while remaining flexible enough to be adapted for other use cases, such as loading each component individually as building blocks to assemble your own diffusion system.

-In the following we explain in-detail how to easily load:
+Everything you need for inference or training is accessible with the `from_pretrained()` method.

- *Complete Diffusion Pipelines* via the [`DiffusionPipeline.from_pretrained`]
- *Diffusion Models* via [`ModelMixin.from_pretrained`]
- *Schedulers* via [`SchedulerMixin.from_pretrained`]
+This guide will show you how to load:

-## Loading pipelines
+- pipelines from the Hub and locally
+- different components into a pipeline
+- checkpoint variants such as different floating point types or non-exponential mean averaged (EMA) weights
+- models and schedulers

-The [`DiffusionPipeline`] class is the easiest way to access any diffusion model that is [available on the Hub](https://huggingface.co/models?library=diffusers). Let's look at an example on how to download [Runway's Stable Diffusion model](https://huggingface.co/runwayml/stable-diffusion-v1-5).
+## Diffusion Pipeline
+
+<Tip>
+
+💡 Skip to the [DiffusionPipeline explained](#diffusionpipeline-explained) section if you interested in learning in more detail about how the [`DiffusionPipeline`] class works.
+
+</Tip>
+
+The [`DiffusionPipeline`] class is the simplest and most generic way to load any diffusion model from the [Hub](https://huggingface.co/models?library=diffusers). The [`DiffusionPipeline.from_pretrained`] method automatically detects the correct pipeline class from the checkpoint, downloads and caches all the required configuration and weight files, and returns a pipeline instance ready for inference.

 ```python
 from diffusers import DiffusionPipeline
@@ -32,10 +40,7 @@ repo_id = "runwayml/stable-diffusion-v1-5"
 pipe = DiffusionPipeline.from_pretrained(repo_id)
 ```

-Here [`DiffusionPipeline`] automatically detects the correct pipeline (*i.e.* [`StableDiffusionPipeline`]), downloads and caches all required configuration and weight files (if not already done so), and finally returns a pipeline instance, called `pipe`.
-The pipeline instance can then be called using [`StableDiffusionPipeline.__call__`] (i.e., `pipe("image of a astronaut riding a horse")`) for text-to-image generation.
-
-Instead of using the generic [`DiffusionPipeline`] class for loading, you can also load the appropriate pipeline class directly. The code snippet above yields the same instance as when doing:
+You can also load a checkpoint with it's specific pipeline class. The example above loaded a Stable Diffusion model; to get the same result, use the [`StableDiffusionPipeline`] class:

 ```python
 from diffusers import StableDiffusionPipeline
@@ -44,10 +49,7 @@ repo_id = "runwayml/stable-diffusion-v1-5"
 pipe = StableDiffusionPipeline.from_pretrained(repo_id)
 ```

-<Tip>
-
-Many checkpoints, such as [CompVis/stable-diffusion-v1-4](https://huggingface.co/CompVis/stable-diffusion-v1-4) and [runwayml/stable-diffusion-v1-5](https://huggingface.co/runwayml/stable-diffusion-v1-5) can be used for multiple tasks, *e.g.* *text-to-image* or *image-to-image*.
-If you want to use those checkpoints for a task that is different from the default one, you have to load it directly from the corresponding task-specific pipeline class:
+A checkpoint (such as [`CompVis/stable-diffusion-v1-4`](https://huggingface.co/CompVis/stable-diffusion-v1-4) or [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5)) may also be used for more than one task, like text-to-image or image-to-image. To differentiate what task you want to use the checkpoint for, you have to load it directly with it's corresponding task-specific pipeline class:

 ```python
 from diffusers import StableDiffusionImg2ImgPipeline
@@ -56,82 +58,16 @@ repo_id = "runwayml/stable-diffusion-v1-5"
 pipe = StableDiffusionImg2ImgPipeline.from_pretrained(repo_id)
 ```

-</Tip>
+### Local pipeline

+To load a diffusion pipeline locally, use [`git-lfs`](https://git-lfs.github.com/) to manually download the checkpoint (in this case, [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5)) to your local disk. This creates a local folder, `./stable-diffusion-v1-5`, on your disk:

-Diffusion pipelines like `StableDiffusionPipeline` or `StableDiffusionImg2ImgPipeline` consist of multiple components. These components can be both parameterized models, such as `"unet"`, `"vae"` and `"text_encoder"`, tokenizers or schedulers. 
-These components often interact in complex ways with each other when using the pipeline in inference, *e.g.* for [`StableDiffusionPipeline`] the inference call is explained [here](https://huggingface.co/blog/stable_diffusion#how-does-stable-diffusion-work).
-The purpose of the [pipeline classes](./api/overview#diffusers-summary) is to wrap the complexity of these diffusion systems and give the user an easy-to-use API while staying flexible for customization, as will be shown later.
-
-<!---
-THE FOLLOWING CAN BE UNCOMMENTED ONCE WE HAVE NEW MODELS WITH ACCESS REQUIREMENT
-
-# Loading pipelines that require access request
-
-Due to the capabilities of diffusion models to generate extremely realistic images, there is a certain danger that such models might be misused for unwanted applications, *e.g.* generating pornography or violent images.
-In order to minimize the possibility of such unsolicited use cases, some of the most powerful diffusion models require users to acknowledge a license before being able to use the model. If the user does not agree to the license, the pipeline cannot be downloaded.
-If you try to load [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5) the same way as done previously:
-
-```python
-from diffusers import DiffusionPipeline
-
-repo_id = "runwayml/stable-diffusion-v1-5"
-stable_diffusion = DiffusionPipeline.from_pretrained(repo_id)
-```
-
-it will only work if you have both *click-accepted* the license on [the model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) and are logged into the Hugging Face Hub. Otherwise you will get an error message
-such as the following:
-
-```
-OSError: runwayml/stable-diffusion-v1-5 is not a local folder and is not a valid model identifier listed on 'https://huggingface.co/models'
-If this is a private repository, make sure to pass a token having permission to this repo with `use_auth_token` or log in with `huggingface-cli login`
-```
-
-Therefore, we need to make sure to *click-accept* the license. You can do this by simply visiting 
-the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) and clicking on "Agree and access repository":
-
-<p align="center">
-    <br>
-    <img src="https://raw.githubusercontent.com/huggingface/diffusers/main/docs/source/imgs/access_request.png" width="400"/>
-    <br>
-</p>
-
-Second, you need to login with your access token:
-
-```
-huggingface-cli login
-```
-
-before trying to load the model. Or alternatively, you can pass [your access token](https://huggingface.co/docs/hub/security-tokens#user-access-tokens) directly via the flag `use_auth_token`. In this case you do **not** need 
-to run `huggingface-cli login` before:
-
-```python
-from diffusers import DiffusionPipeline
-
-repo_id = "runwayml/stable-diffusion-v1-5"
-stable_diffusion = DiffusionPipeline.from_pretrained(repo_id, use_auth_token="<your-access-token>")
-```
-
-The final option to use pipelines that require access without having to rely on the Hugging Face Hub is to load the pipeline locally as explained in the next section.
-->
-
-### Loading pipelines locally
-
-If you prefer to have complete control over the pipeline and its corresponding files or, as said before, if you want to use pipelines that require an access request without having to be connected to the Hugging Face Hub, 
-we recommend loading pipelines locally.
-
-To load a diffusion pipeline locally, you first need to manually download the whole folder structure on your local disk and then pass a local path to the [`DiffusionPipeline.from_pretrained`]. Let's again look at an example for 
-[Runway's Stable Diffusion Diffusion model](https://huggingface.co/runwayml/stable-diffusion-v1-5).
-
-First, you should make use of [`git-lfs`](https://git-lfs.github.com/) to download the whole folder structure that has been uploaded to the [model repository](https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main):
-
-```
+```bash
 git lfs install
 git clone https://huggingface.co/runwayml/stable-diffusion-v1-5
 ```

-The command above will create a local folder called `./stable-diffusion-v1-5` on your disk.
-Now, all you have to do is to simply pass the local folder path to `from_pretrained`:
+Then pass the local path to [`~DiffusionPipeline.from_pretrained`]:

 ```python
 from diffusers import DiffusionPipeline
@@ -140,17 +76,29 @@ repo_id = "./stable-diffusion-v1-5"
 stable_diffusion = DiffusionPipeline.from_pretrained(repo_id)
 ```

-If `repo_id` is a local path, as it is the case here, [`DiffusionPipeline.from_pretrained`] will automatically detect it and therefore not try to download any files from the Hub.
-While we usually recommend to load weights directly from the Hub to be certain to stay up to date with the newest changes, loading pipelines locally should be preferred if one 
-wants to stay anonymous, self-contained applications, etc...
+The [`~DiffusionPipeline.from_pretrained`] method won't download any files from the Hub when it detects a local path, but this also means it won't download and cache the latest changes to a checkpoint.

-### Loading customized pipelines
+### Swap components in a pipeline

-Advanced users that want to load customized versions of diffusion pipelines can do so by swapping any of the default components, *e.g.* the scheduler, with other scheduler classes.
-A classical use case of this functionality is to swap the scheduler. [Stable Diffusion v1-5](https://huggingface.co/runwayml/stable-diffusion-v1-5) uses the [`PNDMScheduler`] by default which is generally not the most performant scheduler. Since the release
-of stable diffusion, multiple improved schedulers have been published. To use those, the user has to manually load their preferred scheduler and pass it into [`DiffusionPipeline.from_pretrained`].
+You can customize the default components of any pipeline with another compatible component. Customization is important because:

-*E.g.* to use [`EulerDiscreteScheduler`] or [`DPMSolverMultistepScheduler`] to have a better quality vs. generation speed trade-off for inference, one could load them as follows:
+- Changing the scheduler is important for exploring the trade-off between generation speed and quality.
+- Different components of a model are typically trained independently and you can swap out a component with a better-performing one.
+- During finetuning, usually only some components - like the UNet or text encoder - are trained.
+
+To find out which schedulers are compatible for customization, you can use the `compatibles` method:
+
+```py
+from diffusers import DiffusionPipeline
+
+repo_id = "runwayml/stable-diffusion-v1-5"
+stable_diffusion = DiffusionPipeline.from_pretrained(repo_id)
+stable_diffusion.scheduler.compatibles
+```
+
+Let's use the [`SchedulerMixin.from_pretrained`] method to replace the default [`PNDMScheduler`] with a more performant scheduler, [`EulerDiscreteScheduler`]. The `subfolder="scheduler"` argument is required to load the scheduler configuration from the correct [subfolder](https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main/scheduler) of the pipeline repository.
+
+Then you can pass the new [`EulerDiscreteScheduler`] instance to the `scheduler` argument in [`DiffusionPipeline`]:

 ```python
 from diffusers import DiffusionPipeline, EulerDiscreteScheduler, DPMSolverMultistepScheduler
@@ -158,31 +106,24 @@ from diffusers import DiffusionPipeline, EulerDiscreteScheduler, DPMSolverMultis
 repo_id = "runwayml/stable-diffusion-v1-5"

 scheduler = EulerDiscreteScheduler.from_pretrained(repo_id, subfolder="scheduler")
-# or
-# scheduler = DPMSolverMultistepScheduler.from_pretrained(repo_id, subfolder="scheduler")

 stable_diffusion = DiffusionPipeline.from_pretrained(repo_id, scheduler=scheduler)
 ```

-Three things are worth paying attention to here.
- First, the scheduler is loaded with [`SchedulerMixin.from_pretrained`]
- Second, the scheduler is loaded with a function argument, called `subfolder="scheduler"` as the configuration of stable diffusion's scheduling is defined in a [subfolder of the official pipeline repository](https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main/scheduler)
- Third, the scheduler instance can simply be passed with the `scheduler` keyword argument to [`DiffusionPipeline.from_pretrained`]. This works because the [`StableDiffusionPipeline`] defines its scheduler with the `scheduler` attribute. It's not possible to use a different name, such as `sampler=scheduler` since `sampler` is not a defined keyword for [`StableDiffusionPipeline.__init__`]
+### Safety checker

-Not only the scheduler components can be customized for diffusion pipelines; in theory, all components of a pipeline can be customized. In practice, however, it often only makes sense to switch out a component that has **compatible** alternatives to what the pipeline expects.
-Many scheduler classes are compatible with each other as can be seen [here](https://github.com/huggingface/diffusers/blob/0dd8c6b4dbab4069de9ed1cafb53cbd495873879/src/diffusers/schedulers/scheduling_ddim.py#L112). This is not always the case for other components, such as the `"unet"`.
-
-One special case that can also be customized is the `"safety_checker"` of stable diffusion. If you believe the safety checker doesn't serve you any good, you can simply disable it by passing `None`:
+Diffusion models like Stable Diffusion can generate harmful content, which is why 🧨 Diffusers has a [safety checker](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/safety_checker.py) to check generated outputs against known hardcoded NSFW content. If you'd like to disable the safety checker for whatever reason, pass `None` to the `safety_checker` argument:

 ```python
-from diffusers import DiffusionPipeline, EulerDiscreteScheduler, DPMSolverMultistepScheduler
+from diffusers import DiffusionPipeline

+repo_id = "runwayml/stable-diffusion-v1-5"
 stable_diffusion = DiffusionPipeline.from_pretrained(repo_id, safety_checker=None)
 ```

-Another common use case is to reuse the same components in multiple pipelines, *e.g.* the weights and configurations of [`"runwayml/stable-diffusion-v1-5"`](https://huggingface.co/runwayml/stable-diffusion-v1-5) can be used for both [`StableDiffusionPipeline`] and [`StableDiffusionImg2ImgPipeline`] and we might not want to 
-use the exact same weights into RAM twice. In this case, customizing all the input instances would help us 
-to only load the weights into RAM once:
+### Reuse components across pipelines
+
+You can also reuse the same components in multiple pipelines to avoid loading the weights into RAM twice. Use the [`~DiffusionPipeline.components`] method to save the components:

 ```python
 from diffusers import StableDiffusionPipeline, StableDiffusionImg2ImgPipeline
@@ -191,349 +132,92 @@ model_id = "runwayml/stable-diffusion-v1-5"
 stable_diffusion_txt2img = StableDiffusionPipeline.from_pretrained(model_id)

 components = stable_diffusion_txt2img.components
+```

-# weights are not reloaded into RAM
+Then you can pass the `components` to another pipeline without reloading the weights into RAM:
+
+```py
 stable_diffusion_img2img = StableDiffusionImg2ImgPipeline(**components)
 ```

-Note how the above code snippet makes use of [`DiffusionPipeline.components`].
-
-### Loading variants
-
-Diffusion Pipeline checkpoints can offer variants of the "main" diffusion pipeline checkpoint.
-Such checkpoint variants are usually variations of the checkpoint that have advantages for specific use-cases and that are so similar to the "main" checkpoint that they **should not** be put in a new checkpoint.
-A variation of a checkpoint has to have **exactly** the same serialization format and **exactly** the same model structure, including all weights having the same tensor shapes.
-
-Examples of variations are different floating point types and non-ema weights. I.e. "fp16", "bf16", and "no_ema" are common variations.
-
-#### Let's first talk about whats **not** checkpoint variant,
-
-Checkpoint variants do **not** include different serialization formats (such as [safetensors](https://huggingface.co/docs/diffusers/main/en/using-diffusers/using_safetensors)) as weights in different serialization formats are 
-identical to the weights of the "main" checkpoint, just loaded in a different framework.
-
-Also variants do not correspond to different model structures, *e.g.* [stable-diffusion-v1-5](https://huggingface.co/runwayml/stable-diffusion-v1-5) is not a variant of [stable-diffusion-2-0](https://huggingface.co/stabilityai/stable-diffusion-2) since the model structure is different (Stable Diffusion 1-5 uses a different `CLIPTextModel` compared to Stable Diffusion 2.0).
-
-Pipeline checkpoints that are identical in model structure, but have been trained on different datasets, trained with vastly different training setups and thus correspond to different official releases (such as [Stable Diffusion v1-4](https://huggingface.co/CompVis/stable-diffusion-v1-4) and [Stable Diffusion v1-5](https://huggingface.co/runwayml/stable-diffusion-v1-5)) should probably be stored in individual repositories instead of as variations of each other.
-
-#### So what are checkpoint variants then? 
-
-Checkpoint variants usually consist of the checkpoint stored in "*low-precision, low-storage*" dtype so that less bandwith is required to download them, or of *non-exponential-averaged* weights that shall be used when continuing fine-tuning from the checkpoint.
-Both use cases have clear advantages when their weights are considered variants: they share the same serialization format as the reference weights, and they correspond to a specialization of the "main" checkpoint which does not warrant a new model repository. 
-A checkpoint stored in [torch's half-precision / float16 format](https://pytorch.org/blog/accelerating-training-on-nvidia-gpus-with-pytorch-automatic-mixed-precision/) requires only half the bandwith and storage when downloading the checkpoint,
-**but** cannot be used when continuing training or when running the checkpoint on CPU.
-Similarly the *non-exponential-averaged* (or non-EMA) version of the checkpoint should be used when continuing fine-tuning of the model checkpoint, **but** should not be used when using the checkpoint for inference. 
-
-#### How to save and load variants
-
-Saving a diffusion pipeline as a variant can be done by providing [`DiffusionPipeline.save_pretrained`] with the `variant` argument. 
-The `variant` extends the weight name by the provided variation, by changing the default weight name from `diffusion_pytorch_model.bin` to `diffusion_pytorch_model.{variant}.bin` or from `diffusion_pytorch_model.safetensors` to `diffusion_pytorch_model.{variant}.safetensors`. By doing so, one creates a variant of the pipeline checkpoint that can be loaded **instead** of the "main" pipeline checkpoint.
-
-Let's have a look at how we could create a float16 variant of a pipeline. First, we load 
-the "main" variant of a checkpoint (stored in `float32` precision) into mixed precision format, using `torch_dtype=torch.float16`.
+You can also pass the components individually to the pipeline if you want more flexibility over which components to reuse or disable. For example, to reuse the same components in the text-to-image pipeline, except for the safety checker and feature extractor, in the image-to-image pipeline:

 ```py
-from diffusers import DiffusionPipeline
-import torch
+from diffusers import StableDiffusionPipeline, StableDiffusionImg2ImgPipeline

-pipe = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16)
+model_id = "runwayml/stable-diffusion-v1-5"
+stable_diffusion_txt2img = StableDiffusionPipeline.from_pretrained(model_id)
+stable_diffusion_img2img = StableDiffusionImg2ImgPipeline(
+    vae=stable_diffusion_txt2img.vae,
+    text_encoder=stable_diffusion_txt2img.text_encoder,
+    tokenizer=stable_diffusion_txt2img.tokenizer,
+    unet=stable_diffusion_txt2img.unet,
+    scheduler=stable_diffusion_txt2img.scheduler,
+    safety_checker=None,
+    feature_extractor=None,
+    requires_safety_checker=False,
+)
 ```

-Now all model components of the pipeline are stored in half-precision dtype. We can now save the 
-pipeline under a `"fp16"` variant as follows:
+## Checkpoint variants

-```py
-pipe.save_pretrained("./stable-diffusion-v1-5", variant="fp16")
-```
+A checkpoint variant is usually a checkpoint where it's weights are:

-If we don't save into an existing `stable-diffusion-v1-5` folder the new folder would look as follows:
-
-```
-stable-diffusion-v1-5
-├── feature_extractor
-│   └── preprocessor_config.json
-├── model_index.json
-├── safety_checker
-│   ├── config.json
-│   └── pytorch_model.fp16.bin
-├── scheduler
-│   └── scheduler_config.json
-├── text_encoder
-│   ├── config.json
-│   └── pytorch_model.fp16.bin
-├── tokenizer
-│   ├── merges.txt
-│   ├── special_tokens_map.json
-│   ├── tokenizer_config.json
-│   └── vocab.json
-├── unet
-│   ├── config.json
-│   └── diffusion_pytorch_model.fp16.bin
-└── vae
-    ├── config.json
-    └── diffusion_pytorch_model.fp16.bin
-```
-
-As one can see, all model files now have a `.fp16.bin` extension instead of just `.bin`.
-The variant now has to be loaded by also passing a `variant="fp16"` to [`DiffusionPipeline.from_pretrained`], e.g.:
-
-
-```py
-DiffusionPipeline.from_pretrained("./stable-diffusion-v1-5", variant="fp16", torch_dtype=torch.float16)
-```
-
-works just fine, while:
-
-```py
-DiffusionPipeline.from_pretrained("./stable-diffusion-v1-5", torch_dtype=torch.float16)
-```
-
-throws an Exception:
-```
-OSError: Error no file named diffusion_pytorch_model.bin found in directory ./stable-diffusion-v1-45/vae since we **only** stored the model 
-```
-
-This is expected as we don't have any "non-variant" checkpoint files saved locally.
-However, the whole idea of pipeline variants is that they can co-exist with the "main" variant, 
-so one would typically also save the "main" variant in the same folder. Let's do this:
-
-```py
-pipe = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5")
-pipe.save_pretrained("./stable-diffusion-v1-5")
-```
-
-and upload the pipeline to the Hub under [diffusers/stable-diffusion-variants](https://huggingface.co/diffusers/stable-diffusion-variants).
-The file structure [on the Hub](https://huggingface.co/diffusers/stable-diffusion-variants/tree/main) now looks as follows:
-
-```
-├── feature_extractor
-│   └── preprocessor_config.json
-├── model_index.json
-├── safety_checker
-│   ├── config.json
-│   ├── pytorch_model.bin
-│   └── pytorch_model.fp16.bin
-├── scheduler
-│   └── scheduler_config.json
-├── text_encoder
-│   ├── config.json
-│   ├── pytorch_model.bin
-│   └── pytorch_model.fp16.bin
-├── tokenizer
-│   ├── merges.txt
-│   ├── special_tokens_map.json
-│   ├── tokenizer_config.json
-│   └── vocab.json
-├── unet
-│   ├── config.json
-│   ├── diffusion_pytorch_model.bin
-│   ├── diffusion_pytorch_model.fp16.bin
-└── vae
-    ├── config.json
-    ├── diffusion_pytorch_model.bin
-    └── diffusion_pytorch_model.fp16.bin
-```
-
-We can now both download the "main" and the "fp16" variant from the Hub. Both:
-
-```py
-pipe = DiffusionPipeline.from_pretrained("diffusers/stable-diffusion-variants")
-```
-
-and 
-
-```py
-pipe = DiffusionPipeline.from_pretrained("diffusers/stable-diffusion-variants", variant="fp16")
-```
-
-work.
+- Stored in a different floating point type for lower precision and lower storage, such as [`torch.float16`](https://pytorch.org/docs/stable/tensors.html#data-types), because it only requires half the bandwidth and storage to download. You can't use this variant if you're continuing training or using a CPU.
+- Non-exponential mean averaged (EMA) weights which shouldn't be used for inference. You should use these to continue finetuning a model.

 <Tip>

-Note that Diffusers never downloads more checkpoints than needed. E.g. when downloading 
-the "main" variant, none of the "fp16.bin" files are downloaded and cached.
-Only when the user specifies `variant="fp16"` are those files downloaded and cached.
+💡 When the checkpoints have identical model structures, but they were trained on different datasets and with a different training setup, they should be stored in separate repositories instead of variations (for example, [`stable-diffusion-v1-4`] and [`stable-diffusion-v1-5`]).

 </Tip>

-Finally, there are cases where only some of the checkpoint files of the pipeline are of a certain 
-variation. E.g. it's usually only the UNet checkpoint that has both a *exponential-mean-averaged* (EMA) and a *non-exponential-mean-averaged* (non-EMA) version. All other model components, e.g. the text encoder, safety checker or variational auto-encoder usually don't have such a variation.
-In such a case, one would upload just the UNet's checkpoint file with a `non_ema` version format (as done [here](https://huggingface.co/diffusers/stable-diffusion-variants/blob/main/unet/diffusion_pytorch_model.non_ema.bin)) and upon calling:
+Otherwise, a variant is **identical** to the original checkpoint. They have exactly the same serialization format (like [Safetensors](./using-diffusers/using_safetensors)), model structure, and weights have identical tensor shapes.

-```python
-pipe = DiffusionPipeline.from_pretrained("diffusers/stable-diffusion-variants", variant="non_ema")
-```
+| **checkpoint type** | **weight name**                     | **argument for loading weights** |
+|---------------------|-------------------------------------|----------------------------------|
+| original            | diffusion_pytorch_model.bin         |                                  |
+| floating point      | diffusion_pytorch_model.fp16.bin    | `variant`, `torch_dtype`         |
+| non-EMA             | diffusion_pytorch_model.non_ema.bin | `variant`                        |

-the model will use only the "non_ema" checkpoint variant if it is available - otherwise it'll load the 
-"main" variation. In the above example, `variant="non_ema"` would therefore download the following file structure:
+There are two important arguments to know for loading variants:

-```
-├── feature_extractor
-│   └── preprocessor_config.json
-├── model_index.json
-├── safety_checker
-│   ├── config.json
-│   ├── pytorch_model.bin
-├── scheduler
-│   └── scheduler_config.json
-├── text_encoder
-│   ├── config.json
-│   ├── pytorch_model.bin
-├── tokenizer
-│   ├── merges.txt
-│   ├── special_tokens_map.json
-│   ├── tokenizer_config.json
-│   └── vocab.json
-├── unet
-│   ├── config.json
-│   └── diffusion_pytorch_model.non_ema.bin
-└── vae
-    ├── config.json
-    ├── diffusion_pytorch_model.bin
-```
+- `torch_dtype` defines the floating point precision of the loaded checkpoints. For example, if you want to save bandwidth by loading a `fp16` variant, you should specify `torch_dtype=torch.float16` to *convert the weights* to `fp16`. Otherwise, the `fp16` weights are converted to the default `fp32` precision. You can also load the original checkpoint without defining the `variant` argument, and convert it to `fp16` with `torch_dtype=torch.float16`. In this case, the default `fp32` weights are downloaded first, and then they're converted to `fp16` after loading.

-In a nutshell, using `variant="{variant}"` will download all files that match the `{variant}` and if for a model component such a file variant is not present it will download the "main" variant. If neither a "main" or `{variant}` variant is available, an error will the thrown.
-
-### How does loading work?
-
-As a class method, [`DiffusionPipeline.from_pretrained`] is responsible for two things:
- Download the latest version of the folder structure required to run the `repo_id` with `diffusers` and cache them. If the latest folder structure is available in the local cache, [`DiffusionPipeline.from_pretrained`] will simply reuse the cache and **not** re-download the files.
- Load the cached weights into the _correct_ pipeline class – one of the [officially supported pipeline classes](./api/overview#diffusers-summary) - and return an instance of the class. The _correct_ pipeline class is thereby retrieved from the `model_index.json` file.
-
-The underlying folder structure of diffusion pipelines corresponds 1-to-1 to their corresponding class instances, *e.g.* [`StableDiffusionPipeline`] for [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5).
-This can be better understood by looking at an example. Let's load a pipeline class instance `pipe` and print it:
+- `variant` defines which files should be loaded from the repository. For example, if you want to load a `non_ema` variant from the [`diffusers/stable-diffusion-variants`](https://huggingface.co/diffusers/stable-diffusion-variants/tree/main/unet) repository, you should specify `variant="non_ema"` to download the `non_ema` files.

 ```python
 from diffusers import DiffusionPipeline

-repo_id = "runwayml/stable-diffusion-v1-5"
-pipe = DiffusionPipeline.from_pretrained(repo_id)
-print(pipe)
+# load fp16 variant
+stable_diffusion = DiffusionPipeline.from_pretrained(
+    "runwayml/stable-diffusion-v1-5", variant="fp16", torch_dtype=torch.float16
+)
+# load non_ema variant
+stable_diffusion = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", variant="non_ema")
 ```

-*Output*:
-```
-StableDiffusionPipeline {
-  "feature_extractor": [
-    "transformers",
-    "CLIPFeatureExtractor"
-  ],
-  "safety_checker": [
-    "stable_diffusion",
-    "StableDiffusionSafetyChecker"
-  ],
-  "scheduler": [
-    "diffusers",
-    "PNDMScheduler"
-  ],
-  "text_encoder": [
-    "transformers",
-    "CLIPTextModel"
-  ],
-  "tokenizer": [
-    "transformers",
-    "CLIPTokenizer"
-  ],
-  "unet": [
-    "diffusers",
-    "UNet2DConditionModel"
-  ],
-  "vae": [
-    "diffusers",
-    "AutoencoderKL"
-  ]
-}
+To save a checkpoint stored in a different floating point type or as a non-EMA variant, use the [`DiffusionPipeline.save_pretrained`] method and specify the `variant` argument. You should try and save a variant to the same folder as the original checkpoint, so you can load both from the same folder:
+
+```python
+from diffusers import DiffusionPipeline
+
+# save as fp16 variant
+stable_diffusion.save_pretrained("runwayml/stable-diffusion-v1-5", variant="fp16")
+# save as non-ema variant
+stable_diffusion.save_pretrained("runwayml/stable-diffusion-v1-5", variant="non_ema")
 ```

-First, we see that the official pipeline is the [`StableDiffusionPipeline`], and second we see that the `StableDiffusionPipeline` consists of 7 components:
- `"feature_extractor"` of class `CLIPFeatureExtractor` as defined [in `transformers`](https://huggingface.co/docs/transformers/main/en/model_doc/clip#transformers.CLIPFeatureExtractor).
- `"safety_checker"` as defined [here](https://github.com/huggingface/diffusers/blob/e55687e1e15407f60f32242027b7bb8170e58266/src/diffusers/pipelines/stable_diffusion/safety_checker.py#L32).
- `"scheduler"` of class [`PNDMScheduler`].
- `"text_encoder"` of class `CLIPTextModel` as defined [in `transformers`](https://huggingface.co/docs/transformers/main/en/model_doc/clip#transformers.CLIPTextModel).
- `"tokenizer"` of class `CLIPTokenizer` as defined [in `transformers`](https://huggingface.co/docs/transformers/main/en/model_doc/clip#transformers.CLIPTokenizer).
- `"unet"` of class [`UNet2DConditionModel`].
- `"vae"` of class [`AutoencoderKL`].
-
-Let's now compare the pipeline instance to the folder structure of the model repository `runwayml/stable-diffusion-v1-5`. Looking at the folder structure of [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main) on the Hub and  excluding model and saving format variants, we can see it matches 1-to-1 the printed out instance of `StableDiffusionPipeline` above:
+If you don't save the variant to an existing folder, you must specify the `variant` argument otherwise it'll throw an `Exception` because it can't find the original checkpoint:

+```python
+# 👎 this won't work
+stable_diffusion = DiffusionPipeline.from_pretrained("./stable-diffusion-v1-5", torch_dtype=torch.float16)
+# 👍 this works
+stable_diffusion = DiffusionPipeline.from_pretrained(
+    "./stable-diffusion-v1-5", variant="fp16", torch_dtype=torch.float16
+)
 ```
-.
-├── feature_extractor
-│   └── preprocessor_config.json
-├── model_index.json
-├── safety_checker
-│   ├── config.json
-│   └── pytorch_model.bin
-├── scheduler
-│   └── scheduler_config.json
-├── text_encoder
-│   ├── config.json
-│   └── pytorch_model.bin
-├── tokenizer
-│   ├── merges.txt
-│   ├── special_tokens_map.json
-│   ├── tokenizer_config.json
-│   └── vocab.json
-├── unet
-│   ├── config.json
-│   ├── diffusion_pytorch_model.bin
-└── vae
-    ├── config.json
-    ├── diffusion_pytorch_model.bin
-```
-
-Each attribute of the instance of `StableDiffusionPipeline` has its configuration and possibly weights defined in a subfolder that is called **exactly** like the class attribute (`"feature_extractor"`, `"safety_checker"`, `"scheduler"`, `"text_encoder"`, `"tokenizer"`, `"unet"`, `"vae"`). Importantly, every pipeline expects a `model_index.json` file that tells the `DiffusionPipeline` both:
- which pipeline class should be loaded, and
- what sub-classes from which library are stored in which subfolders
-
-In the case of `runwayml/stable-diffusion-v1-5` the `model_index.json` is therefore defined as follows:
-
-```
-{
-  "_class_name": "StableDiffusionPipeline",
-  "_diffusers_version": "0.6.0",
-  "feature_extractor": [
-    "transformers",
-    "CLIPFeatureExtractor"
-  ],
-  "safety_checker": [
-    "stable_diffusion",
-    "StableDiffusionSafetyChecker"
-  ],
-  "scheduler": [
-    "diffusers",
-    "PNDMScheduler"
-  ],
-  "text_encoder": [
-    "transformers",
-    "CLIPTextModel"
-  ],
-  "tokenizer": [
-    "transformers",
-    "CLIPTokenizer"
-  ],
-  "unet": [
-    "diffusers",
-    "UNet2DConditionModel"
-  ],
-  "vae": [
-    "diffusers",
-    "AutoencoderKL"
-  ]
-}
-```
-
- `_class_name` tells `DiffusionPipeline` which pipeline class should be loaded. 
- `_diffusers_version` can be useful to know under which `diffusers` version this model was created.
- Every component of the pipeline is then defined under the form:
-```
-"name" : [
-  "library",
-  "class"
-]
-```
-	- The `"name"` field corresponds both to the name of the subfolder in which the configuration and weights are stored as well as the attribute name of the pipeline class (as can be seen [here](https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main/bert) and [here](https://github.com/huggingface/diffusers/blob/cd502b25cf0debac6f98d27a6638ef95208d1ea2/src/diffusers/pipelines/latent_diffusion/pipeline_latent_diffusion.py#L42))
-	- The `"library"` field corresponds to the name of the library, *e.g.* `diffusers` or `transformers` from which the `"class"` should be loaded
-	- The `"class"` field corresponds to the name of the class, *e.g.* [`CLIPTokenizer`](https://huggingface.co/docs/transformers/main/en/model_doc/clip#transformers.CLIPTokenizer) or [`UNet2DConditionModel`]

 <!--
 TODO(Patrick) - Make sure to uncomment this part as soon as things are deprecated.
@@ -562,15 +246,11 @@ instead.
 </Tip>
 -->

-## Loading models
+## Models

-Models as defined under [src/diffusers/models](https://github.com/huggingface/diffusers/tree/main/src/diffusers/models) can be loaded via the [`ModelMixin.from_pretrained`] function. The API is very similar the [`DiffusionPipeline.from_pretrained`] and works in the same way:
- Download the latest version of the model weights and configuration with `diffusers` and cache them. If the latest files are available in the local cache, [`ModelMixin.from_pretrained`] will simply reuse the cache and **not** re-download the files.
- Load the cached weights into the _defined_ model class - one of [the existing model classes](./api/models) - and return an instance of the class.
+Models are loaded from the [`ModelMixin.from_pretrained`] method, which downloads and caches the latest version of the model weights and configurations. If the latest files are available in the local cache, [`~ModelMixin.from_pretrained`] reuses files in the cache instead of redownloading them.

-In constrast to [`DiffusionPipeline.from_pretrained`], models rely on fewer files that usually don't require a folder structure, but just a `diffusion_pytorch_model.bin` and `config.json` file.
-
-Let's look at an example:
+Models can be loaded from a subfolder with the `subfolder` argument. For example, the model weights for `runwayml/stable-diffusion-v1-5` are stored in the [`unet`](https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main/unet) subfolder:

 ```python
 from diffusers import UNet2DConditionModel
@@ -579,19 +259,7 @@ repo_id = "runwayml/stable-diffusion-v1-5"
 model = UNet2DConditionModel.from_pretrained(repo_id, subfolder="unet")
 ```

-Note how we have to define the `subfolder="unet"` argument to tell [`ModelMixin.from_pretrained`] that the model weights are located in a [subfolder of the repository](https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main/unet).
-
-As explained in [Loading customized pipelines]("./using-diffusers/loading#loading-customized-pipelines"), one can pass a loaded model to a diffusion pipeline, via [`DiffusionPipeline.from_pretrained`]:
-
-```python
-from diffusers import DiffusionPipeline
-
-repo_id = "runwayml/stable-diffusion-v1-5"
-pipe = DiffusionPipeline.from_pretrained(repo_id, unet=model)
-```
-
-If the model files can be found directly at the root level, which is usually only the case for some very simple diffusion models, such as [`google/ddpm-cifar10-32`](https://huggingface.co/google/ddpm-cifar10-32), we don't 
-need to pass a `subfolder` argument:
+Or directly from a repository's [directory](https://huggingface.co/google/ddpm-cifar10-32/tree/main):

 ```python
 from diffusers import UNet2DModel
@@ -600,35 +268,21 @@ repo_id = "google/ddpm-cifar10-32"
 model = UNet2DModel.from_pretrained(repo_id)
 ```

-As motivated in [How to save and load variants?](#how-to-save-and-load-variants), models can load and 
-save variants. To load a model variant, one should pass the `variant` function argument to [`ModelMixin.from_pretrained`]. Analogous, to save a model variant, one should pass the `variant` function argument to [`ModelMixin.save_pretrained`]:
+You can also load and save model variants by specifying the `variant` argument in [`ModelMixin.from_pretrained`] and [`ModelMixin.save_pretrained`]:

 ```python
 from diffusers import UNet2DConditionModel

-model = UNet2DConditionModel.from_pretrained(
-    "diffusers/stable-diffusion-variants", subfolder="unet", variant="non_ema"
-)
-model.save_pretrained("./local-unet", variant="non_ema")
+model = UNet2DConditionModel.from_pretrained("runwayml/stable-diffusion-v1-5", subfolder="unet", variant="non-ema")
+model.save_pretrained("./local-unet", variant="non-ema")
 ```

-## Loading schedulers
+## Schedulers

-Schedulers rely on [`SchedulerMixin.from_pretrained`]. Schedulers are **not parameterized** or **trained**, but instead purely defined by a configuration file.
-For consistency, we use the same method name as we do for models or pipelines, but no weights are loaded in this case.
+Schedulers are loaded from the [`SchedulerMixin.from_pretrained`] method, and unlike models, schedulers are **not parameterized** or **trained**; they are defined by a configuration file.

-In constrast to pipelines or models, loading schedulers does not consume any significant amount of memory and the same configuration file can often be used for a variety of different schedulers.
-For example, all of:
-
- [`DDPMScheduler`]
- [`DDIMScheduler`]
- [`PNDMScheduler`]
- [`LMSDiscreteScheduler`]
- [`EulerDiscreteScheduler`]
- [`EulerAncestralDiscreteScheduler`]
- [`DPMSolverMultistepScheduler`]
-
-are compatible with [`StableDiffusionPipeline`] and therefore the same scheduler configuration file can be loaded in any of those classes:
+Loading schedulers does not consume any significant amount of memory and the same configuration file can be used for a variety of different schedulers.
+For example, the following schedulers are compatible with [`StableDiffusionPipeline`] which means you can load the same scheduler configuration file in any of these classes:

 ```python
 from diffusers import StableDiffusionPipeline
@@ -655,3 +309,152 @@ dpm = DPMSolverMultistepScheduler.from_pretrained(repo_id, subfolder="scheduler"
 # replace `dpm` with any of `ddpm`, `ddim`, `pndm`, `lms`, `euler_anc`, `euler`
 pipeline = StableDiffusionPipeline.from_pretrained(repo_id, scheduler=dpm)
 ```
+
+## DiffusionPipeline explained
+
+As a class method, [`DiffusionPipeline.from_pretrained`] is responsible for two things:
+
+- Download the latest version of the folder structure required for inference and cache it. If the latest folder structure is available in the local cache, [`DiffusionPipeline.from_pretrained`] reuses the cache and won't redownload the files.
+- Load the cached weights into the correct pipeline [class](./api/pipelines/overview#diffusers-summary) - retrieved from the `model_index.json` file - and return an instance of it.
+
+The pipelines underlying folder structure corresponds directly with their class instances. For example, the [`StableDiffusionPipeline`] corresponds to the folder structure in [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5).
+
+```python
+from diffusers import DiffusionPipeline
+
+repo_id = "runwayml/stable-diffusion-v1-5"
+pipeline = DiffusionPipeline.from_pretrained(repo_id)
+print(pipeline)
+```
+
+You'll see pipeline is an instance of [`StableDiffusionPipeline`], which consists of seven components:
+
+- `"feature_extractor"`: a [`~transformers.CLIPFeatureExtractor`] from 🤗 Transformers.
+- `"safety_checker"`: a [component](https://github.com/huggingface/diffusers/blob/e55687e1e15407f60f32242027b7bb8170e58266/src/diffusers/pipelines/stable_diffusion/safety_checker.py#L32) for screening against harmful content.
+- `"scheduler"`: an instance of [`PNDMScheduler`].
+- `"text_encoder"`: a [`~transformers.CLIPTextModel`] from 🤗 Transformers.
+- `"tokenizer"`: a [`~transformers.CLIPTokenizer`] from 🤗 Transformers.
+- `"unet"`: an instance of [`UNet2DConditionModel`].
+- `"vae"` an instance of [`AutoencoderKL`].
+
+```json
+StableDiffusionPipeline {
+  "feature_extractor": [
+    "transformers",
+    "CLIPImageProcessor"
+  ],
+  "safety_checker": [
+    "stable_diffusion",
+    "StableDiffusionSafetyChecker"
+  ],
+  "scheduler": [
+    "diffusers",
+    "PNDMScheduler"
+  ],
+  "text_encoder": [
+    "transformers",
+    "CLIPTextModel"
+  ],
+  "tokenizer": [
+    "transformers",
+    "CLIPTokenizer"
+  ],
+  "unet": [
+    "diffusers",
+    "UNet2DConditionModel"
+  ],
+  "vae": [
+    "diffusers",
+    "AutoencoderKL"
+  ]
+}
+```
+
+Compare the components of the pipeline instance to the [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5) folder structure, and you'll see there is a separate folder for each of the components in the repository:
+
+```
+.
+├── feature_extractor
+│   └── preprocessor_config.json
+├── model_index.json
+├── safety_checker
+│   ├── config.json
+│   └── pytorch_model.bin
+├── scheduler
+│   └── scheduler_config.json
+├── text_encoder
+│   ├── config.json
+│   └── pytorch_model.bin
+├── tokenizer
+│   ├── merges.txt
+│   ├── special_tokens_map.json
+│   ├── tokenizer_config.json
+│   └── vocab.json
+├── unet
+│   ├── config.json
+│   ├── diffusion_pytorch_model.bin
+└── vae
+    ├── config.json
+    ├── diffusion_pytorch_model.bin
+```
+
+You can access each of the components of the pipeline as an attribute to view its configuration:
+
+```py
+pipeline.tokenizer
+CLIPTokenizer(
+    name_or_path="/root/.cache/huggingface/hub/models--runwayml--stable-diffusion-v1-5/snapshots/39593d5650112b4cc580433f6b0435385882d819/tokenizer",
+    vocab_size=49408,
+    model_max_length=77,
+    is_fast=False,
+    padding_side="right",
+    truncation_side="right",
+    special_tokens={
+        "bos_token": AddedToken("<|startoftext|>", rstrip=False, lstrip=False, single_word=False, normalized=True),
+        "eos_token": AddedToken("<|endoftext|>", rstrip=False, lstrip=False, single_word=False, normalized=True),
+        "unk_token": AddedToken("<|endoftext|>", rstrip=False, lstrip=False, single_word=False, normalized=True),
+        "pad_token": "<|endoftext|>",
+    },
+)
+```
+
+Every pipeline expects a `model_index.json` file that tells the [`DiffusionPipeline`]:
+
+- which pipeline class to load from `_class_name`
+- which version of 🧨 Diffusers was used to create the model in `_diffusers_version`
+- what components from which library are stored in the subfolders (`name` corresponds to the component and subfolder name, `library` corresponds to the name of the library to load the class from, and `class` corresponds to the class name)
+
+```json
+{
+  "_class_name": "StableDiffusionPipeline",
+  "_diffusers_version": "0.6.0",
+  "feature_extractor": [
+    "transformers",
+    "CLIPImageProcessor"
+  ],
+  "safety_checker": [
+    "stable_diffusion",
+    "StableDiffusionSafetyChecker"
+  ],
+  "scheduler": [
+    "diffusers",
+    "PNDMScheduler"
+  ],
+  "text_encoder": [
+    "transformers",
+    "CLIPTextModel"
+  ],
+  "tokenizer": [
+    "transformers",
+    "CLIPTokenizer"
+  ],
+  "unet": [
+    "diffusers",
+    "UNet2DConditionModel"
+  ],
+  "vae": [
+    "diffusers",
+    "AutoencoderKL"
+  ]
+}
+```
@@ -10,26 +10,26 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Reproducibility
+# Create reproducible pipelines

-Before reading about reproducibility for Diffusers, it is strongly recommended to take a look at 
-[PyTorch's statement about reproducibility](https://pytorch.org/docs/stable/notes/randomness.html).
+Reproducibility is important for testing, replicating results, and can even be used to [improve image quality](reusing_seeds). However, the randomness in diffusion models is a desired property because it allows the pipeline to generate different images every time it is run. While you can't expect to get the exact same results across platforms, you can expect results to be reproducible across releases and platforms within a certain tolerance range. Even then, tolerance varies depending on the diffusion pipeline and checkpoint.

-PyTorch states that 
-> *completely reproducible results are not guaranteed across PyTorch releases, individual commits, or different platforms.*
-While one can never expect the same results across platforms, one can expect results to be reproducible 
-across releases, platforms, etc... within a certain tolerance. However, this tolerance strongly varies 
-depending on the diffusion pipeline and checkpoint.
+This is why it's important to understand how to control sources of randomness in diffusion models.

-In the following, we show how to best control sources of randomness for diffusion models.
+<Tip>
+
+💡 We strongly recommend reading PyTorch's [statement about reproducibility](https://pytorch.org/docs/stable/notes/randomness.html):
+
+> Completely reproducible results are not guaranteed across PyTorch releases, individual commits, or different platforms. Furthermore, results may not be reproducible between CPU and GPU executions, even when using identical seeds.
+
+</Tip>

 ## Inference

-During inference, diffusion pipelines heavily rely on random sampling operations, such as the creating the 
-gaussian noise tensors to be denoised and adding noise to the scheduling step.
+During inference, pipelines rely heavily on random sampling operations which include creating the 
+Gaussian noise tensors to denoise and adding noise to the scheduling step.

-Let's have a look at an example. We run the [DDIM pipeline](./api/pipelines/ddim.mdx) 
-for just two inference steps and return a numpy tensor to look into the numerical values of the output.
+Take a look at the tensor values in the [`DDIMPipeline`] after two inference steps:

 ```python
 from diffusers import DDIMPipeline
@@ -45,11 +45,15 @@ image = ddim(num_inference_steps=2, output_type="np").images
 print(np.abs(image).sum())
 ```

-Running the above prints a value of 1464.2076, but running it again prints a different 
-value of 1495.1768. What is going on here? Every time the pipeline is run, gaussian noise 
-is created and step-wise denoised. To create the gaussian noise with [`torch.randn`](https://pytorch.org/docs/stable/generated/torch.randn.html), a different random seed is taken every time, thus leading to a different result.
-This is a desired property of diffusion pipelines, as it means that the pipeline can create a different random image every time it is run. In many cases, one would like to generate the exact same image of a certain 
-run, for which case an instance of a [PyTorch generator](https://pytorch.org/docs/stable/generated/torch.randn.html) has to be passed:
+Running the code above prints one value, but if you run it again you get a different value. What is going on here? 
+
+Every time the pipeline is run, [`torch.randn`](https://pytorch.org/docs/stable/generated/torch.randn.html) uses a different random seed to create Gaussian noise which is denoised stepwise. This leads to a different result each time it is run, which is great for diffusion pipelines since it generates a different random image each time.
+
+But if you need to reliably generate the same image, that'll depend on whether you're running the pipeline on a CPU or GPU.
+
+### CPU
+
+To generate reproducible results on a CPU, you'll need to use a PyTorch [`Generator`](https://pytorch.org/docs/stable/generated/torch.randn.html) and set a seed:

 ```python
 import torch
@@ -69,28 +73,22 @@ image = ddim(num_inference_steps=2, output_type="np", generator=generator).image
 print(np.abs(image).sum())
 ```

-Running the above always prints a value of 1491.1711 - also upon running it again because we 
-define the generator object to be passed to all random functions of the pipeline.
+Now when you run the code above, it always prints a value of `1491.1711` no matter what because the `Generator` object with the seed is passed to all the random functions of the pipeline.

-If you run this code snippet on your specific hardware and version, you should get a similar, if not the same, result.
+If you run this code example on your specific hardware and PyTorch version, you should get a similar, if not the same, result.

 <Tip>

-It might be a bit unintuitive at first to pass `generator` objects to the pipelines instead of 
+💡 It might be a bit unintuitive at first to pass `Generator` objects to the pipeline instead of 
 just integer values representing the seed, but this is the recommended design when dealing with 
-probabilistic models in PyTorch as generators are *random states* that are advanced and can thus be 
+probabilistic models in PyTorch as `Generator`'s are *random states* that can be 
 passed to multiple pipelines in a sequence.

 </Tip>

-Great! Now, we know how to write reproducible pipelines, but it gets a bit trickier since the above example only runs on the CPU. How do we also achieve reproducibility on GPU? 
-In short, one should not expect full reproducibility across different hardware when running pipelines on GPU 
-as matrix multiplications are less deterministic on GPU than on CPU and diffusion pipelines tend to require
-a lot of matrix multiplications. Let's see what we can do to keep the randomness within limits across 
-different GPU hardware.
+### GPU

-To achieve maximum speed performance, it is recommended to create the generator directly on GPU when running 
-the pipeline on GPU:
+Writing a reproducible pipeline on a GPU is a bit trickier, and full reproducibility across different hardware is not guaranteed because matrix multiplication - which diffusion pipelines require a lot of - is less deterministic on a GPU than a CPU. For example, if you run the same code example above on a GPU: 

 ```python
 import torch
@@ -111,12 +109,11 @@ image = ddim(num_inference_steps=2, output_type="np", generator=generator).image
 print(np.abs(image).sum())
 ```

-Running the above now prints a value of 1389.8634 - even though we're using the exact same seed! 
-This is unfortunate as it means we cannot reproduce the results we achieved on GPU, also on CPU.
-Nevertheless, it should be expected since the GPU uses a different random number generator than the CPU.
+The result is not the same even though you're using an identical seed because the GPU uses a different random number generator than the CPU.

-To circumvent this problem, we created a [`randn_tensor`](#diffusers.utils.randn_tensor) function, which can create random noise 
-on the CPU and then move the tensor to GPU if necessary. The function is used everywhere inside the pipelines allowing the user to **always** pass a CPU generator even if the pipeline is run on GPU:
+To circumvent this problem, 🧨 Diffusers has a [`randn_tensor`](#diffusers.utils.randn_tensor) function for creating random noise on the CPU, and then moving the tensor to a GPU if necessary. The `randn_tensor` function is used everywhere inside the pipeline, allowing the user to **always** pass a CPU `Generator` even if the pipeline is run on a GPU. 
+
+You'll see the results are much closer now!

 ```python
 import torch
@@ -129,7 +126,7 @@ model_id = "google/ddpm-cifar10-32"
 ddim = DDIMPipeline.from_pretrained(model_id)
 ddim.to("cuda")

-# create a generator for reproducibility
+# create a generator for reproducibility; notice you don't place it on the GPU!
 generator = torch.manual_seed(0)

 # run pipeline for just two steps and return numpy tensor
@@ -137,23 +134,18 @@ image = ddim(num_inference_steps=2, output_type="np", generator=generator).image
 print(np.abs(image).sum())
 ```

-Running the above now prints a value of 1491.1713, much closer to the value of 1491.1711 when 
-the pipeline is fully run on the CPU.
-
 <Tip>

-As a consequence, we recommend always passing a CPU generator if Reproducibility is important.
-The loss of performance is often neglectable, but one can be sure to generate much more similar 
-values than if the pipeline would have been run on CPU.
+💡 If reproducibility is important, we recommend always passing a CPU generator.
+The performance loss is often neglectable, and you'll generate much more similar 
+values than if the pipeline had been run on a GPU.

 </Tip>

-Finally, we noticed that more complex pipelines, such as [`UnCLIPPipeline`] are often extremely 
-susceptible to precision error propagation and thus one cannot expect even similar results across 
-different GPU hardware or PyTorch versions. In such cases, one has to make sure to run 
-exactly the same hardware and PyTorch version for full Reproducibility.
+Finally, for more complex pipelines such as [`UnCLIPPipeline`], these are often extremely 
+susceptible to precision error propagation. Don't expect similar results across 
+different GPU hardware or PyTorch versions. In this case, you'll need to run 
+exactly the same hardware and PyTorch version for full reproducibility.

-## Randomness utilities
-
-### randn_tensor
+## randn_tensor
 [[autodoc]] diffusers.utils.randn_tensor
@@ -10,23 +10,17 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

-# Re-using seeds for fast prompt engineering
+# Improve image quality with deterministic generation

-A common use case when generating images is to generate a batch of images, select one image and improve it with a better, more detailed prompt in a second run.
-To do this, one needs to make each generated image of the batch deterministic.
-Images are generated by denoising gaussian random noise which can be instantiated by passing a [torch generator](https://pytorch.org/docs/stable/generated/torch.Generator.html#generator).
+A common way to improve the quality of generated images is with *deterministic batch generation*, generate a batch of images and select one image to improve with a more detailed prompt in a second round of inference. The key is to pass a list of [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html#generator)'s to the pipeline for batched image generation, and tie each `Generator` to a seed so you can reuse it for an image.

-Now, for batched generation, we need to make sure that every single generated image in the batch is tied exactly to one seed. In 🧨 Diffusers, this can be achieved by not passing one `generator`, but a list 
-of `generators` to the pipeline.
-
-Let's go through an example using [`runwayml/stable-diffusion-v1-5`](runwayml/stable-diffusion-v1-5).
-We want to generate several versions of the prompt:
+Let's use [`runwayml/stable-diffusion-v1-5`](runwayml/stable-diffusion-v1-5) for example, and generate several versions of the following prompt:

 ```py
 prompt = "Labrador in the style of Vermeer"
 ```

-Let's load the pipeline
+Instantiate a pipeline with [`DiffusionPipeline.from_pretrained`] and place it on a GPU (if available):

 ```python
 >>> from diffusers import DiffusionPipeline
@@ -35,7 +29,7 @@ Let's load the pipeline
 >>> pipe = pipe.to("cuda")
 ```

-Now, let's define 4 different generators, since we would like to reproduce a certain image. We'll use seeds `0` to `3` to create our generators.
+Now, define four different `Generator`'s and assign each `Generator` a seed (`0` to `3`) so you can reuse a `Generator` later for a specific image:

 ```python
 >>> import torch
@@ -43,7 +37,7 @@ Now, let's define 4 different generators, since we would like to reproduce a cer
 >>> generator = [torch.Generator(device="cuda").manual_seed(i) for i in range(4)]
 ```

-Let's generate 4 images:
+Generate the images and have a look:

 ```python
 >>> images = pipe(prompt, generator=generator, num_images_per_prompt=4).images
@@ -52,18 +46,14 @@ Let's generate 4 images:

 ![img](https://huggingface.co/datasets/diffusers/diffusers-images-docs/resolve/main/reusabe_seeds.jpg)

-Ok, the last images has some double eyes, but the first image looks good! 
-Let's try to make the prompt a bit better **while keeping the first seed** 
-so that the images are similar to the first image.
+In this example, you'll improve upon the first image - but in reality, you can use any image you want (even the image with double sets of eyes!). The first image used the `Generator` with seed `0`, so you'll reuse that `Generator` for the second round of inference. To improve the quality of the image, add some additional text to the prompt:

 ```python
 prompt = [prompt + t for t in [", highly realistic", ", artsy", ", trending", ", colorful"]]
 generator = [torch.Generator(device="cuda").manual_seed(0) for i in range(4)]
 ```

-We create 4 generators with seed `0`, which is the first seed we used before.
-
-Let's run the pipeline again.
+Create four generators with seed `0`, and generate another batch of images, all of which should look like the first image from the previous round!

 ```python
 >>> images = pipe(prompt, generator=generator).images
@@ -0,0 +1,250 @@
+# 🧨 Stable Diffusion in JAX / Flax !
+
+[[open-in-colab]]
+
+🤗 Hugging Face [Diffusers](https://github.com/huggingface/diffusers) supports Flax since version `0.5.1`! This allows for super fast inference on Google TPUs, such as those available in Colab, Kaggle or Google Cloud Platform.
+
+This notebook shows how to run inference using JAX / Flax. If you want more details about how Stable Diffusion works or want to run it in GPU, please refer to [this notebook](https://huggingface.co/docs/diffusers/stable_diffusion).
+
+First, make sure you are using a TPU backend. If you are running this notebook in Colab, select `Runtime` in the menu above, then select the option "Change runtime type" and then select `TPU` under the `Hardware accelerator` setting.
+
+Note that JAX is not exclusive to TPUs, but it shines on that hardware because each TPU server has 8 TPU accelerators working in parallel.
+
+## Setup
+
+First make sure diffusers is installed.
+
+```bash
+!pip install jax==0.3.25 jaxlib==0.3.25 flax transformers ftfy
+!pip install diffusers
+```
+
+```python
+import jax.tools.colab_tpu
+
+jax.tools.colab_tpu.setup_tpu()
+import jax
+```
+
+```python
+num_devices = jax.device_count()
+device_type = jax.devices()[0].device_kind
+
+print(f"Found {num_devices} JAX devices of type {device_type}.")
+assert (
+    "TPU" in device_type
+), "Available device is not a TPU, please select TPU from Edit > Notebook settings > Hardware accelerator"
+```
+
+```python out
+Found 8 JAX devices of type Cloud TPU.
+```
+
+Then we import all the dependencies.
+
+```python
+import numpy as np
+import jax
+import jax.numpy as jnp
+
+from pathlib import Path
+from jax import pmap
+from flax.jax_utils import replicate
+from flax.training.common_utils import shard
+from PIL import Image
+
+from huggingface_hub import notebook_login
+from diffusers import FlaxStableDiffusionPipeline
+```
+
+## Model Loading
+
+TPU devices support `bfloat16`, an efficient half-float type. We'll use it for our tests, but you can also use `float32` to use full precision instead.
+
+```python
+dtype = jnp.bfloat16
+```
+
+Flax is a functional framework, so models are stateless and parameters are stored outside them. Loading the pre-trained Flax pipeline will return both the pipeline itself and the model weights (or parameters). We are using a `bf16` version of the weights, which leads to type warnings that you can safely ignore.
+
+```python
+pipeline, params = FlaxStableDiffusionPipeline.from_pretrained(
+    "CompVis/stable-diffusion-v1-4",
+    revision="bf16",
+    dtype=dtype,
+)
+```
+
+## Inference
+
+Since TPUs usually have 8 devices working in parallel, we'll replicate our prompt as many times as devices we have. Then we'll perform inference on the 8 devices at once, each responsible for generating one image. Thus, we'll get 8 images in the same amount of time it takes for one chip to generate a single one.
+
+After replicating the prompt, we obtain the tokenized text ids by invoking the `prepare_inputs` function of the pipeline. The length of the tokenized text is set to 77 tokens, as required by the configuration of the underlying CLIP Text model.
+
+```python
+prompt = "A cinematic film still of Morgan Freeman starring as Jimi Hendrix, portrait, 40mm lens, shallow depth of field, close up, split lighting, cinematic"
+prompt = [prompt] * jax.device_count()
+prompt_ids = pipeline.prepare_inputs(prompt)
+prompt_ids.shape
+```
+
+```python out
+(8, 77)
+```
+
+### Replication and parallelization
+
+Model parameters and inputs have to be replicated across the 8 parallel devices we have. The parameters dictionary is replicated using `flax.jax_utils.replicate`, which traverses the dictionary and changes the shape of the weights so they are repeated 8 times. Arrays are replicated using `shard`.
+
+```python
+p_params = replicate(params)
+```
+
+```python
+prompt_ids = shard(prompt_ids)
+prompt_ids.shape
+```
+
+```python out
+(8, 1, 77)
+```
+
+That shape means that each one of the `8` devices will receive as an input a `jnp` array with shape `(1, 77)`. `1` is therefore the batch size per device. In TPUs with sufficient memory, it could be larger than `1` if we wanted to generate multiple images (per chip) at once.
+
+We are almost ready to generate images! We just need to create a random number generator to pass to the generation function. This is the standard procedure in Flax, which is very serious and opinionated about random numbers – all functions that deal with random numbers are expected to receive a generator. This ensures reproducibility, even when we are training across multiple distributed devices.
+
+The helper function below uses a seed to initialize a random number generator. As long as we use the same seed, we'll get the exact same results. Feel free to use different seeds when exploring results later in the notebook.
+
+```python
+def create_key(seed=0):
+    return jax.random.PRNGKey(seed)
+```
+
+We obtain a rng and then "split" it 8 times so each device receives a different generator. Therefore, each device will create a different image, and the full process is reproducible.
+
+```python
+rng = create_key(0)
+rng = jax.random.split(rng, jax.device_count())
+```
+
+JAX code can be compiled to an efficient representation that runs very fast. However, we need to ensure that all inputs have the same shape in subsequent calls; otherwise, JAX will have to recompile the code, and we wouldn't be able to take advantage of the optimized speed.
+
+The Flax pipeline can compile the code for us if we pass `jit = True` as an argument. It will also ensure that the model runs in parallel in the 8 available devices.
+
+The first time we run the following cell it will take a long time to compile, but subequent calls (even with different inputs) will be much faster. For example, it took more than a minute to compile in a TPU v2-8 when I tested, but then it takes about **`7s`** for future inference runs.
+
+```
+%%time
+images = pipeline(prompt_ids, p_params, rng, jit=True)[0]
+```
+
+```python out
+CPU times: user 56.2 s, sys: 42.5 s, total: 1min 38s
+Wall time: 1min 29s
+```
+
+The returned array has shape `(8, 1, 512, 512, 3)`. We reshape it to get rid of the second dimension and obtain 8 images of `512 × 512 × 3` and then convert them to PIL.
+
+```python
+images = images.reshape((images.shape[0] * images.shape[1],) + images.shape[-3:])
+images = pipeline.numpy_to_pil(images)
+```
+
+### Visualization
+
+Let's create a helper function to display images in a grid.
+
+```python
+def image_grid(imgs, rows, cols):
+    w, h = imgs[0].size
+    grid = Image.new("RGB", size=(cols * w, rows * h))
+    for i, img in enumerate(imgs):
+        grid.paste(img, box=(i % cols * w, i // cols * h))
+    return grid
+```
+
+```python
+image_grid(images, 2, 4)
+```
+
+![img](https://huggingface.co/datasets/YiYiXu/test-doc-assets/resolve/main/stable_diffusion_jax_how_to_cell_38_output_0.jpeg)
+
+
+## Using different prompts
+
+We don't have to replicate the _same_ prompt in all the devices. We can do whatever we want: generate 2 prompts 4 times each, or even generate 8 different prompts at once. Let's do that!
+
+First, we'll refactor the input preparation code into a handy function:
+
+```python
+prompts = [
+    "Labrador in the style of Hokusai",
+    "Painting of a squirrel skating in New York",
+    "HAL-9000 in the style of Van Gogh",
+    "Times Square under water, with fish and a dolphin swimming around",
+    "Ancient Roman fresco showing a man working on his laptop",
+    "Close-up photograph of young black woman against urban background, high quality, bokeh",
+    "Armchair in the shape of an avocado",
+    "Clown astronaut in space, with Earth in the background",
+]
+```
+
+```python
+prompt_ids = pipeline.prepare_inputs(prompts)
+prompt_ids = shard(prompt_ids)
+
+images = pipeline(prompt_ids, p_params, rng, jit=True).images
+images = images.reshape((images.shape[0] * images.shape[1],) + images.shape[-3:])
+images = pipeline.numpy_to_pil(images)
+
+image_grid(images, 2, 4)
+```
+
+![img](https://huggingface.co/datasets/YiYiXu/test-doc-assets/resolve/main/stable_diffusion_jax_how_to_cell_43_output_0.jpeg)
+
+
+## How does parallelization work?
+
+We said before that the `diffusers` Flax pipeline automatically compiles the model and runs it in parallel on all available devices. We'll now briefly look inside that process to show how it works.
+
+JAX parallelization can be done in multiple ways. The easiest one revolves around using the `jax.pmap` function to achieve single-program, multiple-data (SPMD) parallelization. It means we'll run several copies of the same code, each on different data inputs. More sophisticated approaches are possible, we invite you to go over the [JAX documentation](https://jax.readthedocs.io/en/latest/index.html) and the [`pjit` pages](https://jax.readthedocs.io/en/latest/jax-101/08-pjit.html?highlight=pjit) to explore this topic if you are interested!
+
+`jax.pmap` does two things for us:
+- Compiles (or `jit`s) the code, as if we had invoked `jax.jit()`. This does not happen when we call `pmap`, but the first time the pmapped function is invoked.
+- Ensures the compiled code runs in parallel in all the available devices.
+
+To show how it works we `pmap` the `_generate` method of the pipeline, which is the private method that runs generates images. Please, note that this method may be renamed or removed in future releases of `diffusers`.
+
+```python
+p_generate = pmap(pipeline._generate)
+```
+
+After we use `pmap`, the prepared function `p_generate` will conceptually do the following:
+* Invoke a copy of the underlying function `pipeline._generate` in each device.
+* Send each device a different portion of the input arguments. That's what sharding is used for. In our case, `prompt_ids` has shape `(8, 1, 77, 768)`. This array will be split in `8` and each copy of `_generate` will receive an input with shape `(1, 77, 768)`.
+
+We can code `_generate` completely ignoring the fact that it will be invoked in parallel. We just care about our batch size (`1` in this example) and the dimensions that make sense for our code, and don't have to change anything to make it work in parallel.
+
+The same way as when we used the pipeline call, the first time we run the following cell it will take a while, but then it will be much faster.
+
+```
+%%time
+images = p_generate(prompt_ids, p_params, rng)
+images = images.block_until_ready()
+images.shape
+```
+
+```python out
+CPU times: user 1min 15s, sys: 18.2 s, total: 1min 34s
+Wall time: 1min 15s
+```
+
+```python
+images.shape
+```
+
+```python out
+(8, 1, 512, 512, 3)
+```
+
+We use `block_until_ready()` to correctly measure inference time, because JAX uses asynchronous dispatch and returns control to the Python loop as soon as it can. You don't need to use that in your code; blocking will occur automatically when you want to use the result of a computation that has not yet been materialized.
@@ -10,43 +10,60 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->

+# Unconditional image generation

+[[open-in-colab]]

-# Unconditional Image Generation
+Unconditional image generation is a relatively straightforward task. The model only generates images - without any additional context like text or an image - resembling the training data it was trained on.

 The [`DiffusionPipeline`] is the easiest way to use a pre-trained diffusion system for inference.

 Start by creating an instance of [`DiffusionPipeline`] and specify which pipeline checkpoint you would like to download.
-You can use the [`DiffusionPipeline`] for any [Diffusers' checkpoint](https://huggingface.co/models?library=diffusers&sort=downloads).
-In this guide though, you'll use [`DiffusionPipeline`] for unconditional image generation with [DDPM](https://arxiv.org/abs/2006.11239):
+You can use any of the 🧨 Diffusers [checkpoints](https://huggingface.co/models?library=diffusers&sort=downloads) from the Hub (the checkpoint you'll use generates images of butterflies).
+
+<Tip>
+
+💡 Want to train your own unconditional image generation model? Take a look at the training [guide](training/unconditional_training) to learn how to generate your own images.
+
+</Tip>
+
+In this guide, you'll use [`DiffusionPipeline`] for unconditional image generation with [DDPM](https://arxiv.org/abs/2006.11239):

 ```python
 >>> from diffusers import DiffusionPipeline

->>> generator = DiffusionPipeline.from_pretrained("google/ddpm-celebahq-256")
+>>> generator = DiffusionPipeline.from_pretrained("anton-l/ddpm-butterflies-128")
 ```
+
 The [`DiffusionPipeline`] downloads and caches all modeling, tokenization, and scheduling components. 
-Because the model consists of roughly 1.4 billion parameters, we strongly recommend running it on GPU.
-You can move the generator object to GPU, just like you would in PyTorch.
+Because the model consists of roughly 1.4 billion parameters, we strongly recommend running it on a GPU.
+You can move the generator object to a GPU, just like you would in PyTorch:

 ```python
 >>> generator.to("cuda")
 ```

-Now you can use the `generator` on your text prompt:
+Now you can use the `generator` to generate an image:

 ```python
 >>> image = generator().images[0]
 ```

-The output is by default wrapped into a [PIL Image object](https://pillow.readthedocs.io/en/stable/reference/Image.html?highlight=image#the-image-class).
+The output is by default wrapped into a [`PIL.Image`](https://pillow.readthedocs.io/en/stable/reference/Image.html?highlight=image#the-image-class) object.

-You can save the image by simply calling:
+You can save the image by calling:

 ```python
 >>> image.save("generated_image.png")
 ```

-
+Try out the Spaces below, and feel free to play around with the inference steps parameter to see how it affects the image quality!
+
+<iframe
+	src="https://stevhliu-ddpm-butterflies-128.hf.space"
+	frameborder="0"
+	width="850"
+	height="500"
+></iframe>


@@ -75,9 +75,9 @@ And we're equipped with dealing with it.
 Then in order to use the model, even before the branch gets accepted by the original author you can do:

 ```python
-from diffusers import StableDiffusionPipeline
+from diffusers import DiffusionPipeline

-pipe = StableDiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1", revision="refs/pr/22")
+pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1", revision="refs/pr/22")
 ```

 or you can test it directly online with this [space](https://huggingface.co/spaces/diffusers/check_pr).
@@ -42,6 +42,8 @@ Training examples show how to pretrain or fine-tune diffusion models for a varie
 | [**Text-to-Image fine-tuning**](./text_to_image) | ✅ | ✅ | 
 | [**Textual Inversion**](./textual_inversion) | ✅ | - | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_textual_inversion_training.ipynb)
 | [**Dreambooth**](./dreambooth) | ✅ | - | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_dreambooth_training.ipynb)
+| [**ControlNet**](./controlnet) | ✅ | ✅ | -
+| [**InstructPix2Pix**](./instruct_pix2pix) | ✅ | ✅ | -
 | [**Reinforcement Learning for Control**](https://github.com/huggingface/diffusers/blob/main/examples/rl/run_diffusers_locomotion.py)                    | - | - | coming soon.

 ## Community
@@ -30,7 +30,8 @@ MagicMix | Diffusion Pipeline for semantic mixing of an image and a text prompt
 | UnCLIP Text Interpolation Pipeline | Diffusion Pipeline that allows passing two prompts and produces images while interpolating between the text-embeddings of the two prompts | [UnCLIP Text Interpolation Pipeline](#unclip-text-interpolation-pipeline)                   | -                                                                                                                                                                                                                  | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | 
 | UnCLIP Image Interpolation Pipeline | Diffusion Pipeline that allows passing two images/image_embeddings and produces images while interpolating between their image-embeddings | [UnCLIP Image Interpolation Pipeline](#unclip-image-interpolation-pipeline)                   | -                                                                                                                                                                                                                  | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | 
 | DDIM Noise Comparative Analysis Pipeline | Investigating how the diffusion models learn visual concepts from each noise level (which is a contribution of [P2 weighting (CVPR 2022)](https://arxiv.org/abs/2204.00227)) | [DDIM Noise Comparative Analysis Pipeline](#ddim-noise-comparative-analysis-pipeline) | - |[Aengus (Duc-Anh)](https://github.com/aengusng8) |
-
+| CLIP Guided Img2Img Stable Diffusion Pipeline | Doing CLIP guidance for image to image generation with Stable Diffusion | [CLIP Guided Img2Img Stable Diffusion](#clip-guided-img2img-stable-diffusion) | - | [Nipun Jindal](https://github.com/nipunjindal/) | 
+| TensorRT Stable Diffusion Pipeline | Accelerates the Stable Diffusion Text2Image Pipeline using TensorRT | [TensorRT Stable Diffusion Pipeline](#tensorrt-text2image-stable-diffusion-pipeline) | - |[Asfiya Baig](https://github.com/asfiyab-nvidia) |


 To load a custom pipeline you just need to pass the `custom_pipeline` argument to `DiffusionPipeline`, as one of the files in `diffusers/examples/community`. Feel free to send a PR with your own pipelines, we will merge them quickly.
@@ -49,11 +50,11 @@ The following code requires roughly 12GB of GPU RAM.

 ```python
 from diffusers import DiffusionPipeline
-from transformers import CLIPFeatureExtractor, CLIPModel
+from transformers import CLIPImageProcessor, CLIPModel
 import torch


-feature_extractor = CLIPFeatureExtractor.from_pretrained("laion/CLIP-ViT-B-32-laion2B-s34B-b79K")
+feature_extractor = CLIPImageProcessor.from_pretrained("laion/CLIP-ViT-B-32-laion2B-s34B-b79K")
 clip_model = CLIPModel.from_pretrained("laion/CLIP-ViT-B-32-laion2B-s34B-b79K", torch_dtype=torch.float16)


@@ -1074,3 +1075,89 @@ for strength in np.linspace(0.1, 1, 25):
 Here is the result of this pipeline (which is DDIM) on CelebA-HQ dataset.

 ![noise-comparative-analysis](https://user-images.githubusercontent.com/67547213/224677066-4474b2ed-56ab-4c27-87c6-de3c0255eb9c.jpeg)
+
+### CLIP Guided Img2Img Stable Diffusion
+
+CLIP guided Img2Img stable diffusion can help to generate more realistic images with an initial image 
+by guiding stable diffusion at every denoising step with an additional CLIP model.
+
+The following code requires roughly 12GB of GPU RAM.
+
+```python
+from io import BytesIO
+import requests
+import torch
+from diffusers import DiffusionPipeline
+from PIL import Image
+from transformers import CLIPFeatureExtractor, CLIPModel
+feature_extractor = CLIPFeatureExtractor.from_pretrained(
+    "laion/CLIP-ViT-B-32-laion2B-s34B-b79K"
+)
+clip_model = CLIPModel.from_pretrained(
+    "laion/CLIP-ViT-B-32-laion2B-s34B-b79K", torch_dtype=torch.float16
+)
+guided_pipeline = DiffusionPipeline.from_pretrained(
+    "CompVis/stable-diffusion-v1-4",
+    # custom_pipeline="clip_guided_stable_diffusion",
+    custom_pipeline="/home/njindal/diffusers/examples/community/clip_guided_stable_diffusion.py",
+    clip_model=clip_model,
+    feature_extractor=feature_extractor,
+    torch_dtype=torch.float16,
+)
+guided_pipeline.enable_attention_slicing()
+guided_pipeline = guided_pipeline.to("cuda")
+prompt = "fantasy book cover, full moon, fantasy forest landscape, golden vector elements, fantasy magic, dark light night, intricate, elegant, sharp focus, illustration, highly detailed, digital painting, concept art, matte, art by WLOP and Artgerm and Albert Bierstadt, masterpiece"
+url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg"
+response = requests.get(url)
+init_image = Image.open(BytesIO(response.content)).convert("RGB")
+image = guided_pipeline(
+    prompt=prompt,
+    num_inference_steps=30,
+    image=init_image,
+    strength=0.75,
+    guidance_scale=7.5,
+    clip_guidance_scale=100,
+    num_cutouts=4,
+    use_cutouts=False,
+).images[0]
+display(image)
+```
+
+Init Image
+
+![img2img_init_clip_guidance](https://huggingface.co/datasets/njindal/images/resolve/main/clip_guided_img2img_init.jpg)
+
+Output Image
+
+![img2img_clip_guidance](https://huggingface.co/datasets/njindal/images/resolve/main/clip_guided_img2img.jpg)
+
+### TensorRT Text2Image Stable Diffusion Pipeline
+
+The TensorRT Pipeline can be used to accelerate the Text2Image Stable Diffusion Inference run.
+
+NOTE: The ONNX conversions and TensorRT engine build may take up to 30 minutes.
+
+```python
+import torch
+from diffusers import DDIMScheduler
+from diffusers.pipelines.stable_diffusion import StableDiffusionPipeline
+
+# Use the DDIMScheduler scheduler here instead
+scheduler = DDIMScheduler.from_pretrained("stabilityai/stable-diffusion-2-1",
+                                            subfolder="scheduler")
+
+pipe = StableDiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1",
+                                                custom_pipeline="stable_diffusion_tensorrt_txt2img",
+                                                revision='fp16',
+                                                torch_dtype=torch.float16,
+                                                scheduler=scheduler,)
+
+# re-use cached folder to save ONNX models and TensorRT Engines
+pipe.set_cached_folder("stabilityai/stable-diffusion-2-1", revision='fp16',)
+
+pipe = pipe.to("cuda")
+
+prompt = "a beautiful photograph of Mt. Fuji during cherry blossom"
+image = pipe(prompt).images[0]
+image.save('tensorrt_mt_fuji.png')
+```
@@ -238,7 +238,7 @@ class BitDiffusion(DiffusionPipeline):
        **kwargs,
    ) -> Union[Tuple, ImagePipelineOutput]:
        latents = torch.randn(
-            (batch_size, self.unet.in_channels, height, width),
+            (batch_size, self.unet.config.in_channels, height, width),
            generator=generator,
        )
        latents = decimal_to_bits(latents) * self.bit_scale
@@ -199,24 +199,20 @@ class CheckpointMergerPipeline(DiffusionPipeline):
            if not attr.startswith("_"):
                checkpoint_path_1 = os.path.join(cached_folders[1], attr)
                if os.path.exists(checkpoint_path_1):
-                    files = list(
-                        (
-                            *glob.glob(os.path.join(checkpoint_path_1, "*.safetensors")),
-                            *glob.glob(os.path.join(checkpoint_path_1, "*.bin")),
-                        )
-                    )
+                    files = [
+                        *glob.glob(os.path.join(checkpoint_path_1, "*.safetensors")),
+                        *glob.glob(os.path.join(checkpoint_path_1, "*.bin")),
+                    ]
                    checkpoint_path_1 = files[0] if len(files) > 0 else None
                if len(cached_folders) < 3:
                    checkpoint_path_2 = None
                else:
                    checkpoint_path_2 = os.path.join(cached_folders[2], attr)
                    if os.path.exists(checkpoint_path_2):
-                        files = list(
-                            (
-                                *glob.glob(os.path.join(checkpoint_path_2, "*.safetensors")),
-                                *glob.glob(os.path.join(checkpoint_path_2, "*.bin")),
-                            )
-                        )
+                        files = [
+                            *glob.glob(os.path.join(checkpoint_path_2, "*.safetensors")),
+                            *glob.glob(os.path.join(checkpoint_path_2, "*.bin")),
+                        ]
                        checkpoint_path_2 = files[0] if len(files) > 0 else None
                # For an attr if both checkpoint_path_1 and 2 are None, ignore.
                # If atleast one is present, deal with it according to interp method, of course only if the state_dict keys match.
@@ -5,12 +5,13 @@ import torch
 from torch import nn
 from torch.nn import functional as F
 from torchvision import transforms
-from transformers import CLIPFeatureExtractor, CLIPModel, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPModel, CLIPTextModel, CLIPTokenizer

 from diffusers import (
    AutoencoderKL,
    DDIMScheduler,
    DiffusionPipeline,
+    DPMSolverMultistepScheduler,
    LMSDiscreteScheduler,
    PNDMScheduler,
    UNet2DConditionModel,
@@ -63,8 +64,8 @@ class CLIPGuidedStableDiffusion(DiffusionPipeline):
        clip_model: CLIPModel,
        tokenizer: CLIPTokenizer,
        unet: UNet2DConditionModel,
-        scheduler: Union[PNDMScheduler, LMSDiscreteScheduler, DDIMScheduler],
-        feature_extractor: CLIPFeatureExtractor,
+        scheduler: Union[PNDMScheduler, LMSDiscreteScheduler, DDIMScheduler, DPMSolverMultistepScheduler],
+        feature_extractor: CLIPImageProcessor,
    ):
        super().__init__()
        self.register_modules(
@@ -125,17 +126,12 @@ class CLIPGuidedStableDiffusion(DiffusionPipeline):
    ):
        latents = latents.detach().requires_grad_()

-        if isinstance(self.scheduler, LMSDiscreteScheduler):
-            sigma = self.scheduler.sigmas[index]
-            # the model input needs to be scaled to match the continuous ODE formulation in K-LMS
-            latent_model_input = latents / ((sigma**2 + 1) ** 0.5)
-        else:
-            latent_model_input = latents
+        latent_model_input = self.scheduler.scale_model_input(latents, timestep)

        # predict the noise residual
        noise_pred = self.unet(latent_model_input, timestep, encoder_hidden_states=text_embeddings).sample

-        if isinstance(self.scheduler, (PNDMScheduler, DDIMScheduler)):
+        if isinstance(self.scheduler, (PNDMScheduler, DDIMScheduler, DPMSolverMultistepScheduler)):
            alpha_prod_t = self.scheduler.alphas_cumprod[timestep]
            beta_prod_t = 1 - alpha_prod_t
            # compute predicted original sample from predicted noise also called
@@ -258,7 +254,7 @@ class CLIPGuidedStableDiffusion(DiffusionPipeline):
        # Unlike in other pipelines, latents need to be generated in the target device
        # for 1-to-1 results reproducibility with the CompVis implementation.
        # However this currently doesn't work in `mps`.
-        latents_shape = (batch_size * num_images_per_prompt, self.unet.in_channels, height // 8, width // 8)
+        latents_shape = (batch_size * num_images_per_prompt, self.unet.config.in_channels, height // 8, width // 8)
        latents_dtype = text_embeddings.dtype
        if latents is None:
            if self.device.type == "mps":
@@ -0,0 +1,496 @@
+import inspect
+from typing import List, Optional, Union
+
+import numpy as np
+import PIL
+import torch
+from torch import nn
+from torch.nn import functional as F
+from torchvision import transforms
+from transformers import CLIPFeatureExtractor, CLIPModel, CLIPTextModel, CLIPTokenizer
+
+from diffusers import (
+    AutoencoderKL,
+    DDIMScheduler,
+    DiffusionPipeline,
+    DPMSolverMultistepScheduler,
+    LMSDiscreteScheduler,
+    PNDMScheduler,
+    UNet2DConditionModel,
+)
+from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput
+from diffusers.utils import (
+    PIL_INTERPOLATION,
+    deprecate,
+    randn_tensor,
+)
+
+
+EXAMPLE_DOC_STRING = """
+    Examples:
+        ```
+        from io import BytesIO
+
+        import requests
+        import torch
+        from diffusers import DiffusionPipeline
+        from PIL import Image
+        from transformers import CLIPFeatureExtractor, CLIPModel
+
+        feature_extractor = CLIPFeatureExtractor.from_pretrained(
+            "laion/CLIP-ViT-B-32-laion2B-s34B-b79K"
+        )
+        clip_model = CLIPModel.from_pretrained(
+            "laion/CLIP-ViT-B-32-laion2B-s34B-b79K", torch_dtype=torch.float16
+        )
+
+
+        guided_pipeline = DiffusionPipeline.from_pretrained(
+            "CompVis/stable-diffusion-v1-4",
+            # custom_pipeline="clip_guided_stable_diffusion",
+            custom_pipeline="/home/njindal/diffusers/examples/community/clip_guided_stable_diffusion.py",
+            clip_model=clip_model,
+            feature_extractor=feature_extractor,
+            torch_dtype=torch.float16,
+        )
+        guided_pipeline.enable_attention_slicing()
+        guided_pipeline = guided_pipeline.to("cuda")
+
+        prompt = "fantasy book cover, full moon, fantasy forest landscape, golden vector elements, fantasy magic, dark light night, intricate, elegant, sharp focus, illustration, highly detailed, digital painting, concept art, matte, art by WLOP and Artgerm and Albert Bierstadt, masterpiece"
+
+        url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg"
+
+        response = requests.get(url)
+        init_image = Image.open(BytesIO(response.content)).convert("RGB")
+
+        image = guided_pipeline(
+            prompt=prompt,
+            num_inference_steps=30,
+            image=init_image,
+            strength=0.75,
+            guidance_scale=7.5,
+            clip_guidance_scale=100,
+            num_cutouts=4,
+            use_cutouts=False,
+        ).images[0]
+        display(image)
+        ```
+"""
+
+
+def preprocess(image, w, h):
+    if isinstance(image, torch.Tensor):
+        return image
+    elif isinstance(image, PIL.Image.Image):
+        image = [image]
+
+    if isinstance(image[0], PIL.Image.Image):
+        image = [np.array(i.resize((w, h), resample=PIL_INTERPOLATION["lanczos"]))[None, :] for i in image]
+        image = np.concatenate(image, axis=0)
+        image = np.array(image).astype(np.float32) / 255.0
+        image = image.transpose(0, 3, 1, 2)
+        image = 2.0 * image - 1.0
+        image = torch.from_numpy(image)
+    elif isinstance(image[0], torch.Tensor):
+        image = torch.cat(image, dim=0)
+    return image
+
+
+class MakeCutouts(nn.Module):
+    def __init__(self, cut_size, cut_power=1.0):
+        super().__init__()
+
+        self.cut_size = cut_size
+        self.cut_power = cut_power
+
+    def forward(self, pixel_values, num_cutouts):
+        sideY, sideX = pixel_values.shape[2:4]
+        max_size = min(sideX, sideY)
+        min_size = min(sideX, sideY, self.cut_size)
+        cutouts = []
+        for _ in range(num_cutouts):
+            size = int(torch.rand([]) ** self.cut_power * (max_size - min_size) + min_size)
+            offsetx = torch.randint(0, sideX - size + 1, ())
+            offsety = torch.randint(0, sideY - size + 1, ())
+            cutout = pixel_values[:, :, offsety : offsety + size, offsetx : offsetx + size]
+            cutouts.append(F.adaptive_avg_pool2d(cutout, self.cut_size))
+        return torch.cat(cutouts)
+
+
+def spherical_dist_loss(x, y):
+    x = F.normalize(x, dim=-1)
+    y = F.normalize(y, dim=-1)
+    return (x - y).norm(dim=-1).div(2).arcsin().pow(2).mul(2)
+
+
+def set_requires_grad(model, value):
+    for param in model.parameters():
+        param.requires_grad = value
+
+
+class CLIPGuidedStableDiffusion(DiffusionPipeline):
+    """CLIP guided stable diffusion based on the amazing repo by @crowsonkb and @Jack000
+    - https://github.com/Jack000/glid-3-xl
+    - https://github.dev/crowsonkb/k-diffusion
+    """
+
+    def __init__(
+        self,
+        vae: AutoencoderKL,
+        text_encoder: CLIPTextModel,
+        clip_model: CLIPModel,
+        tokenizer: CLIPTokenizer,
+        unet: UNet2DConditionModel,
+        scheduler: Union[PNDMScheduler, LMSDiscreteScheduler, DDIMScheduler, DPMSolverMultistepScheduler],
+        feature_extractor: CLIPFeatureExtractor,
+    ):
+        super().__init__()
+        self.register_modules(
+            vae=vae,
+            text_encoder=text_encoder,
+            clip_model=clip_model,
+            tokenizer=tokenizer,
+            unet=unet,
+            scheduler=scheduler,
+            feature_extractor=feature_extractor,
+        )
+
+        self.normalize = transforms.Normalize(mean=feature_extractor.image_mean, std=feature_extractor.image_std)
+        self.cut_out_size = (
+            feature_extractor.size
+            if isinstance(feature_extractor.size, int)
+            else feature_extractor.size["shortest_edge"]
+        )
+        self.make_cutouts = MakeCutouts(self.cut_out_size)
+
+        set_requires_grad(self.text_encoder, False)
+        set_requires_grad(self.clip_model, False)
+
+    def enable_attention_slicing(self, slice_size: Optional[Union[str, int]] = "auto"):
+        if slice_size == "auto":
+            # half the attention head size is usually a good trade-off between
+            # speed and memory
+            slice_size = self.unet.config.attention_head_dim // 2
+        self.unet.set_attention_slice(slice_size)
+
+    def disable_attention_slicing(self):
+        self.enable_attention_slicing(None)
+
+    def freeze_vae(self):
+        set_requires_grad(self.vae, False)
+
+    def unfreeze_vae(self):
+        set_requires_grad(self.vae, True)
+
+    def freeze_unet(self):
+        set_requires_grad(self.unet, False)
+
+    def unfreeze_unet(self):
+        set_requires_grad(self.unet, True)
+
+    def get_timesteps(self, num_inference_steps, strength, device):
+        # get the original timestep using init_timestep
+        init_timestep = min(int(num_inference_steps * strength), num_inference_steps)
+
+        t_start = max(num_inference_steps - init_timestep, 0)
+        timesteps = self.scheduler.timesteps[t_start:]
+
+        return timesteps, num_inference_steps - t_start
+
+    def prepare_latents(self, image, timestep, batch_size, num_images_per_prompt, dtype, device, generator=None):
+        if not isinstance(image, (torch.Tensor, PIL.Image.Image, list)):
+            raise ValueError(
+                f"`image` has to be of type `torch.Tensor`, `PIL.Image.Image` or list but is {type(image)}"
+            )
+
+        image = image.to(device=device, dtype=dtype)
+
+        batch_size = batch_size * num_images_per_prompt
+        if isinstance(generator, list) and len(generator) != batch_size:
+            raise ValueError(
+                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
+                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
+            )
+
+        if isinstance(generator, list):
+            init_latents = [
+                self.vae.encode(image[i : i + 1]).latent_dist.sample(generator[i]) for i in range(batch_size)
+            ]
+            init_latents = torch.cat(init_latents, dim=0)
+        else:
+            init_latents = self.vae.encode(image).latent_dist.sample(generator)
+
+        init_latents = self.vae.config.scaling_factor * init_latents
+
+        if batch_size > init_latents.shape[0] and batch_size % init_latents.shape[0] == 0:
+            # expand init_latents for batch_size
+            deprecation_message = (
+                f"You have passed {batch_size} text prompts (`prompt`), but only {init_latents.shape[0]} initial"
+                " images (`image`). Initial images are now duplicating to match the number of text prompts. Note"
+                " that this behavior is deprecated and will be removed in a version 1.0.0. Please make sure to update"
+                " your script to pass as many initial images as text prompts to suppress this warning."
+            )
+            deprecate("len(prompt) != len(image)", "1.0.0", deprecation_message, standard_warn=False)
+            additional_image_per_prompt = batch_size // init_latents.shape[0]
+            init_latents = torch.cat([init_latents] * additional_image_per_prompt, dim=0)
+        elif batch_size > init_latents.shape[0] and batch_size % init_latents.shape[0] != 0:
+            raise ValueError(
+                f"Cannot duplicate `image` of batch size {init_latents.shape[0]} to {batch_size} text prompts."
+            )
+        else:
+            init_latents = torch.cat([init_latents], dim=0)
+
+        shape = init_latents.shape
+        noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
+
+        # get latents
+        init_latents = self.scheduler.add_noise(init_latents, noise, timestep)
+        latents = init_latents
+
+        return latents
+
+    @torch.enable_grad()
+    def cond_fn(
+        self,
+        latents,
+        timestep,
+        index,
+        text_embeddings,
+        noise_pred_original,
+        text_embeddings_clip,
+        clip_guidance_scale,
+        num_cutouts,
+        use_cutouts=True,
+    ):
+        latents = latents.detach().requires_grad_()
+
+        latent_model_input = self.scheduler.scale_model_input(latents, timestep)
+
+        # predict the noise residual
+        noise_pred = self.unet(latent_model_input, timestep, encoder_hidden_states=text_embeddings).sample
+
+        if isinstance(self.scheduler, (PNDMScheduler, DDIMScheduler, DPMSolverMultistepScheduler)):
+            alpha_prod_t = self.scheduler.alphas_cumprod[timestep]
+            beta_prod_t = 1 - alpha_prod_t
+            # compute predicted original sample from predicted noise also called
+            # "predicted x_0" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf
+            pred_original_sample = (latents - beta_prod_t ** (0.5) * noise_pred) / alpha_prod_t ** (0.5)
+
+            fac = torch.sqrt(beta_prod_t)
+            sample = pred_original_sample * (fac) + latents * (1 - fac)
+        elif isinstance(self.scheduler, LMSDiscreteScheduler):
+            sigma = self.scheduler.sigmas[index]
+            sample = latents - sigma * noise_pred
+        else:
+            raise ValueError(f"scheduler type {type(self.scheduler)} not supported")
+
+        sample = 1 / self.vae.config.scaling_factor * sample
+        image = self.vae.decode(sample).sample
+        image = (image / 2 + 0.5).clamp(0, 1)
+
+        if use_cutouts:
+            image = self.make_cutouts(image, num_cutouts)
+        else:
+            image = transforms.Resize(self.cut_out_size)(image)
+        image = self.normalize(image).to(latents.dtype)
+
+        image_embeddings_clip = self.clip_model.get_image_features(image)
+        image_embeddings_clip = image_embeddings_clip / image_embeddings_clip.norm(p=2, dim=-1, keepdim=True)
+
+        if use_cutouts:
+            dists = spherical_dist_loss(image_embeddings_clip, text_embeddings_clip)
+            dists = dists.view([num_cutouts, sample.shape[0], -1])
+            loss = dists.sum(2).mean(0).sum() * clip_guidance_scale
+        else:
+            loss = spherical_dist_loss(image_embeddings_clip, text_embeddings_clip).mean() * clip_guidance_scale
+
+        grads = -torch.autograd.grad(loss, latents)[0]
+
+        if isinstance(self.scheduler, LMSDiscreteScheduler):
+            latents = latents.detach() + grads * (sigma**2)
+            noise_pred = noise_pred_original
+        else:
+            noise_pred = noise_pred_original - torch.sqrt(beta_prod_t) * grads
+        return noise_pred, latents
+
+    @torch.no_grad()
+    def __call__(
+        self,
+        prompt: Union[str, List[str]],
+        height: Optional[int] = 512,
+        width: Optional[int] = 512,
+        image: Union[torch.FloatTensor, PIL.Image.Image] = None,
+        strength: float = 0.8,
+        num_inference_steps: Optional[int] = 50,
+        guidance_scale: Optional[float] = 7.5,
+        num_images_per_prompt: Optional[int] = 1,
+        eta: float = 0.0,
+        clip_guidance_scale: Optional[float] = 100,
+        clip_prompt: Optional[Union[str, List[str]]] = None,
+        num_cutouts: Optional[int] = 4,
+        use_cutouts: Optional[bool] = True,
+        generator: Optional[torch.Generator] = None,
+        latents: Optional[torch.FloatTensor] = None,
+        output_type: Optional[str] = "pil",
+        return_dict: bool = True,
+    ):
+        if isinstance(prompt, str):
+            batch_size = 1
+        elif isinstance(prompt, list):
+            batch_size = len(prompt)
+        else:
+            raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}")
+
+        if height % 8 != 0 or width % 8 != 0:
+            raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.")
+
+        # get prompt text embeddings
+        text_input = self.tokenizer(
+            prompt,
+            padding="max_length",
+            max_length=self.tokenizer.model_max_length,
+            truncation=True,
+            return_tensors="pt",
+        )
+        text_embeddings = self.text_encoder(text_input.input_ids.to(self.device))[0]
+        # duplicate text embeddings for each generation per prompt
+        text_embeddings = text_embeddings.repeat_interleave(num_images_per_prompt, dim=0)
+
+        # set timesteps
+        accepts_offset = "offset" in set(inspect.signature(self.scheduler.set_timesteps).parameters.keys())
+        extra_set_kwargs = {}
+        if accepts_offset:
+            extra_set_kwargs["offset"] = 1
+
+        self.scheduler.set_timesteps(num_inference_steps, **extra_set_kwargs)
+        # Some schedulers like PNDM have timesteps as arrays
+        # It's more optimized to move all timesteps to correct device beforehand
+        self.scheduler.timesteps.to(self.device)
+
+        timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength, self.device)
+        latent_timestep = timesteps[:1].repeat(batch_size * num_images_per_prompt)
+
+        # Preprocess image
+        image = preprocess(image, width, height)
+        latents = self.prepare_latents(
+            image, latent_timestep, batch_size, num_images_per_prompt, text_embeddings.dtype, self.device, generator
+        )
+
+        if clip_guidance_scale > 0:
+            if clip_prompt is not None:
+                clip_text_input = self.tokenizer(
+                    clip_prompt,
+                    padding="max_length",
+                    max_length=self.tokenizer.model_max_length,
+                    truncation=True,
+                    return_tensors="pt",
+                ).input_ids.to(self.device)
+            else:
+                clip_text_input = text_input.input_ids.to(self.device)
+            text_embeddings_clip = self.clip_model.get_text_features(clip_text_input)
+            text_embeddings_clip = text_embeddings_clip / text_embeddings_clip.norm(p=2, dim=-1, keepdim=True)
+            # duplicate text embeddings clip for each generation per prompt
+            text_embeddings_clip = text_embeddings_clip.repeat_interleave(num_images_per_prompt, dim=0)
+
+        # 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`
+        # corresponds to doing no classifier free guidance.
+        do_classifier_free_guidance = guidance_scale > 1.0
+        # get unconditional embeddings for classifier free guidance
+        if do_classifier_free_guidance:
+            max_length = text_input.input_ids.shape[-1]
+            uncond_input = self.tokenizer([""], padding="max_length", max_length=max_length, return_tensors="pt")
+            uncond_embeddings = self.text_encoder(uncond_input.input_ids.to(self.device))[0]
+            # duplicate unconditional embeddings for each generation per prompt
+            uncond_embeddings = uncond_embeddings.repeat_interleave(num_images_per_prompt, dim=0)
+
+            # For classifier free guidance, we need to do two forward passes.
+            # Here we concatenate the unconditional and text embeddings into a single batch
+            # to avoid doing two forward passes
+            text_embeddings = torch.cat([uncond_embeddings, text_embeddings])
+
+        # get the initial random noise unless the user supplied it
+
+        # Unlike in other pipelines, latents need to be generated in the target device
+        # for 1-to-1 results reproducibility with the CompVis implementation.
+        # However this currently doesn't work in `mps`.
+        latents_shape = (batch_size * num_images_per_prompt, self.unet.config.in_channels, height // 8, width // 8)
+        latents_dtype = text_embeddings.dtype
+        if latents is None:
+            if self.device.type == "mps":
+                # randn does not work reproducibly on mps
+                latents = torch.randn(latents_shape, generator=generator, device="cpu", dtype=latents_dtype).to(
+                    self.device
+                )
+            else:
+                latents = torch.randn(latents_shape, generator=generator, device=self.device, dtype=latents_dtype)
+        else:
+            if latents.shape != latents_shape:
+                raise ValueError(f"Unexpected latents shape, got {latents.shape}, expected {latents_shape}")
+            latents = latents.to(self.device)
+
+        # scale the initial noise by the standard deviation required by the scheduler
+        latents = latents * self.scheduler.init_noise_sigma
+
+        # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
+        # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
+        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
+        # and should be between [0, 1]
+        accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
+        extra_step_kwargs = {}
+        if accepts_eta:
+            extra_step_kwargs["eta"] = eta
+
+        # check if the scheduler accepts generator
+        accepts_generator = "generator" in set(inspect.signature(self.scheduler.step).parameters.keys())
+        if accepts_generator:
+            extra_step_kwargs["generator"] = generator
+
+        with self.progress_bar(total=num_inference_steps):
+            for i, t in enumerate(timesteps):
+                # expand the latents if we are doing classifier free guidance
+                latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
+                latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)
+
+                # predict the noise residual
+                noise_pred = self.unet(latent_model_input, t, encoder_hidden_states=text_embeddings).sample
+
+                # perform classifier free guidance
+                if do_classifier_free_guidance:
+                    noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
+                    noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
+
+                # perform clip guidance
+                if clip_guidance_scale > 0:
+                    text_embeddings_for_guidance = (
+                        text_embeddings.chunk(2)[1] if do_classifier_free_guidance else text_embeddings
+                    )
+                    noise_pred, latents = self.cond_fn(
+                        latents,
+                        t,
+                        i,
+                        text_embeddings_for_guidance,
+                        noise_pred,
+                        text_embeddings_clip,
+                        clip_guidance_scale,
+                        num_cutouts,
+                        use_cutouts,
+                    )
+
+                # compute the previous noisy sample x_t -> x_t-1
+                latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs).prev_sample
+
+        # scale and decode the image latents with vae
+        latents = 1 / self.vae.config.scaling_factor * latents
+        image = self.vae.decode(latents).sample
+
+        image = (image / 2 + 0.5).clamp(0, 1)
+        image = image.cpu().permute(0, 2, 3, 1).numpy()
+
+        if output_type == "pil":
+            image = self.numpy_to_pil(image)
+
+        if not return_dict:
+            return (image, None)
+
+        return StableDiffusionPipelineOutput(images=image, nsfw_content_detected=None)
@@ -17,11 +17,13 @@ from typing import Callable, List, Optional, Union

 import torch
 from packaging import version
-from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

 from diffusers import DiffusionPipeline
 from diffusers.configuration_utils import FrozenDict
 from diffusers.models import AutoencoderKL, UNet2DConditionModel
+from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput
+from diffusers.pipelines.stable_diffusion.safety_checker import StableDiffusionSafetyChecker
 from diffusers.schedulers import (
    DDIMScheduler,
    DPMSolverMultistepScheduler,
@@ -30,11 +32,7 @@ from diffusers.schedulers import (
    LMSDiscreteScheduler,
    PNDMScheduler,
 )
-from diffusers.utils import is_accelerate_available
-
-from ...utils import deprecate, logging
-from . import StableDiffusionPipelineOutput
-from .safety_checker import StableDiffusionSafetyChecker
+from diffusers.utils import deprecate, is_accelerate_available, logging


 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
@@ -64,7 +62,7 @@ class ComposableStableDiffusionPipeline(DiffusionPipeline):
        safety_checker ([`StableDiffusionSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offensive or harmful.
            Please, refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for details.
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """
    _optional_components = ["safety_checker", "feature_extractor"]
@@ -84,7 +82,7 @@ class ComposableStableDiffusionPipeline(DiffusionPipeline):
            DPMSolverMultistepScheduler,
        ],
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
        requires_safety_checker: bool = True,
    ):
        super().__init__()
@@ -513,7 +511,7 @@ class ComposableStableDiffusionPipeline(DiffusionPipeline):
        timesteps = self.scheduler.timesteps

        # 5. Prepare latent variables
-        num_channels_latents = self.unet.in_channels
+        num_channels_latents = self.unet.config.in_channels
        latents = self.prepare_latents(
            batch_size * num_images_per_prompt,
            num_channels_latents,
@@ -15,7 +15,7 @@ from accelerate import Accelerator
 # TODO: remove and import from diffusers.utils when the new version of diffusers is released
 from packaging import version
 from tqdm.auto import tqdm
-from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

 from diffusers import DiffusionPipeline
 from diffusers.models import AutoencoderKL, UNet2DConditionModel
@@ -48,7 +48,7 @@ logger = logging.get_logger(__name__)  # pylint: disable=invalid-name

 def preprocess(image):
    w, h = image.size
-    w, h = map(lambda x: x - x % 32, (w, h))  # resize to integer multiple of 32
+    w, h = (x - x % 32 for x in (w, h))  # resize to integer multiple of 32
    image = image.resize((w, h), resample=PIL_INTERPOLATION["lanczos"])
    image = np.array(image).astype(np.float32) / 255.0
    image = image[None].transpose(0, 3, 1, 2)
@@ -80,7 +80,7 @@ class ImagicStableDiffusionPipeline(DiffusionPipeline):
        safety_checker ([`StableDiffusionSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offsensive or harmful.
            Please, refer to the [model card](https://huggingface.co/CompVis/stable-diffusion-v1-4) for details.
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """

@@ -92,7 +92,7 @@ class ImagicStableDiffusionPipeline(DiffusionPipeline):
        unet: UNet2DConditionModel,
        scheduler: Union[DDIMScheduler, PNDMScheduler, LMSDiscreteScheduler],
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
    ):
        super().__init__()
        self.register_modules(
@@ -424,7 +424,7 @@ class ImagicStableDiffusionPipeline(DiffusionPipeline):
        # Unlike in other pipelines, latents need to be generated in the target device
        # for 1-to-1 results reproducibility with the CompVis implementation.
        # However this currently doesn't work in `mps`.
-        latents_shape = (1, self.unet.in_channels, height // 8, width // 8)
+        latents_shape = (1, self.unet.config.in_channels, height // 8, width // 8)
        latents_dtype = text_embeddings.dtype
        if self.device.type == "mps":
            # randn does not exist on mps
@@ -4,7 +4,7 @@ from typing import Callable, List, Optional, Tuple, Union
 import numpy as np
 import PIL
 import torch
-from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

 from diffusers import DiffusionPipeline
 from diffusers.configuration_utils import FrozenDict
@@ -79,7 +79,7 @@ class ImageToImageInpaintingPipeline(DiffusionPipeline):
        safety_checker ([`StableDiffusionSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offensive or harmful.
            Please, refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for details.
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """

@@ -91,7 +91,7 @@ class ImageToImageInpaintingPipeline(DiffusionPipeline):
        unet: UNet2DConditionModel,
        scheduler: Union[DDIMScheduler, PNDMScheduler, LMSDiscreteScheduler],
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
    ):
        super().__init__()

@@ -5,7 +5,7 @@ from typing import Callable, List, Optional, Union

 import numpy as np
 import torch
-from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

 from diffusers import DiffusionPipeline
 from diffusers.configuration_utils import FrozenDict
@@ -70,7 +70,7 @@ class StableDiffusionWalkPipeline(DiffusionPipeline):
        safety_checker ([`StableDiffusionSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offensive or harmful.
            Please, refer to the [model card](https://huggingface.co/CompVis/stable-diffusion-v1-4) for details.
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """

@@ -82,7 +82,7 @@ class StableDiffusionWalkPipeline(DiffusionPipeline):
        unet: UNet2DConditionModel,
        scheduler: Union[DDIMScheduler, PNDMScheduler, LMSDiscreteScheduler],
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
    ):
        super().__init__()

@@ -320,7 +320,7 @@ class StableDiffusionWalkPipeline(DiffusionPipeline):
        # Unlike in other pipelines, latents need to be generated in the target device
        # for 1-to-1 results reproducibility with the CompVis implementation.
        # However this currently doesn't work in `mps`.
-        latents_shape = (batch_size * num_images_per_prompt, self.unet.in_channels, height // 8, width // 8)
+        latents_shape = (batch_size * num_images_per_prompt, self.unet.config.in_channels, height // 8, width // 8)
        latents_dtype = text_embeddings.dtype
        if latents is None:
            if self.device.type == "mps":
@@ -416,7 +416,7 @@ class StableDiffusionWalkPipeline(DiffusionPipeline):
    def get_noise(self, seed, dtype=torch.float32, height=512, width=512):
        """Takes in random seed and returns corresponding noise vector"""
        return torch.randn(
-            (1, self.unet.in_channels, height // 8, width // 8),
+            (1, self.unet.config.in_channels, height // 8, width // 8),
            generator=torch.Generator(device=self.device).manual_seed(seed),
            device=self.device,
            dtype=dtype,
@@ -6,7 +6,7 @@ import numpy as np
 import PIL
 import torch
 from packaging import version
-from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

 import diffusers
 from diffusers import SchedulerMixin, StableDiffusionPipeline
@@ -179,14 +179,14 @@ def get_prompts_with_weights(pipe: StableDiffusionPipeline, prompt: List[str], m
    return tokens, weights


-def pad_tokens_and_weights(tokens, weights, max_length, bos, eos, no_boseos_middle=True, chunk_length=77):
+def pad_tokens_and_weights(tokens, weights, max_length, bos, eos, pad, no_boseos_middle=True, chunk_length=77):
    r"""
    Pad the tokens (with starting and ending tokens) and weights (with 1.0) to max_length.
    """
    max_embeddings_multiples = (max_length - 2) // (chunk_length - 2)
    weights_length = max_length if no_boseos_middle else max_embeddings_multiples * chunk_length
    for i in range(len(tokens)):
-        tokens[i] = [bos] + tokens[i] + [eos] * (max_length - 1 - len(tokens[i]))
+        tokens[i] = [bos] + tokens[i] + [pad] * (max_length - 1 - len(tokens[i]) - 1) + [eos]
        if no_boseos_middle:
            weights[i] = [1.0] + weights[i] + [1.0] * (max_length - 1 - len(weights[i]))
        else:
@@ -317,12 +317,14 @@ def get_weighted_text_embeddings(
    # pad the length of tokens and weights
    bos = pipe.tokenizer.bos_token_id
    eos = pipe.tokenizer.eos_token_id
+    pad = getattr(pipe.tokenizer, "pad_token_id", eos)
    prompt_tokens, prompt_weights = pad_tokens_and_weights(
        prompt_tokens,
        prompt_weights,
        max_length,
        bos,
        eos,
+        pad,
        no_boseos_middle=no_boseos_middle,
        chunk_length=pipe.tokenizer.model_max_length,
    )
@@ -334,6 +336,7 @@ def get_weighted_text_embeddings(
            max_length,
            bos,
            eos,
+            pad,
            no_boseos_middle=no_boseos_middle,
            chunk_length=pipe.tokenizer.model_max_length,
        )
@@ -376,7 +379,7 @@ def get_weighted_text_embeddings(

 def preprocess_image(image):
    w, h = image.size
-    w, h = map(lambda x: x - x % 32, (w, h))  # resize to integer multiple of 32
+    w, h = (x - x % 32 for x in (w, h))  # resize to integer multiple of 32
    image = image.resize((w, h), resample=PIL_INTERPOLATION["lanczos"])
    image = np.array(image).astype(np.float32) / 255.0
    image = image[None].transpose(0, 3, 1, 2)
@@ -387,7 +390,7 @@ def preprocess_image(image):
 def preprocess_mask(mask, scale_factor=8):
    mask = mask.convert("L")
    w, h = mask.size
-    w, h = map(lambda x: x - x % 32, (w, h))  # resize to integer multiple of 32
+    w, h = (x - x % 32 for x in (w, h))  # resize to integer multiple of 32
    mask = mask.resize((w // scale_factor, h // scale_factor), resample=PIL_INTERPOLATION["nearest"])
    mask = np.array(mask).astype(np.float32) / 255.0
    mask = np.tile(mask, (4, 1, 1))
@@ -422,7 +425,7 @@ class StableDiffusionLongPromptWeightingPipeline(StableDiffusionPipeline):
        safety_checker ([`StableDiffusionSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offensive or harmful.
            Please, refer to the [model card](https://huggingface.co/CompVis/stable-diffusion-v1-4) for details.
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """

@@ -436,7 +439,7 @@ class StableDiffusionLongPromptWeightingPipeline(StableDiffusionPipeline):
            unet: UNet2DConditionModel,
            scheduler: SchedulerMixin,
            safety_checker: StableDiffusionSafetyChecker,
-            feature_extractor: CLIPFeatureExtractor,
+            feature_extractor: CLIPImageProcessor,
            requires_safety_checker: bool = True,
        ):
            super().__init__(
@@ -461,7 +464,7 @@ class StableDiffusionLongPromptWeightingPipeline(StableDiffusionPipeline):
            unet: UNet2DConditionModel,
            scheduler: SchedulerMixin,
            safety_checker: StableDiffusionSafetyChecker,
-            feature_extractor: CLIPFeatureExtractor,
+            feature_extractor: CLIPImageProcessor,
        ):
            super().__init__(
                vae=vae,
@@ -624,7 +627,7 @@ class StableDiffusionLongPromptWeightingPipeline(StableDiffusionPipeline):
        if image is None:
            shape = (
                batch_size,
-                self.unet.in_channels,
+                self.unet.config.in_channels,
                height // self.vae_scale_factor,
                width // self.vae_scale_factor,
            )
@@ -6,7 +6,7 @@ import numpy as np
 import PIL
 import torch
 from packaging import version
-from transformers import CLIPFeatureExtractor, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTokenizer

 import diffusers
 from diffusers import OnnxRuntimeModel, OnnxStableDiffusionPipeline, SchedulerMixin
@@ -196,14 +196,14 @@ def get_prompts_with_weights(pipe, prompt: List[str], max_length: int):
    return tokens, weights


-def pad_tokens_and_weights(tokens, weights, max_length, bos, eos, no_boseos_middle=True, chunk_length=77):
+def pad_tokens_and_weights(tokens, weights, max_length, bos, eos, pad, no_boseos_middle=True, chunk_length=77):
    r"""
    Pad the tokens (with starting and ending tokens) and weights (with 1.0) to max_length.
    """
    max_embeddings_multiples = (max_length - 2) // (chunk_length - 2)
    weights_length = max_length if no_boseos_middle else max_embeddings_multiples * chunk_length
    for i in range(len(tokens)):
-        tokens[i] = [bos] + tokens[i] + [eos] * (max_length - 1 - len(tokens[i]))
+        tokens[i] = [bos] + tokens[i] + [pad] * (max_length - 1 - len(tokens[i]) - 1) + [eos]
        if no_boseos_middle:
            weights[i] = [1.0] + weights[i] + [1.0] * (max_length - 1 - len(weights[i]))
        else:
@@ -342,12 +342,14 @@ def get_weighted_text_embeddings(
    # pad the length of tokens and weights
    bos = pipe.tokenizer.bos_token_id
    eos = pipe.tokenizer.eos_token_id
+    pad = getattr(pipe.tokenizer, "pad_token_id", eos)
    prompt_tokens, prompt_weights = pad_tokens_and_weights(
        prompt_tokens,
        prompt_weights,
        max_length,
        bos,
        eos,
+        pad,
        no_boseos_middle=no_boseos_middle,
        chunk_length=pipe.tokenizer.model_max_length,
    )
@@ -359,6 +361,7 @@ def get_weighted_text_embeddings(
            max_length,
            bos,
            eos,
+            pad,
            no_boseos_middle=no_boseos_middle,
            chunk_length=pipe.tokenizer.model_max_length,
        )
@@ -403,7 +406,7 @@ def get_weighted_text_embeddings(

 def preprocess_image(image):
    w, h = image.size
-    w, h = map(lambda x: x - x % 32, (w, h))  # resize to integer multiple of 32
+    w, h = (x - x % 32 for x in (w, h))  # resize to integer multiple of 32
    image = image.resize((w, h), resample=PIL_INTERPOLATION["lanczos"])
    image = np.array(image).astype(np.float32) / 255.0
    image = image[None].transpose(0, 3, 1, 2)
@@ -413,7 +416,7 @@ def preprocess_image(image):
 def preprocess_mask(mask, scale_factor=8):
    mask = mask.convert("L")
    w, h = mask.size
-    w, h = map(lambda x: x - x % 32, (w, h))  # resize to integer multiple of 32
+    w, h = (x - x % 32 for x in (w, h))  # resize to integer multiple of 32
    mask = mask.resize((w // scale_factor, h // scale_factor), resample=PIL_INTERPOLATION["nearest"])
    mask = np.array(mask).astype(np.float32) / 255.0
    mask = np.tile(mask, (4, 1, 1))
@@ -441,7 +444,7 @@ class OnnxStableDiffusionLongPromptWeightingPipeline(OnnxStableDiffusionPipeline
            unet: OnnxRuntimeModel,
            scheduler: SchedulerMixin,
            safety_checker: OnnxRuntimeModel,
-            feature_extractor: CLIPFeatureExtractor,
+            feature_extractor: CLIPImageProcessor,
            requires_safety_checker: bool = True,
        ):
            super().__init__(
@@ -468,7 +471,7 @@ class OnnxStableDiffusionLongPromptWeightingPipeline(OnnxStableDiffusionPipeline
            unet: OnnxRuntimeModel,
            scheduler: SchedulerMixin,
            safety_checker: OnnxRuntimeModel,
-            feature_extractor: CLIPFeatureExtractor,
+            feature_extractor: CLIPImageProcessor,
        ):
            super().__init__(
                vae_encoder=vae_encoder,
@@ -483,7 +486,7 @@ class OnnxStableDiffusionLongPromptWeightingPipeline(OnnxStableDiffusionPipeline
            self.__init__additional__()

    def __init__additional__(self):
-        self.unet_in_channels = 4
+        self.unet.config.in_channels = 4
        self.vae_scale_factor = 8

    def _encode_prompt(
@@ -618,7 +621,7 @@ class OnnxStableDiffusionLongPromptWeightingPipeline(OnnxStableDiffusionPipeline
        if image is None:
            shape = (
                batch_size,
-                self.unet_in_channels,
+                self.unet.config.in_channels,
                height // self.vae_scale_factor,
                width // self.vae_scale_factor,
            )
@@ -93,7 +93,7 @@ class MagicMixPipeline(DiffusionPipeline):

        torch.manual_seed(seed)
        noise = torch.randn(
-            (1, self.unet.in_channels, height // 8, width // 8),
+            (1, self.unet.config.in_channels, height // 8, width // 8),
        ).to(self.device)

        latents = self.scheduler.add_noise(
@@ -3,7 +3,7 @@ from typing import Callable, List, Optional, Union

 import torch
 from transformers import (
-    CLIPFeatureExtractor,
+    CLIPImageProcessor,
    CLIPTextModel,
    CLIPTokenizer,
    MBart50TokenizerFast,
@@ -79,7 +79,7 @@ class MultilingualStableDiffusion(DiffusionPipeline):
        safety_checker ([`StableDiffusionSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offensive or harmful.
            Please, refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for details.
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """

@@ -94,7 +94,7 @@ class MultilingualStableDiffusion(DiffusionPipeline):
        unet: UNet2DConditionModel,
        scheduler: Union[DDIMScheduler, PNDMScheduler, LMSDiscreteScheduler],
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
    ):
        super().__init__()

@@ -355,7 +355,7 @@ class MultilingualStableDiffusion(DiffusionPipeline):
        # Unlike in other pipelines, latents need to be generated in the target device
        # for 1-to-1 results reproducibility with the CompVis implementation.
        # However this currently doesn't work in `mps`.
-        latents_shape = (batch_size * num_images_per_prompt, self.unet.in_channels, height // 8, width // 8)
+        latents_shape = (batch_size * num_images_per_prompt, self.unet.config.in_channels, height // 8, width // 8)
        latents_dtype = text_embeddings.dtype
        if latents is None:
            if self.device.type == "mps":
@@ -12,7 +12,7 @@ class UnetSchedulerOneForwardPipeline(DiffusionPipeline):

    def __call__(self):
        image = torch.randn(
-            (1, self.unet.in_channels, self.unet.sample_size, self.unet.sample_size),
+            (1, self.unet.config.in_channels, self.unet.config.sample_size, self.unet.config.sample_size),
        )
        timestep = 1

@@ -65,7 +65,7 @@ class StableDiffusionPipeline(DiffusionPipeline):
        safety_checker ([`StableDiffusionSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offensive or harmful.
            Please, refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for details.
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """
    _optional_components = ["safety_checker", "feature_extractor"]
@@ -105,7 +105,7 @@ class StableDiffusionPipeline(DiffusionPipeline):
        )

        model = ModelWrapper(unet, scheduler.alphas_cumprod)
-        if scheduler.prediction_type == "v_prediction":
+        if scheduler.config.prediction_type == "v_prediction":
            self.k_diffusion_model = CompVisVDenoiser(model)
        else:
            self.k_diffusion_model = CompVisDenoiser(model)
@@ -433,7 +433,7 @@ class StableDiffusionPipeline(DiffusionPipeline):
        sigmas = sigmas.to(text_embeddings.dtype)

        # 5. Prepare latent variables
-        num_channels_latents = self.unet.in_channels
+        num_channels_latents = self.unet.config.in_channels
        latents = self.prepare_latents(
            batch_size * num_images_per_prompt,
            num_channels_latents,
@@ -5,7 +5,7 @@ import inspect
 from typing import Callable, List, Optional, Union

 import torch
-from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

 from diffusers import DiffusionPipeline
 from diffusers.models import AutoencoderKL, UNet2DConditionModel
@@ -42,7 +42,7 @@ class SeedResizeStableDiffusionPipeline(DiffusionPipeline):
        safety_checker ([`StableDiffusionSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offensive or harmful.
            Please, refer to the [model card](https://huggingface.co/CompVis/stable-diffusion-v1-4) for details.
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """

@@ -54,7 +54,7 @@ class SeedResizeStableDiffusionPipeline(DiffusionPipeline):
        unet: UNet2DConditionModel,
        scheduler: Union[DDIMScheduler, PNDMScheduler, LMSDiscreteScheduler],
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
    ):
        super().__init__()
        self.register_modules(
@@ -262,8 +262,8 @@ class SeedResizeStableDiffusionPipeline(DiffusionPipeline):
        # Unlike in other pipelines, latents need to be generated in the target device
        # for 1-to-1 results reproducibility with the CompVis implementation.
        # However this currently doesn't work in `mps`.
-        latents_shape = (batch_size * num_images_per_prompt, self.unet.in_channels, height // 8, width // 8)
-        latents_shape_reference = (batch_size * num_images_per_prompt, self.unet.in_channels, 64, 64)
+        latents_shape = (batch_size * num_images_per_prompt, self.unet.config.in_channels, height // 8, width // 8)
+        latents_shape_reference = (batch_size * num_images_per_prompt, self.unet.config.in_channels, 64, 64)
        latents_dtype = text_embeddings.dtype
        if latents is None:
            if self.device.type == "mps":
@@ -3,7 +3,7 @@ from typing import Callable, List, Optional, Union

 import torch
 from transformers import (
-    CLIPFeatureExtractor,
+    CLIPImageProcessor,
    CLIPTextModel,
    CLIPTokenizer,
    WhisperForConditionalGeneration,
@@ -37,7 +37,7 @@ class SpeechToImagePipeline(DiffusionPipeline):
        unet: UNet2DConditionModel,
        scheduler: Union[DDIMScheduler, PNDMScheduler, LMSDiscreteScheduler],
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
    ):
        super().__init__()

@@ -190,7 +190,7 @@ class SpeechToImagePipeline(DiffusionPipeline):
        # Unlike in other pipelines, latents need to be generated in the target device
        # for 1-to-1 results reproducibility with the CompVis implementation.
        # However this currently doesn't work in `mps`.
-        latents_shape = (batch_size * num_images_per_prompt, self.unet.in_channels, height // 8, width // 8)
+        latents_shape = (batch_size * num_images_per_prompt, self.unet.config.in_channels, height // 8, width // 8)
        latents_dtype = text_embeddings.dtype
        if latents is None:
            if self.device.type == "mps":
@@ -1,7 +1,7 @@
 from typing import Any, Callable, Dict, List, Optional, Union

 import torch
-from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

 from diffusers import (
    AutoencoderKL,
@@ -46,7 +46,7 @@ class StableDiffusionComparisonPipeline(DiffusionPipeline):
        safety_checker ([`StableDiffusionMegaSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offensive or harmful.
            Please, refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for details.
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """

@@ -58,7 +58,7 @@ class StableDiffusionComparisonPipeline(DiffusionPipeline):
        unet: UNet2DConditionModel,
        scheduler: Union[DDIMScheduler, PNDMScheduler, LMSDiscreteScheduler],
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
        requires_safety_checker: bool = True,
    ):
        super()._init_()
@@ -1,15 +1,16 @@
 # Inspired by: https://github.com/haofanwang/ControlNet-for-Diffusers/

 import inspect
-from typing import Any, Callable, Dict, List, Optional, Union
+from typing import Any, Callable, Dict, List, Optional, Tuple, Union

 import numpy as np
 import PIL.Image
 import torch
-from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

 from diffusers import AutoencoderKL, ControlNetModel, DiffusionPipeline, UNet2DConditionModel, logging
 from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput, StableDiffusionSafetyChecker
+from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_controlnet import MultiControlNetModel
 from diffusers.schedulers import KarrasDiffusionSchedulers
 from diffusers.utils import (
    PIL_INTERPOLATION,
@@ -86,7 +87,14 @@ def prepare_image(image):


 def prepare_controlnet_conditioning_image(
-    controlnet_conditioning_image, width, height, batch_size, num_images_per_prompt, device, dtype
+    controlnet_conditioning_image,
+    width,
+    height,
+    batch_size,
+    num_images_per_prompt,
+    device,
+    dtype,
+    do_classifier_free_guidance,
 ):
    if not isinstance(controlnet_conditioning_image, torch.Tensor):
        if isinstance(controlnet_conditioning_image, PIL.Image.Image):
@@ -116,6 +124,9 @@ def prepare_controlnet_conditioning_image(

    controlnet_conditioning_image = controlnet_conditioning_image.to(device=device, dtype=dtype)

+    if do_classifier_free_guidance:
+        controlnet_conditioning_image = torch.cat([controlnet_conditioning_image] * 2)
+
    return controlnet_conditioning_image


@@ -132,10 +143,10 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
        text_encoder: CLIPTextModel,
        tokenizer: CLIPTokenizer,
        unet: UNet2DConditionModel,
-        controlnet: ControlNetModel,
+        controlnet: Union[ControlNetModel, List[ControlNetModel], Tuple[ControlNetModel], MultiControlNetModel],
        scheduler: KarrasDiffusionSchedulers,
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
        requires_safety_checker: bool = True,
    ):
        super().__init__()
@@ -156,6 +167,9 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
                " checker. If you do not want to use the safety checker, you can pass `'safety_checker=None'` instead."
            )

+        if isinstance(controlnet, (list, tuple)):
+            controlnet = MultiControlNetModel(controlnet)
+
        self.register_modules(
            vae=vae,
            text_encoder=text_encoder,
@@ -216,7 +230,7 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
        if is_accelerate_available() and is_accelerate_version(">=", "0.17.0.dev0"):
            from accelerate import cpu_offload_with_hook
        else:
-            raise ImportError("`enable_model_offload` requires `accelerate v0.17.0` or higher.")
+            raise ImportError("`enable_model_cpu_offload` requires `accelerate v0.17.0` or higher.")

        device = torch.device(f"cuda:{gpu_id}")

@@ -276,8 +290,7 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
                whether to use classifier free guidance or not
            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. If not defined, one has to pass `negative_prompt_embeds`. instead.
-                Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`).
+                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`).
            prompt_embeds (`torch.FloatTensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                provided, text embeddings will be generated from `prompt` input argument.
@@ -425,6 +438,42 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
            extra_step_kwargs["generator"] = generator
        return extra_step_kwargs

+    def check_controlnet_conditioning_image(self, image, prompt, prompt_embeds):
+        image_is_pil = isinstance(image, PIL.Image.Image)
+        image_is_tensor = isinstance(image, torch.Tensor)
+        image_is_pil_list = isinstance(image, list) and isinstance(image[0], PIL.Image.Image)
+        image_is_tensor_list = isinstance(image, list) and isinstance(image[0], torch.Tensor)
+
+        if not image_is_pil and not image_is_tensor and not image_is_pil_list and not image_is_tensor_list:
+            raise TypeError(
+                "image must be passed and be one of PIL image, torch tensor, list of PIL images, or list of torch tensors"
+            )
+
+        if image_is_pil:
+            image_batch_size = 1
+        elif image_is_tensor:
+            image_batch_size = image.shape[0]
+        elif image_is_pil_list:
+            image_batch_size = len(image)
+        elif image_is_tensor_list:
+            image_batch_size = len(image)
+        else:
+            raise ValueError("controlnet condition image is not valid")
+
+        if prompt is not None and isinstance(prompt, str):
+            prompt_batch_size = 1
+        elif prompt is not None and isinstance(prompt, list):
+            prompt_batch_size = len(prompt)
+        elif prompt_embeds is not None:
+            prompt_batch_size = prompt_embeds.shape[0]
+        else:
+            raise ValueError("prompt or prompt_embeds are not valid")
+
+        if image_batch_size != 1 and image_batch_size != prompt_batch_size:
+            raise ValueError(
+                f"If image batch size is not 1, image batch size must be same as prompt batch size. image batch size: {image_batch_size}, prompt batch size: {prompt_batch_size}"
+            )
+
    def check_inputs(
        self,
        prompt,
@@ -437,6 +486,9 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
        prompt_embeds=None,
        negative_prompt_embeds=None,
        strength=None,
+        controlnet_guidance_start=None,
+        controlnet_guidance_end=None,
+        controlnet_conditioning_scale=None,
    ):
        if height % 8 != 0 or width % 8 != 0:
            raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.")
@@ -475,58 +527,51 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
                    f" {negative_prompt_embeds.shape}."
                )

-        controlnet_cond_image_is_pil = isinstance(controlnet_conditioning_image, PIL.Image.Image)
-        controlnet_cond_image_is_tensor = isinstance(controlnet_conditioning_image, torch.Tensor)
-        controlnet_cond_image_is_pil_list = isinstance(controlnet_conditioning_image, list) and isinstance(
-            controlnet_conditioning_image[0], PIL.Image.Image
-        )
-        controlnet_cond_image_is_tensor_list = isinstance(controlnet_conditioning_image, list) and isinstance(
-            controlnet_conditioning_image[0], torch.Tensor
-        )
+        # check controlnet condition image

-        if (
-            not controlnet_cond_image_is_pil
-            and not controlnet_cond_image_is_tensor
-            and not controlnet_cond_image_is_pil_list
-            and not controlnet_cond_image_is_tensor_list
-        ):
-            raise TypeError(
-                "image must be passed and be one of PIL image, torch tensor, list of PIL images, or list of torch tensors"
-            )
+        if isinstance(self.controlnet, ControlNetModel):
+            self.check_controlnet_conditioning_image(controlnet_conditioning_image, prompt, prompt_embeds)
+        elif isinstance(self.controlnet, MultiControlNetModel):
+            if not isinstance(controlnet_conditioning_image, list):
+                raise TypeError("For multiple controlnets: `image` must be type `list`")

-        if controlnet_cond_image_is_pil:
-            controlnet_cond_image_batch_size = 1
-        elif controlnet_cond_image_is_tensor:
-            controlnet_cond_image_batch_size = controlnet_conditioning_image.shape[0]
-        elif controlnet_cond_image_is_pil_list:
-            controlnet_cond_image_batch_size = len(controlnet_conditioning_image)
-        elif controlnet_cond_image_is_tensor_list:
-            controlnet_cond_image_batch_size = len(controlnet_conditioning_image)
+            if len(controlnet_conditioning_image) != len(self.controlnet.nets):
+                raise ValueError(
+                    "For multiple controlnets: `image` must have the same length as the number of controlnets."
+                )

-        if prompt is not None and isinstance(prompt, str):
-            prompt_batch_size = 1
-        elif prompt is not None and isinstance(prompt, list):
-            prompt_batch_size = len(prompt)
-        elif prompt_embeds is not None:
-            prompt_batch_size = prompt_embeds.shape[0]
+            for image_ in controlnet_conditioning_image:
+                self.check_controlnet_conditioning_image(image_, prompt, prompt_embeds)
+        else:
+            assert False

-        if controlnet_cond_image_batch_size != 1 and controlnet_cond_image_batch_size != prompt_batch_size:
-            raise ValueError(
-                f"If image batch size is not 1, image batch size must be same as prompt batch size. image batch size: {controlnet_cond_image_batch_size}, prompt batch size: {prompt_batch_size}"
-            )
+        # Check `controlnet_conditioning_scale`
+
+        if isinstance(self.controlnet, ControlNetModel):
+            if not isinstance(controlnet_conditioning_scale, float):
+                raise TypeError("For single controlnet: `controlnet_conditioning_scale` must be type `float`.")
+        elif isinstance(self.controlnet, MultiControlNetModel):
+            if isinstance(controlnet_conditioning_scale, list) and len(controlnet_conditioning_scale) != len(
+                self.controlnet.nets
+            ):
+                raise ValueError(
+                    "For multiple controlnets: When `controlnet_conditioning_scale` is specified as `list`, it must have"
+                    " the same length as the number of controlnets"
+                )
+        else:
+            assert False

        if isinstance(image, torch.Tensor):
            if image.ndim != 3 and image.ndim != 4:
                raise ValueError("`image` must have 3 or 4 dimensions")

-            # if mask_image.ndim != 2 and mask_image.ndim != 3 and mask_image.ndim != 4:
-            #     raise ValueError("`mask_image` must have 2, 3, or 4 dimensions")
-
            if image.ndim == 3:
                image_batch_size = 1
                image_channels, image_height, image_width = image.shape
            elif image.ndim == 4:
                image_batch_size, image_channels, image_height, image_width = image.shape
+            else:
+                assert False

            if image_channels != 3:
                raise ValueError("`image` must have 3 channels")
@@ -542,7 +587,23 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
            )

        if strength < 0 or strength > 1:
-            raise ValueError(f"The value of strength should in [0.0, 1.0] but is {strength}")
+            raise ValueError(f"The value of `strength` should in [0.0, 1.0] but is {strength}")
+
+        if controlnet_guidance_start < 0 or controlnet_guidance_start > 1:
+            raise ValueError(
+                f"The value of `controlnet_guidance_start` should in [0.0, 1.0] but is {controlnet_guidance_start}"
+            )
+
+        if controlnet_guidance_end < 0 or controlnet_guidance_end > 1:
+            raise ValueError(
+                f"The value of `controlnet_guidance_end` should in [0.0, 1.0] but is {controlnet_guidance_end}"
+            )
+
+        if controlnet_guidance_start > controlnet_guidance_end:
+            raise ValueError(
+                "The value of `controlnet_guidance_start` should be less than `controlnet_guidance_end`, but got"
+                f" `controlnet_guidance_start` {controlnet_guidance_start} >= `controlnet_guidance_end` {controlnet_guidance_end}"
+            )

    def get_timesteps(self, num_inference_steps, strength, device):
        # get the original timestep using init_timestep
@@ -642,7 +703,9 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
        callback_steps: int = 1,
        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
-        controlnet_conditioning_scale: float = 1.0,
+        controlnet_conditioning_scale: Union[float, List[float]] = 1.0,
+        controlnet_guidance_start: float = 0.0,
+        controlnet_guidance_end: float = 1.0,
    ):
        r"""
        Function invoked when calling the pipeline for generation.
@@ -679,8 +742,7 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
                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. If not defined, one has to pass `negative_prompt_embeds`. instead.
-                Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`).
+                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`).
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
            eta (`float`, *optional*, defaults to 0.0):
@@ -719,6 +781,11 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
            controlnet_conditioning_scale (`float`, *optional*, defaults to 1.0):
                The outputs of the controlnet are multiplied by `controlnet_conditioning_scale` before they are added
                to the residual in the original unet.
+            controlnet_guidance_start ('float', *optional*, defaults to 0.0):
+                The percentage of total steps the controlnet starts applying. Must be between 0 and 1.
+            controlnet_guidance_end ('float', *optional*, defaults to 1.0):
+                The percentage of total steps the controlnet ends applying. Must be between 0 and 1. Must be greater
+                than `controlnet_guidance_start`.

        Examples:

@@ -736,7 +803,6 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
        self.check_inputs(
            prompt,
            image,
-            # mask_image,
            controlnet_conditioning_image,
            height,
            width,
@@ -745,6 +811,9 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
            prompt_embeds,
            negative_prompt_embeds,
            strength,
+            controlnet_guidance_start,
+            controlnet_guidance_end,
+            controlnet_conditioning_scale,
        )

        # 2. Define call parameters
@@ -761,6 +830,9 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
        # corresponds to doing no classifier free guidance.
        do_classifier_free_guidance = guidance_scale > 1.0

+        if isinstance(self.controlnet, MultiControlNetModel) and isinstance(controlnet_conditioning_scale, float):
+            controlnet_conditioning_scale = [controlnet_conditioning_scale] * len(self.controlnet.nets)
+
        # 3. Encode input prompt
        prompt_embeds = self._encode_prompt(
            prompt,
@@ -772,22 +844,41 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
            negative_prompt_embeds=negative_prompt_embeds,
        )

-        # 4. Prepare mask, image, and controlnet_conditioning_image
+        # 4. Prepare image, and controlnet_conditioning_image
        image = prepare_image(image)

-        # mask_image = prepare_mask_image(mask_image)
+        # condition image(s)
+        if isinstance(self.controlnet, ControlNetModel):
+            controlnet_conditioning_image = prepare_controlnet_conditioning_image(
+                controlnet_conditioning_image=controlnet_conditioning_image,
+                width=width,
+                height=height,
+                batch_size=batch_size * num_images_per_prompt,
+                num_images_per_prompt=num_images_per_prompt,
+                device=device,
+                dtype=self.controlnet.dtype,
+                do_classifier_free_guidance=do_classifier_free_guidance,
+            )
+        elif isinstance(self.controlnet, MultiControlNetModel):
+            controlnet_conditioning_images = []

-        controlnet_conditioning_image = prepare_controlnet_conditioning_image(
-            controlnet_conditioning_image,
-            width,
-            height,
-            batch_size * num_images_per_prompt,
-            num_images_per_prompt,
-            device,
-            self.controlnet.dtype,
-        )
+            for image_ in controlnet_conditioning_image:
+                image_ = prepare_controlnet_conditioning_image(
+                    controlnet_conditioning_image=image_,
+                    width=width,
+                    height=height,
+                    batch_size=batch_size * num_images_per_prompt,
+                    num_images_per_prompt=num_images_per_prompt,
+                    device=device,
+                    dtype=self.controlnet.dtype,
+                    do_classifier_free_guidance=do_classifier_free_guidance,
+                )

-        # masked_image = image * (mask_image < 0.5)
+                controlnet_conditioning_images.append(image_)
+
+            controlnet_conditioning_image = controlnet_conditioning_images
+        else:
+            assert False

        # 5. Prepare timesteps
        self.scheduler.set_timesteps(num_inference_steps, device=device)
@@ -805,9 +896,6 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
            generator,
        )

-        if do_classifier_free_guidance:
-            controlnet_conditioning_image = torch.cat([controlnet_conditioning_image] * 2)
-
        # 7. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
        extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

@@ -820,19 +908,26 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):

                latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)

-                down_block_res_samples, mid_block_res_sample = self.controlnet(
-                    latent_model_input,
-                    t,
-                    encoder_hidden_states=prompt_embeds,
-                    controlnet_cond=controlnet_conditioning_image,
-                    return_dict=False,
-                )
+                # compute the percentage of total steps we are at
+                current_sampling_percent = i / len(timesteps)

-                down_block_res_samples = [
-                    down_block_res_sample * controlnet_conditioning_scale
-                    for down_block_res_sample in down_block_res_samples
-                ]
-                mid_block_res_sample *= controlnet_conditioning_scale
+                if (
+                    current_sampling_percent < controlnet_guidance_start
+                    or current_sampling_percent > controlnet_guidance_end
+                ):
+                    # do not apply the controlnet
+                    down_block_res_samples = None
+                    mid_block_res_sample = None
+                else:
+                    # apply the controlnet
+                    down_block_res_samples, mid_block_res_sample = self.controlnet(
+                        latent_model_input,
+                        t,
+                        encoder_hidden_states=prompt_embeds,
+                        controlnet_cond=controlnet_conditioning_image,
+                        conditioning_scale=controlnet_conditioning_scale,
+                        return_dict=False,
+                    )

                # predict the noise residual
                noise_pred = self.unet(
@@ -7,7 +7,7 @@ import numpy as np
 import PIL.Image
 import torch
 import torch.nn.functional as F
-from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

 from diffusers import AutoencoderKL, ControlNetModel, DiffusionPipeline, UNet2DConditionModel, logging
 from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput, StableDiffusionSafetyChecker
@@ -233,7 +233,7 @@ class StableDiffusionControlNetInpaintPipeline(DiffusionPipeline):
        controlnet: ControlNetModel,
        scheduler: KarrasDiffusionSchedulers,
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
        requires_safety_checker: bool = True,
    ):
        super().__init__()
@@ -314,7 +314,7 @@ class StableDiffusionControlNetInpaintPipeline(DiffusionPipeline):
        if is_accelerate_available() and is_accelerate_version(">=", "0.17.0.dev0"):
            from accelerate import cpu_offload_with_hook
        else:
-            raise ImportError("`enable_model_offload` requires `accelerate v0.17.0` or higher.")
+            raise ImportError("`enable_model_cpu_offload` requires `accelerate v0.17.0` or higher.")

        device = torch.device(f"cuda:{gpu_id}")

@@ -373,8 +373,7 @@ class StableDiffusionControlNetInpaintPipeline(DiffusionPipeline):
            do_classifier_free_guidance (`bool`):
                whether to use classifier free guidance or not
            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. If not defined, one has to pass `negative_prompt_embeds`. instead.
+                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 less than `1`).
            prompt_embeds (`torch.FloatTensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
@@ -833,8 +832,7 @@ class StableDiffusionControlNetInpaintPipeline(DiffusionPipeline):
                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. If not defined, one has to pass `negative_prompt_embeds`. instead.
+                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 less than `1`).
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
@@ -7,7 +7,7 @@ import numpy as np
 import PIL.Image
 import torch
 import torch.nn.functional as F
-from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

 from diffusers import AutoencoderKL, ControlNetModel, DiffusionPipeline, UNet2DConditionModel, logging
 from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput, StableDiffusionSafetyChecker
@@ -233,7 +233,7 @@ class StableDiffusionControlNetInpaintImg2ImgPipeline(DiffusionPipeline):
        controlnet: ControlNetModel,
        scheduler: KarrasDiffusionSchedulers,
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
        requires_safety_checker: bool = True,
    ):
        super().__init__()
@@ -314,7 +314,7 @@ class StableDiffusionControlNetInpaintImg2ImgPipeline(DiffusionPipeline):
        if is_accelerate_available() and is_accelerate_version(">=", "0.17.0.dev0"):
            from accelerate import cpu_offload_with_hook
        else:
-            raise ImportError("`enable_model_offload` requires `accelerate v0.17.0` or higher.")
+            raise ImportError("`enable_model_cpu_offload` requires `accelerate v0.17.0` or higher.")

        device = torch.device(f"cuda:{gpu_id}")

@@ -373,8 +373,7 @@ class StableDiffusionControlNetInpaintImg2ImgPipeline(DiffusionPipeline):
            do_classifier_free_guidance (`bool`):
                whether to use classifier free guidance or not
            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. If not defined, one has to pass `negative_prompt_embeds`. instead.
+                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 less than `1`).
            prompt_embeds (`torch.FloatTensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
@@ -876,8 +875,7 @@ class StableDiffusionControlNetInpaintImg2ImgPipeline(DiffusionPipeline):
                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. If not defined, one has to pass `negative_prompt_embeds`. instead.
+                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 less than `1`).
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
@@ -2,7 +2,7 @@ from typing import Any, Callable, Dict, List, Optional, Union

 import PIL.Image
 import torch
-from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

 from diffusers import (
    AutoencoderKL,
@@ -47,7 +47,7 @@ class StableDiffusionMegaPipeline(DiffusionPipeline):
        safety_checker ([`StableDiffusionMegaSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offensive or harmful.
            Please, refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for details.
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """
    _optional_components = ["safety_checker", "feature_extractor"]
@@ -60,7 +60,7 @@ class StableDiffusionMegaPipeline(DiffusionPipeline):
        unet: UNet2DConditionModel,
        scheduler: Union[DDIMScheduler, PNDMScheduler, LMSDiscreteScheduler],
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
        requires_safety_checker: bool = True,
    ):
        super().__init__()
@@ -0,0 +1,926 @@
+#
+# Copyright 2023 The HuggingFace Inc. team.
+# SPDX-FileCopyrightText: Copyright (c) 1993-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# 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
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import os
+from collections import OrderedDict
+from copy import copy
+from typing import List, Optional, Union
+
+import numpy as np
+import onnx
+import onnx_graphsurgeon as gs
+import tensorrt as trt
+import torch
+from huggingface_hub import snapshot_download
+from onnx import shape_inference
+from polygraphy import cuda
+from polygraphy.backend.common import bytes_from_path
+from polygraphy.backend.onnx.loader import fold_constants
+from polygraphy.backend.trt import (
+    CreateConfig,
+    Profile,
+    engine_from_bytes,
+    engine_from_network,
+    network_from_onnx_path,
+    save_engine,
+)
+from polygraphy.backend.trt import util as trt_util
+from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
+
+from diffusers.models import AutoencoderKL, UNet2DConditionModel
+from diffusers.pipelines.stable_diffusion import (
+    StableDiffusionPipeline,
+    StableDiffusionPipelineOutput,
+    StableDiffusionSafetyChecker,
+)
+from diffusers.schedulers import DDIMScheduler
+from diffusers.utils import DIFFUSERS_CACHE, logging
+
+
+"""
+Installation instructions
+python3 -m pip install --upgrade tensorrt
+python3 -m pip install --upgrade polygraphy onnx-graphsurgeon --extra-index-url https://pypi.ngc.nvidia.com
+python3 -m pip install onnxruntime
+"""
+
+TRT_LOGGER = trt.Logger(trt.Logger.ERROR)
+logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
+
+# Map of numpy dtype -> torch dtype
+numpy_to_torch_dtype_dict = {
+    np.uint8: torch.uint8,
+    np.int8: torch.int8,
+    np.int16: torch.int16,
+    np.int32: torch.int32,
+    np.int64: torch.int64,
+    np.float16: torch.float16,
+    np.float32: torch.float32,
+    np.float64: torch.float64,
+    np.complex64: torch.complex64,
+    np.complex128: torch.complex128,
+}
+if np.version.full_version >= "1.24.0":
+    numpy_to_torch_dtype_dict[np.bool_] = torch.bool
+else:
+    numpy_to_torch_dtype_dict[np.bool] = torch.bool
+
+# Map of torch dtype -> numpy dtype
+torch_to_numpy_dtype_dict = {value: key for (key, value) in numpy_to_torch_dtype_dict.items()}
+
+
+def device_view(t):
+    return cuda.DeviceView(ptr=t.data_ptr(), shape=t.shape, dtype=torch_to_numpy_dtype_dict[t.dtype])
+
+
+class Engine:
+    def __init__(self, engine_path):
+        self.engine_path = engine_path
+        self.engine = None
+        self.context = None
+        self.buffers = OrderedDict()
+        self.tensors = OrderedDict()
+
+    def __del__(self):
+        [buf.free() for buf in self.buffers.values() if isinstance(buf, cuda.DeviceArray)]
+        del self.engine
+        del self.context
+        del self.buffers
+        del self.tensors
+
+    def build(
+        self,
+        onnx_path,
+        fp16,
+        input_profile=None,
+        enable_preview=False,
+        enable_all_tactics=False,
+        timing_cache=None,
+        workspace_size=0,
+    ):
+        logger.warning(f"Building TensorRT engine for {onnx_path}: {self.engine_path}")
+        p = Profile()
+        if input_profile:
+            for name, dims in input_profile.items():
+                assert len(dims) == 3
+                p.add(name, min=dims[0], opt=dims[1], max=dims[2])
+
+        config_kwargs = {}
+
+        config_kwargs["preview_features"] = [trt.PreviewFeature.DISABLE_EXTERNAL_TACTIC_SOURCES_FOR_CORE_0805]
+        if enable_preview:
+            # Faster dynamic shapes made optional since it increases engine build time.
+            config_kwargs["preview_features"].append(trt.PreviewFeature.FASTER_DYNAMIC_SHAPES_0805)
+        if workspace_size > 0:
+            config_kwargs["memory_pool_limits"] = {trt.MemoryPoolType.WORKSPACE: workspace_size}
+        if not enable_all_tactics:
+            config_kwargs["tactic_sources"] = []
+
+        engine = engine_from_network(
+            network_from_onnx_path(onnx_path),
+            config=CreateConfig(fp16=fp16, profiles=[p], load_timing_cache=timing_cache, **config_kwargs),
+            save_timing_cache=timing_cache,
+        )
+        save_engine(engine, path=self.engine_path)
+
+    def load(self):
+        logger.warning(f"Loading TensorRT engine: {self.engine_path}")
+        self.engine = engine_from_bytes(bytes_from_path(self.engine_path))
+
+    def activate(self):
+        self.context = self.engine.create_execution_context()
+
+    def allocate_buffers(self, shape_dict=None, device="cuda"):
+        for idx in range(trt_util.get_bindings_per_profile(self.engine)):
+            binding = self.engine[idx]
+            if shape_dict and binding in shape_dict:
+                shape = shape_dict[binding]
+            else:
+                shape = self.engine.get_binding_shape(binding)
+            dtype = trt.nptype(self.engine.get_binding_dtype(binding))
+            if self.engine.binding_is_input(binding):
+                self.context.set_binding_shape(idx, shape)
+            tensor = torch.empty(tuple(shape), dtype=numpy_to_torch_dtype_dict[dtype]).to(device=device)
+            self.tensors[binding] = tensor
+            self.buffers[binding] = cuda.DeviceView(ptr=tensor.data_ptr(), shape=shape, dtype=dtype)
+
+    def infer(self, feed_dict, stream):
+        start_binding, end_binding = trt_util.get_active_profile_bindings(self.context)
+        # shallow copy of ordered dict
+        device_buffers = copy(self.buffers)
+        for name, buf in feed_dict.items():
+            assert isinstance(buf, cuda.DeviceView)
+            device_buffers[name] = buf
+        bindings = [0] * start_binding + [buf.ptr for buf in device_buffers.values()]
+        noerror = self.context.execute_async_v2(bindings=bindings, stream_handle=stream.ptr)
+        if not noerror:
+            raise ValueError("ERROR: inference failed.")
+
+        return self.tensors
+
+
+class Optimizer:
+    def __init__(self, onnx_graph):
+        self.graph = gs.import_onnx(onnx_graph)
+
+    def cleanup(self, return_onnx=False):
+        self.graph.cleanup().toposort()
+        if return_onnx:
+            return gs.export_onnx(self.graph)
+
+    def select_outputs(self, keep, names=None):
+        self.graph.outputs = [self.graph.outputs[o] for o in keep]
+        if names:
+            for i, name in enumerate(names):
+                self.graph.outputs[i].name = name
+
+    def fold_constants(self, return_onnx=False):
+        onnx_graph = fold_constants(gs.export_onnx(self.graph), allow_onnxruntime_shape_inference=True)
+        self.graph = gs.import_onnx(onnx_graph)
+        if return_onnx:
+            return onnx_graph
+
+    def infer_shapes(self, return_onnx=False):
+        onnx_graph = gs.export_onnx(self.graph)
+        if onnx_graph.ByteSize() > 2147483648:
+            raise TypeError("ERROR: model size exceeds supported 2GB limit")
+        else:
+            onnx_graph = shape_inference.infer_shapes(onnx_graph)
+
+        self.graph = gs.import_onnx(onnx_graph)
+        if return_onnx:
+            return onnx_graph
+
+
+class BaseModel:
+    def __init__(self, model, fp16=False, device="cuda", max_batch_size=16, embedding_dim=768, text_maxlen=77):
+        self.model = model
+        self.name = "SD Model"
+        self.fp16 = fp16
+        self.device = device
+
+        self.min_batch = 1
+        self.max_batch = max_batch_size
+        self.min_image_shape = 256  # min image resolution: 256x256
+        self.max_image_shape = 1024  # max image resolution: 1024x1024
+        self.min_latent_shape = self.min_image_shape // 8
+        self.max_latent_shape = self.max_image_shape // 8
+
+        self.embedding_dim = embedding_dim
+        self.text_maxlen = text_maxlen
+
+    def get_model(self):
+        return self.model
+
+    def get_input_names(self):
+        pass
+
+    def get_output_names(self):
+        pass
+
+    def get_dynamic_axes(self):
+        return None
+
+    def get_sample_input(self, batch_size, image_height, image_width):
+        pass
+
+    def get_input_profile(self, batch_size, image_height, image_width, static_batch, static_shape):
+        return None
+
+    def get_shape_dict(self, batch_size, image_height, image_width):
+        return None
+
+    def optimize(self, onnx_graph):
+        opt = Optimizer(onnx_graph)
+        opt.cleanup()
+        opt.fold_constants()
+        opt.infer_shapes()
+        onnx_opt_graph = opt.cleanup(return_onnx=True)
+        return onnx_opt_graph
+
+    def check_dims(self, batch_size, image_height, image_width):
+        assert batch_size >= self.min_batch and batch_size <= self.max_batch
+        assert image_height % 8 == 0 or image_width % 8 == 0
+        latent_height = image_height // 8
+        latent_width = image_width // 8
+        assert latent_height >= self.min_latent_shape and latent_height <= self.max_latent_shape
+        assert latent_width >= self.min_latent_shape and latent_width <= self.max_latent_shape
+        return (latent_height, latent_width)
+
+    def get_minmax_dims(self, batch_size, image_height, image_width, static_batch, static_shape):
+        min_batch = batch_size if static_batch else self.min_batch
+        max_batch = batch_size if static_batch else self.max_batch
+        latent_height = image_height // 8
+        latent_width = image_width // 8
+        min_image_height = image_height if static_shape else self.min_image_shape
+        max_image_height = image_height if static_shape else self.max_image_shape
+        min_image_width = image_width if static_shape else self.min_image_shape
+        max_image_width = image_width if static_shape else self.max_image_shape
+        min_latent_height = latent_height if static_shape else self.min_latent_shape
+        max_latent_height = latent_height if static_shape else self.max_latent_shape
+        min_latent_width = latent_width if static_shape else self.min_latent_shape
+        max_latent_width = latent_width if static_shape else self.max_latent_shape
+        return (
+            min_batch,
+            max_batch,
+            min_image_height,
+            max_image_height,
+            min_image_width,
+            max_image_width,
+            min_latent_height,
+            max_latent_height,
+            min_latent_width,
+            max_latent_width,
+        )
+
+
+def getOnnxPath(model_name, onnx_dir, opt=True):
+    return os.path.join(onnx_dir, model_name + (".opt" if opt else "") + ".onnx")
+
+
+def getEnginePath(model_name, engine_dir):
+    return os.path.join(engine_dir, model_name + ".plan")
+
+
+def build_engines(
+    models: dict,
+    engine_dir,
+    onnx_dir,
+    onnx_opset,
+    opt_image_height,
+    opt_image_width,
+    opt_batch_size=1,
+    force_engine_rebuild=False,
+    static_batch=False,
+    static_shape=True,
+    enable_preview=False,
+    enable_all_tactics=False,
+    timing_cache=None,
+    max_workspace_size=0,
+):
+    built_engines = {}
+    if not os.path.isdir(onnx_dir):
+        os.makedirs(onnx_dir)
+    if not os.path.isdir(engine_dir):
+        os.makedirs(engine_dir)
+
+    # Export models to ONNX
+    for model_name, model_obj in models.items():
+        engine_path = getEnginePath(model_name, engine_dir)
+        if force_engine_rebuild or not os.path.exists(engine_path):
+            logger.warning("Building Engines...")
+            logger.warning("Engine build can take a while to complete")
+            onnx_path = getOnnxPath(model_name, onnx_dir, opt=False)
+            onnx_opt_path = getOnnxPath(model_name, onnx_dir)
+            if force_engine_rebuild or not os.path.exists(onnx_opt_path):
+                if force_engine_rebuild or not os.path.exists(onnx_path):
+                    logger.warning(f"Exporting model: {onnx_path}")
+                    model = model_obj.get_model()
+                    with torch.inference_mode(), torch.autocast("cuda"):
+                        inputs = model_obj.get_sample_input(opt_batch_size, opt_image_height, opt_image_width)
+                        torch.onnx.export(
+                            model,
+                            inputs,
+                            onnx_path,
+                            export_params=True,
+                            opset_version=onnx_opset,
+                            do_constant_folding=True,
+                            input_names=model_obj.get_input_names(),
+                            output_names=model_obj.get_output_names(),
+                            dynamic_axes=model_obj.get_dynamic_axes(),
+                        )
+                    del model
+                    torch.cuda.empty_cache()
+                    gc.collect()
+                else:
+                    logger.warning(f"Found cached model: {onnx_path}")
+
+                # Optimize onnx
+                if force_engine_rebuild or not os.path.exists(onnx_opt_path):
+                    logger.warning(f"Generating optimizing model: {onnx_opt_path}")
+                    onnx_opt_graph = model_obj.optimize(onnx.load(onnx_path))
+                    onnx.save(onnx_opt_graph, onnx_opt_path)
+                else:
+                    logger.warning(f"Found cached optimized model: {onnx_opt_path} ")
+
+    # Build TensorRT engines
+    for model_name, model_obj in models.items():
+        engine_path = getEnginePath(model_name, engine_dir)
+        engine = Engine(engine_path)
+        onnx_path = getOnnxPath(model_name, onnx_dir, opt=False)
+        onnx_opt_path = getOnnxPath(model_name, onnx_dir)
+
+        if force_engine_rebuild or not os.path.exists(engine.engine_path):
+            engine.build(
+                onnx_opt_path,
+                fp16=True,
+                input_profile=model_obj.get_input_profile(
+                    opt_batch_size,
+                    opt_image_height,
+                    opt_image_width,
+                    static_batch=static_batch,
+                    static_shape=static_shape,
+                ),
+                enable_preview=enable_preview,
+                timing_cache=timing_cache,
+                workspace_size=max_workspace_size,
+            )
+        built_engines[model_name] = engine
+
+    # Load and activate TensorRT engines
+    for model_name, model_obj in models.items():
+        engine = built_engines[model_name]
+        engine.load()
+        engine.activate()
+
+    return built_engines
+
+
+def runEngine(engine, feed_dict, stream):
+    return engine.infer(feed_dict, stream)
+
+
+class CLIP(BaseModel):
+    def __init__(self, model, device, max_batch_size, embedding_dim):
+        super(CLIP, self).__init__(
+            model=model, device=device, max_batch_size=max_batch_size, embedding_dim=embedding_dim
+        )
+        self.name = "CLIP"
+
+    def get_input_names(self):
+        return ["input_ids"]
+
+    def get_output_names(self):
+        return ["text_embeddings", "pooler_output"]
+
+    def get_dynamic_axes(self):
+        return {"input_ids": {0: "B"}, "text_embeddings": {0: "B"}}
+
+    def get_input_profile(self, batch_size, image_height, image_width, static_batch, static_shape):
+        self.check_dims(batch_size, image_height, image_width)
+        min_batch, max_batch, _, _, _, _, _, _, _, _ = self.get_minmax_dims(
+            batch_size, image_height, image_width, static_batch, static_shape
+        )
+        return {
+            "input_ids": [(min_batch, self.text_maxlen), (batch_size, self.text_maxlen), (max_batch, self.text_maxlen)]
+        }
+
+    def get_shape_dict(self, batch_size, image_height, image_width):
+        self.check_dims(batch_size, image_height, image_width)
+        return {
+            "input_ids": (batch_size, self.text_maxlen),
+            "text_embeddings": (batch_size, self.text_maxlen, self.embedding_dim),
+        }
+
+    def get_sample_input(self, batch_size, image_height, image_width):
+        self.check_dims(batch_size, image_height, image_width)
+        return torch.zeros(batch_size, self.text_maxlen, dtype=torch.int32, device=self.device)
+
+    def optimize(self, onnx_graph):
+        opt = Optimizer(onnx_graph)
+        opt.select_outputs([0])  # delete graph output#1
+        opt.cleanup()
+        opt.fold_constants()
+        opt.infer_shapes()
+        opt.select_outputs([0], names=["text_embeddings"])  # rename network output
+        opt_onnx_graph = opt.cleanup(return_onnx=True)
+        return opt_onnx_graph
+
+
+def make_CLIP(model, device, max_batch_size, embedding_dim, inpaint=False):
+    return CLIP(model, device=device, max_batch_size=max_batch_size, embedding_dim=embedding_dim)
+
+
+class UNet(BaseModel):
+    def __init__(
+        self, model, fp16=False, device="cuda", max_batch_size=16, embedding_dim=768, text_maxlen=77, unet_dim=4
+    ):
+        super(UNet, self).__init__(
+            model=model,
+            fp16=fp16,
+            device=device,
+            max_batch_size=max_batch_size,
+            embedding_dim=embedding_dim,
+            text_maxlen=text_maxlen,
+        )
+        self.unet_dim = unet_dim
+        self.name = "UNet"
+
+    def get_input_names(self):
+        return ["sample", "timestep", "encoder_hidden_states"]
+
+    def get_output_names(self):
+        return ["latent"]
+
+    def get_dynamic_axes(self):
+        return {
+            "sample": {0: "2B", 2: "H", 3: "W"},
+            "encoder_hidden_states": {0: "2B"},
+            "latent": {0: "2B", 2: "H", 3: "W"},
+        }
+
+    def get_input_profile(self, batch_size, image_height, image_width, static_batch, static_shape):
+        latent_height, latent_width = self.check_dims(batch_size, image_height, image_width)
+        (
+            min_batch,
+            max_batch,
+            _,
+            _,
+            _,
+            _,
+            min_latent_height,
+            max_latent_height,
+            min_latent_width,
+            max_latent_width,
+        ) = self.get_minmax_dims(batch_size, image_height, image_width, static_batch, static_shape)
+        return {
+            "sample": [
+                (2 * min_batch, self.unet_dim, min_latent_height, min_latent_width),
+                (2 * batch_size, self.unet_dim, latent_height, latent_width),
+                (2 * max_batch, self.unet_dim, max_latent_height, max_latent_width),
+            ],
+            "encoder_hidden_states": [
+                (2 * min_batch, self.text_maxlen, self.embedding_dim),
+                (2 * batch_size, self.text_maxlen, self.embedding_dim),
+                (2 * max_batch, self.text_maxlen, self.embedding_dim),
+            ],
+        }
+
+    def get_shape_dict(self, batch_size, image_height, image_width):
+        latent_height, latent_width = self.check_dims(batch_size, image_height, image_width)
+        return {
+            "sample": (2 * batch_size, self.unet_dim, latent_height, latent_width),
+            "encoder_hidden_states": (2 * batch_size, self.text_maxlen, self.embedding_dim),
+            "latent": (2 * batch_size, 4, latent_height, latent_width),
+        }
+
+    def get_sample_input(self, batch_size, image_height, image_width):
+        latent_height, latent_width = self.check_dims(batch_size, image_height, image_width)
+        dtype = torch.float16 if self.fp16 else torch.float32
+        return (
+            torch.randn(
+                2 * batch_size, self.unet_dim, latent_height, latent_width, dtype=torch.float32, device=self.device
+            ),
+            torch.tensor([1.0], dtype=torch.float32, device=self.device),
+            torch.randn(2 * batch_size, self.text_maxlen, self.embedding_dim, dtype=dtype, device=self.device),
+        )
+
+
+def make_UNet(model, device, max_batch_size, embedding_dim, inpaint=False):
+    return UNet(
+        model,
+        fp16=True,
+        device=device,
+        max_batch_size=max_batch_size,
+        embedding_dim=embedding_dim,
+        unet_dim=(9 if inpaint else 4),
+    )
+
+
+class VAE(BaseModel):
+    def __init__(self, model, device, max_batch_size, embedding_dim):
+        super(VAE, self).__init__(
+            model=model, device=device, max_batch_size=max_batch_size, embedding_dim=embedding_dim
+        )
+        self.name = "VAE decoder"
+
+    def get_input_names(self):
+        return ["latent"]
+
+    def get_output_names(self):
+        return ["images"]
+
+    def get_dynamic_axes(self):
+        return {"latent": {0: "B", 2: "H", 3: "W"}, "images": {0: "B", 2: "8H", 3: "8W"}}
+
+    def get_input_profile(self, batch_size, image_height, image_width, static_batch, static_shape):
+        latent_height, latent_width = self.check_dims(batch_size, image_height, image_width)
+        (
+            min_batch,
+            max_batch,
+            _,
+            _,
+            _,
+            _,
+            min_latent_height,
+            max_latent_height,
+            min_latent_width,
+            max_latent_width,
+        ) = self.get_minmax_dims(batch_size, image_height, image_width, static_batch, static_shape)
+        return {
+            "latent": [
+                (min_batch, 4, min_latent_height, min_latent_width),
+                (batch_size, 4, latent_height, latent_width),
+                (max_batch, 4, max_latent_height, max_latent_width),
+            ]
+        }
+
+    def get_shape_dict(self, batch_size, image_height, image_width):
+        latent_height, latent_width = self.check_dims(batch_size, image_height, image_width)
+        return {
+            "latent": (batch_size, 4, latent_height, latent_width),
+            "images": (batch_size, 3, image_height, image_width),
+        }
+
+    def get_sample_input(self, batch_size, image_height, image_width):
+        latent_height, latent_width = self.check_dims(batch_size, image_height, image_width)
+        return torch.randn(batch_size, 4, latent_height, latent_width, dtype=torch.float32, device=self.device)
+
+
+def make_VAE(model, device, max_batch_size, embedding_dim, inpaint=False):
+    return VAE(model, device=device, max_batch_size=max_batch_size, embedding_dim=embedding_dim)
+
+
+class TensorRTStableDiffusionPipeline(StableDiffusionPipeline):
+    r"""
+    Pipeline for text-to-image generation using TensorRT accelerated Stable Diffusion.
+
+    This model inherits from [`StableDiffusionPipeline`]. Check the superclass documentation for the generic methods the
+    library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.)
+
+    Args:
+        vae ([`AutoencoderKL`]):
+            Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations.
+        text_encoder ([`CLIPTextModel`]):
+            Frozen text-encoder. Stable Diffusion uses the text portion of
+            [CLIP](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel), specifically
+            the [clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14) variant.
+        tokenizer (`CLIPTokenizer`):
+            Tokenizer of class
+            [CLIPTokenizer](https://huggingface.co/docs/transformers/v4.21.0/en/model_doc/clip#transformers.CLIPTokenizer).
+        unet ([`UNet2DConditionModel`]): Conditional U-Net architecture to denoise the encoded image latents.
+        scheduler ([`SchedulerMixin`]):
+            A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
+            [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
+        safety_checker ([`StableDiffusionSafetyChecker`]):
+            Classification module that estimates whether generated images could be considered offensive or harmful.
+            Please, refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for details.
+        feature_extractor ([`CLIPFeatureExtractor`]):
+            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
+    """
+
+    def __init__(
+        self,
+        vae: AutoencoderKL,
+        text_encoder: CLIPTextModel,
+        tokenizer: CLIPTokenizer,
+        unet: UNet2DConditionModel,
+        scheduler: DDIMScheduler,
+        safety_checker: StableDiffusionSafetyChecker,
+        feature_extractor: CLIPFeatureExtractor,
+        requires_safety_checker: bool = True,
+        stages=["clip", "unet", "vae"],
+        image_height: int = 768,
+        image_width: int = 768,
+        max_batch_size: int = 16,
+        # ONNX export parameters
+        onnx_opset: int = 17,
+        onnx_dir: str = "onnx",
+        # TensorRT engine build parameters
+        engine_dir: str = "engine",
+        force_engine_rebuild: bool = False,
+        timing_cache: str = "timing_cache",
+    ):
+        super().__init__(
+            vae, text_encoder, tokenizer, unet, scheduler, safety_checker, feature_extractor, requires_safety_checker
+        )
+
+        self.vae.forward = self.vae.decode
+
+        self.stages = stages
+        self.image_height, self.image_width = image_height, image_width
+        self.inpaint = False
+        self.onnx_opset = onnx_opset
+        self.onnx_dir = onnx_dir
+        self.engine_dir = engine_dir
+        self.force_engine_rebuild = force_engine_rebuild
+        self.timing_cache = timing_cache
+        self.build_static_batch = False
+        self.build_dynamic_shape = False
+        self.build_preview_features = False
+
+        self.max_batch_size = max_batch_size
+        # TODO: Restrict batch size to 4 for larger image dimensions as a WAR for TensorRT limitation.
+        if self.build_dynamic_shape or self.image_height > 512 or self.image_width > 512:
+            self.max_batch_size = 4
+
+        self.stream = None  # loaded in loadResources()
+        self.models = {}  # loaded in __loadModels()
+        self.engine = {}  # loaded in build_engines()
+
+    def __loadModels(self):
+        # Load pipeline models
+        self.embedding_dim = self.text_encoder.config.hidden_size
+        models_args = {
+            "device": self.torch_device,
+            "max_batch_size": self.max_batch_size,
+            "embedding_dim": self.embedding_dim,
+            "inpaint": self.inpaint,
+        }
+        if "clip" in self.stages:
+            self.models["clip"] = make_CLIP(self.text_encoder, **models_args)
+        if "unet" in self.stages:
+            self.models["unet"] = make_UNet(self.unet, **models_args)
+        if "vae" in self.stages:
+            self.models["vae"] = make_VAE(self.vae, **models_args)
+
+    @classmethod
+    def set_cached_folder(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], **kwargs):
+        cache_dir = kwargs.pop("cache_dir", DIFFUSERS_CACHE)
+        resume_download = kwargs.pop("resume_download", False)
+        proxies = kwargs.pop("proxies", None)
+        local_files_only = kwargs.pop("local_files_only", False)
+        use_auth_token = kwargs.pop("use_auth_token", None)
+        revision = kwargs.pop("revision", None)
+
+        cls.cached_folder = (
+            pretrained_model_name_or_path
+            if os.path.isdir(pretrained_model_name_or_path)
+            else snapshot_download(
+                pretrained_model_name_or_path,
+                cache_dir=cache_dir,
+                resume_download=resume_download,
+                proxies=proxies,
+                local_files_only=local_files_only,
+                use_auth_token=use_auth_token,
+                revision=revision,
+            )
+        )
+
+    def to(self, torch_device: Optional[Union[str, torch.device]] = None, silence_dtype_warnings: bool = False):
+        super().to(torch_device, silence_dtype_warnings)
+
+        self.onnx_dir = os.path.join(self.cached_folder, self.onnx_dir)
+        self.engine_dir = os.path.join(self.cached_folder, self.engine_dir)
+        self.timing_cache = os.path.join(self.cached_folder, self.timing_cache)
+
+        # set device
+        self.torch_device = self._execution_device
+        logger.warning(f"Running inference on device: {self.torch_device}")
+
+        # load models
+        self.__loadModels()
+
+        # build engines
+        self.engine = build_engines(
+            self.models,
+            self.engine_dir,
+            self.onnx_dir,
+            self.onnx_opset,
+            opt_image_height=self.image_height,
+            opt_image_width=self.image_width,
+            force_engine_rebuild=self.force_engine_rebuild,
+            static_batch=self.build_static_batch,
+            static_shape=not self.build_dynamic_shape,
+            enable_preview=self.build_preview_features,
+            timing_cache=self.timing_cache,
+        )
+
+        return self
+
+    def __encode_prompt(self, prompt, negative_prompt):
+        r"""
+        Encodes the prompt into text encoder hidden states.
+
+        Args:
+             prompt (`str` or `List[str]`, *optional*):
+                prompt to be encoded
+            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. If not defined, one has to pass `negative_prompt_embeds`. instead.
+                Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`).
+        """
+        # Tokenize prompt
+        text_input_ids = (
+            self.tokenizer(
+                prompt,
+                padding="max_length",
+                max_length=self.tokenizer.model_max_length,
+                truncation=True,
+                return_tensors="pt",
+            )
+            .input_ids.type(torch.int32)
+            .to(self.torch_device)
+        )
+
+        text_input_ids_inp = device_view(text_input_ids)
+        # NOTE: output tensor for CLIP must be cloned because it will be overwritten when called again for negative prompt
+        text_embeddings = runEngine(self.engine["clip"], {"input_ids": text_input_ids_inp}, self.stream)[
+            "text_embeddings"
+        ].clone()
+
+        # Tokenize negative prompt
+        uncond_input_ids = (
+            self.tokenizer(
+                negative_prompt,
+                padding="max_length",
+                max_length=self.tokenizer.model_max_length,
+                truncation=True,
+                return_tensors="pt",
+            )
+            .input_ids.type(torch.int32)
+            .to(self.torch_device)
+        )
+        uncond_input_ids_inp = device_view(uncond_input_ids)
+        uncond_embeddings = runEngine(self.engine["clip"], {"input_ids": uncond_input_ids_inp}, self.stream)[
+            "text_embeddings"
+        ]
+
+        # Concatenate the unconditional and text embeddings into a single batch to avoid doing two forward passes for classifier free guidance
+        text_embeddings = torch.cat([uncond_embeddings, text_embeddings]).to(dtype=torch.float16)
+
+        return text_embeddings
+
+    def __denoise_latent(
+        self, latents, text_embeddings, timesteps=None, step_offset=0, mask=None, masked_image_latents=None
+    ):
+        if not isinstance(timesteps, torch.Tensor):
+            timesteps = self.scheduler.timesteps
+        for step_index, timestep in enumerate(timesteps):
+            # Expand the latents if we are doing classifier free guidance
+            latent_model_input = torch.cat([latents] * 2)
+            latent_model_input = self.scheduler.scale_model_input(latent_model_input, timestep)
+            if isinstance(mask, torch.Tensor):
+                latent_model_input = torch.cat([latent_model_input, mask, masked_image_latents], dim=1)
+
+            # Predict the noise residual
+            timestep_float = timestep.float() if timestep.dtype != torch.float32 else timestep
+
+            sample_inp = device_view(latent_model_input)
+            timestep_inp = device_view(timestep_float)
+            embeddings_inp = device_view(text_embeddings)
+            noise_pred = runEngine(
+                self.engine["unet"],
+                {"sample": sample_inp, "timestep": timestep_inp, "encoder_hidden_states": embeddings_inp},
+                self.stream,
+            )["latent"]
+
+            # Perform guidance
+            noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
+            noise_pred = noise_pred_uncond + self.guidance_scale * (noise_pred_text - noise_pred_uncond)
+
+            latents = self.scheduler.step(noise_pred, timestep, latents).prev_sample
+
+        latents = 1.0 / 0.18215 * latents
+        return latents
+
+    def __decode_latent(self, latents):
+        images = runEngine(self.engine["vae"], {"latent": device_view(latents)}, self.stream)["images"]
+        images = (images / 2 + 0.5).clamp(0, 1)
+        return images.cpu().permute(0, 2, 3, 1).float().numpy()
+
+    def __loadResources(self, image_height, image_width, batch_size):
+        self.stream = cuda.Stream()
+
+        # Allocate buffers for TensorRT engine bindings
+        for model_name, obj in self.models.items():
+            self.engine[model_name].allocate_buffers(
+                shape_dict=obj.get_shape_dict(batch_size, image_height, image_width), device=self.torch_device
+            )
+
+    @torch.no_grad()
+    def __call__(
+        self,
+        prompt: Union[str, List[str]] = None,
+        num_inference_steps: int = 50,
+        guidance_scale: float = 7.5,
+        negative_prompt: Optional[Union[str, List[str]]] = None,
+        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
+    ):
+        r"""
+        Function invoked when calling the pipeline for generation.
+
+        Args:
+            prompt (`str` or `List[str]`, *optional*):
+                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
+                instead.
+            num_inference_steps (`int`, *optional*, defaults to 50):
+                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.
+            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. If not defined, one has to pass `negative_prompt_embeds`. instead.
+                Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`).
+            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.
+
+        """
+        self.generator = generator
+        self.denoising_steps = num_inference_steps
+        self.guidance_scale = guidance_scale
+
+        # Pre-compute latent input scales and linear multistep coefficients
+        self.scheduler.set_timesteps(self.denoising_steps, device=self.torch_device)
+
+        # Define call parameters
+        if prompt is not None and isinstance(prompt, str):
+            batch_size = 1
+            prompt = [prompt]
+        elif prompt is not None and isinstance(prompt, list):
+            batch_size = len(prompt)
+        else:
+            raise ValueError(f"Expected prompt to be of type list or str but got {type(prompt)}")
+
+        if negative_prompt is None:
+            negative_prompt = [""] * batch_size
+
+        if negative_prompt is not None and isinstance(negative_prompt, str):
+            negative_prompt = [negative_prompt]
+
+        assert len(prompt) == len(negative_prompt)
+
+        if batch_size > self.max_batch_size:
+            raise ValueError(
+                f"Batch size {len(prompt)} is larger than allowed {self.max_batch_size}. If dynamic shape is used, then maximum batch size is 4"
+            )
+
+        # load resources
+        self.__loadResources(self.image_height, self.image_width, batch_size)
+
+        with torch.inference_mode(), torch.autocast("cuda"), trt.Runtime(TRT_LOGGER):
+            # CLIP text encoder
+            text_embeddings = self.__encode_prompt(prompt, negative_prompt)
+
+            # Pre-initialize latents
+            num_channels_latents = self.unet.in_channels
+            latents = self.prepare_latents(
+                batch_size,
+                num_channels_latents,
+                self.image_height,
+                self.image_width,
+                torch.float32,
+                self.torch_device,
+                generator,
+            )
+
+            # UNet denoiser
+            latents = self.__denoise_latent(latents, text_embeddings)
+
+            # VAE decode latent
+            images = self.__decode_latent(latents)
+
+        images, has_nsfw_concept = self.run_safety_checker(images, self.torch_device, text_embeddings.dtype)
+        images = self.numpy_to_pil(images)
+        return StableDiffusionPipelineOutput(images=images, nsfw_content_detected=has_nsfw_concept)
@@ -46,7 +46,7 @@ class StableUnCLIPPipeline(DiffusionPipeline):
    ):
        super().__init__()

-        decoder_pipe_kwargs = dict(image_encoder=None) if decoder_pipe_kwargs is None else decoder_pipe_kwargs
+        decoder_pipe_kwargs = {"image_encoder": None} if decoder_pipe_kwargs is None else decoder_pipe_kwargs

        decoder_pipe_kwargs["torch_dtype"] = decoder_pipe_kwargs.get("torch_dtype", None) or prior.dtype

@@ -3,7 +3,7 @@ from typing import Callable, List, Optional, Union
 import PIL
 import torch
 from transformers import (
-    CLIPFeatureExtractor,
+    CLIPImageProcessor,
    CLIPSegForImageSegmentation,
    CLIPSegProcessor,
    CLIPTextModel,
@@ -52,7 +52,7 @@ class TextInpainting(DiffusionPipeline):
        safety_checker ([`StableDiffusionSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offensive or harmful.
            Please, refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for details.
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """

@@ -66,7 +66,7 @@ class TextInpainting(DiffusionPipeline):
        unet: UNet2DConditionModel,
        scheduler: Union[DDIMScheduler, PNDMScheduler, LMSDiscreteScheduler],
        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
    ):
        super().__init__()

@@ -5,7 +5,7 @@ import PIL
 import torch
 from torch.nn import functional as F
 from transformers import (
-    CLIPFeatureExtractor,
+    CLIPImageProcessor,
    CLIPTextModelWithProjection,
    CLIPTokenizer,
    CLIPVisionModelWithProjection,
@@ -50,7 +50,7 @@ class UnCLIPImageInterpolationPipeline(DiffusionPipeline):
        tokenizer (`CLIPTokenizer`):
            Tokenizer of class
            [CLIPTokenizer](https://huggingface.co/docs/transformers/v4.21.0/en/model_doc/clip#transformers.CLIPTokenizer).
-        feature_extractor ([`CLIPFeatureExtractor`]):
+        feature_extractor ([`CLIPImageProcessor`]):
            Model that extracts features from generated images to be used as inputs for the `image_encoder`.
        image_encoder ([`CLIPVisionModelWithProjection`]):
            Frozen CLIP image-encoder. unCLIP Image Variation uses the vision portion of
@@ -75,7 +75,7 @@ class UnCLIPImageInterpolationPipeline(DiffusionPipeline):
    text_proj: UnCLIPTextProjModel
    text_encoder: CLIPTextModelWithProjection
    tokenizer: CLIPTokenizer
-    feature_extractor: CLIPFeatureExtractor
+    feature_extractor: CLIPImageProcessor
    image_encoder: CLIPVisionModelWithProjection
    super_res_first: UNet2DModel
    super_res_last: UNet2DModel
@@ -90,7 +90,7 @@ class UnCLIPImageInterpolationPipeline(DiffusionPipeline):
        text_encoder: CLIPTextModelWithProjection,
        tokenizer: CLIPTokenizer,
        text_proj: UnCLIPTextProjModel,
-        feature_extractor: CLIPFeatureExtractor,
+        feature_extractor: CLIPImageProcessor,
        image_encoder: CLIPVisionModelWithProjection,
        super_res_first: UNet2DModel,
        super_res_last: UNet2DModel,
@@ -270,7 +270,7 @@ class UnCLIPImageInterpolationPipeline(DiffusionPipeline):
                The images to use for the image interpolation. Only accepts a list of two PIL Images or If you provide a tensor, it needs to comply with the
                configuration of
                [this](https://huggingface.co/fusing/karlo-image-variations-diffusers/blob/main/feature_extractor/preprocessor_config.json)
-                `CLIPFeatureExtractor` while still having a shape of two in the 0th dimension. Can be left to `None` only when `image_embeddings` are passed.
+                `CLIPImageProcessor` while still having a shape of two in the 0th dimension. Can be left to `None` only when `image_embeddings` are passed.
            steps (`int`, *optional*, defaults to 5):
                The number of interpolation images to generate.
            decoder_num_inference_steps (`int`, *optional*, defaults to 25):
@@ -372,9 +372,9 @@ class UnCLIPImageInterpolationPipeline(DiffusionPipeline):
        self.decoder_scheduler.set_timesteps(decoder_num_inference_steps, device=device)
        decoder_timesteps_tensor = self.decoder_scheduler.timesteps

-        num_channels_latents = self.decoder.in_channels
-        height = self.decoder.sample_size
-        width = self.decoder.sample_size
+        num_channels_latents = self.decoder.config.in_channels
+        height = self.decoder.config.sample_size
+        width = self.decoder.config.sample_size

        decoder_latents = self.prepare_latents(
            (batch_size, num_channels_latents, height, width),
@@ -425,9 +425,9 @@ class UnCLIPImageInterpolationPipeline(DiffusionPipeline):
        self.super_res_scheduler.set_timesteps(super_res_num_inference_steps, device=device)
        super_res_timesteps_tensor = self.super_res_scheduler.timesteps

-        channels = self.super_res_first.in_channels // 2
-        height = self.super_res_first.sample_size
-        width = self.super_res_first.sample_size
+        channels = self.super_res_first.config.in_channels // 2
+        height = self.super_res_first.config.sample_size
+        width = self.super_res_first.config.sample_size

        super_res_latents = self.prepare_latents(
            (batch_size, channels, height, width),
--- a/Show More
+++ b/Show More