@@ -77,7 +77,7 @@ Please refer to the [How to use Stable Diffusion in Apple Silicon](https://huggi
## Quickstart
Generating outputs is super easy with 🤗 Diffusers. To generate an image from text, use the `from_pretrained` method to load any pretrained diffusion model (browse the [Hub](https://huggingface.co/models?library=diffusers&sort=downloads) for 15000+ checkpoints):
Generating outputs is super easy with 🤗 Diffusers. To generate an image from text, use the `from_pretrained` method to load any pretrained diffusion model (browse the [Hub](https://huggingface.co/models?library=diffusers&sort=downloads) for 16000+ checkpoints):
```python
fromdiffusersimportDiffusionPipeline
...
...
@@ -219,7 +219,7 @@ Also, say 👋 in our public Discord channel <a href="https://discord.gg/G7tWnz9
Diffusers examples are a collection of scripts to demonstrate how to effectively use the `diffusers` library
for a variety of use cases involving training or fine-tuning.
**Note**: If you are looking for **official** examples on how to use `diffusers` for inference,
please have a look at [src/diffusers/pipelines](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines).
**Note**: If you are looking for **official** examples on how to use `diffusers` for inference, please have a look at [src/diffusers/pipelines](https://github.com/huggingface/diffusers/tree/main/src/diffusers/pipelines).
Our examples aspire to be **self-contained**, **easy-to-tweak**, **beginner-friendly** and for **one-purpose-only**.
More specifically, this means:
...
...
@@ -27,11 +26,10 @@ More specifically, this means:
-**Self-contained**: An example script shall only depend on "pip-install-able" Python packages that can be found in a `requirements.txt` file. Example scripts shall **not** depend on any local files. This means that one can simply download an example script, *e.g.*[train_unconditional.py](https://github.com/huggingface/diffusers/blob/main/examples/unconditional_image_generation/train_unconditional.py), install the required dependencies, *e.g.*[requirements.txt](https://github.com/huggingface/diffusers/blob/main/examples/unconditional_image_generation/requirements.txt) and execute the example script.
-**Easy-to-tweak**: While we strive to present as many use cases as possible, the example scripts are just that - examples. It is expected that they won't work out-of-the box on your specific problem and that you will be required to change a few lines of code to adapt them to your needs. To help you with that, most of the examples fully expose the preprocessing of the data and the training loop to allow you to tweak and edit them as required.
-**Beginner-friendly**: We do not aim for providing state-of-the-art training scripts for the newest models, but rather examples that can be used as a way to better understand diffusion models and how to use them with the `diffusers` library. We often purposefully leave out certain state-of-the-art methods if we consider them too complex for beginners.
-**One-purpose-only**: Examples should show one task and one task only. Even if a task is from a modeling
point of view very similar, *e.g.* image super-resolution and image modification tend to use the same model and training method, we want examples to showcase only one task to keep them as readable and easy-to-understand as possible.
-**One-purpose-only**: Examples should show one task and one task only. Even if a task is from a modeling point of view very similar, *e.g.* image super-resolution and image modification tend to use the same model and training method, we want examples to showcase only one task to keep them as readable and easy-to-understand as possible.
We provide **official** examples that cover the most popular tasks of diffusion models.
*Official* examples are **actively** maintained by the `diffusers` maintainers and we try to rigorously follow our example philosophy as defined above.
*Official* examples are **actively** maintained by the `diffusers` maintainers and we try to rigorously follow our example philosophy as defined above.
If you feel like another important example should exist, we are more than happy to welcome a [Feature Request](https://github.com/huggingface/diffusers/issues/new?assignees=&labels=&template=feature_request.md&title=) or directly a [Pull Request](https://github.com/huggingface/diffusers/compare) from you!
Training examples show how to pretrain or fine-tune diffusion models for a variety of tasks. Currently we support:
...
...
@@ -39,7 +37,7 @@ Training examples show how to pretrain or fine-tune diffusion models for a varie
| LLM-grounded Diffusion (LMD+) | LMD greatly improves the prompt following ability of text-to-image generation models by introducing an LLM as a front-end prompt parser and layout planner. [Project page.](https://llm-grounded-diffusion.github.io/)[See our full codebase (also with diffusers).](https://github.com/TonyLianLong/LLM-groundedDiffusion) | [LLM-grounded Diffusion (LMD+)](#llm-grounded-diffusion) | [Huggingface Demo](https://huggingface.co/spaces/longlian/llm-grounded-diffusion)[](https://colab.research.google.com/drive/1SXzMSeAB-LJYISb2yrUOdypLz4OYWUKj) | [Long (Tony) Lian](https://tonylian.com/) |
| CLIP Guided Stable Diffusion | Doing CLIP guidance for text to image generation with Stable Diffusion | [CLIP Guided Stable Diffusion](#clip-guided-stable-diffusion) | [](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/CLIP_Guided_Stable_diffusion_with_diffusers.ipynb) | [Suraj Patil](https://github.com/patil-suraj/) |
| LLM-grounded Diffusion (LMD+) | LMD greatly improves the prompt following ability of text-to-image generation models by introducing an LLM as a front-end prompt parser and layout planner. [Project page.](https://llm-grounded-diffusion.github.io/)[See our full codebase (also with diffusers).](https://github.com/TonyLianLong/LLM-groundedDiffusion) | [LLM-grounded Diffusion (LMD+)](#llm-grounded-diffusion) | [Huggingface Demo](https://huggingface.co/spaces/longlian/llm-grounded-diffusion)[](https://colab.research.google.com/drive/1SXzMSeAB-LJYISb2yrUOdypLz4OYWUKj) | [Long (Tony) Lian](https://tonylian.com/) |
| CLIP Guided Stable Diffusion | Doing CLIP guidance for text to image generation with Stable Diffusion | [CLIP Guided Stable Diffusion](#clip-guided-stable-diffusion) | [](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/CLIP_Guided_Stable_diffusion_with_diffusers.ipynb) | [Suraj Patil](https://github.com/patil-suraj/) |
| One Step U-Net (Dummy) | Example showcasing of how to use Community Pipelines (see https://github.com/huggingface/diffusers/issues/841) | [One Step U-Net](#one-step-unet) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) |
| Stable Diffusion Interpolation | Interpolate the latent space of Stable Diffusion between different prompts/seeds | [Stable Diffusion Interpolation](#stable-diffusion-interpolation) | - | [Nate Raw](https://github.com/nateraw/) |
| Stable Diffusion Mega | **One** Stable Diffusion Pipeline with all functionalities of [Text2Image](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py), [Image2Image](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py) and [Inpainting](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py) | [Stable Diffusion Mega](#stable-diffusion-mega) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) |
| Long Prompt Weighting Stable Diffusion | **One** Stable Diffusion Pipeline without tokens length limit, and support parsing weighting in prompt. | [Long Prompt Weighting Stable Diffusion](#long-prompt-weighting-stable-diffusion) | - | [SkyTNT](https://github.com/SkyTNT) |
| Speech to Image | Using automatic-speech-recognition to transcribe text and Stable Diffusion to generate images | [Speech to Image](#speech-to-image) | - | [Mikail Duzenli](https://github.com/MikailINTech)
| Speech to Image | Using automatic-speech-recognition to transcribe text and Stable Diffusion to generate images | [Speech to Image](#speech-to-image) | - | [Mikail Duzenli](https://github.com/MikailINTech)
| Wild Card Stable Diffusion | Stable Diffusion Pipeline that supports prompts that contain wildcard terms (indicated by surrounding double underscores), with values instantiated randomly from a corresponding txt file or a dictionary of possible values | [Wildcard Stable Diffusion](#wildcard-stable-diffusion) | - | [Shyam Sudhakaran](https://github.com/shyamsn97) |
| [Composable Stable Diffusion](https://energy-based-model.github.io/Compositional-Visual-Generation-with-Composable-Diffusion-Models/) | Stable Diffusion Pipeline that supports prompts that contain "|" in prompts (as an AND condition) and weights (separated by "|" as well) to positively / negatively weight prompts. | [Composable Stable Diffusion](#composable-stable-diffusion) | - | [Mark Rich](https://github.com/MarkRich) |
| Seed Resizing Stable Diffusion | Stable Diffusion Pipeline that supports resizing an image and retaining the concepts of the 512 by 512 generation. | [Seed Resizing](#seed-resizing) | - | [Mark Rich](https://github.com/MarkRich) |
...
...
@@ -24,27 +24,27 @@ If a community doesn't work as expected, please open an issue and ping the autho
| Text Based Inpainting Stable Diffusion | Stable Diffusion Inpainting Pipeline that enables passing a text prompt to generate the mask for inpainting | [Text Based Inpainting Stable Diffusion](#image-to-image-inpainting-stable-diffusion) | - | [Dhruv Karan](https://github.com/unography) |
| Bit Diffusion | Diffusion on discrete data | [Bit Diffusion](#bit-diffusion) | - | [Stuti R.](https://github.com/kingstut) |
| K-Diffusion Stable Diffusion | Run Stable Diffusion with any of [K-Diffusion's samplers](https://github.com/crowsonkb/k-diffusion/blob/master/k_diffusion/sampling.py) | [Stable Diffusion with K Diffusion](#stable-diffusion-with-k-diffusion) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) |
| Checkpoint Merger Pipeline | Diffusion Pipeline that enables merging of saved model checkpoints | [Checkpoint Merger Pipeline](#checkpoint-merger-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) |
| Checkpoint Merger Pipeline | Diffusion Pipeline that enables merging of saved model checkpoints | [Checkpoint Merger Pipeline](#checkpoint-merger-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) |
Stable Diffusion v1.1-1.4 Comparison | Run all 4 model checkpoints for Stable Diffusion and compare their results together | [Stable Diffusion Comparison](#stable-diffusion-comparisons) | - | [Suvaditya Mukherjee](https://github.com/suvadityamuk) |
MagicMix | Diffusion Pipeline for semantic mixing of an image and a text prompt | [MagicMix](#magic-mix) | - | [Partho Das](https://github.com/daspartho) |
| Stable UnCLIP | Diffusion Pipeline for combining prior model (generate clip image embedding from text, UnCLIPPipeline `"kakaobrain/karlo-v1-alpha"`) and decoder pipeline (decode clip image embedding to image, StableDiffusionImageVariationPipeline `"lambdalabs/sd-image-variations-diffusers"` ). | [Stable UnCLIP](#stable-unclip) | - | [Ray Wang](https://wrong.wang) |
| 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/) |
| 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) |
| Zero1to3 Pipeline | Implementation of [Zero-1-to-3: Zero-shot One Image to 3D Object](https://arxiv.org/abs/2303.11328) | [Zero1to3 Pipeline](#Zero1to3-pipeline) | - | [Xin Kong](https://github.com/kxhit) |
Stable Diffusion XL Long Weighted Prompt Pipeline | A pipeline support unlimited length of prompt and negative prompt, use A1111 style of prompt weighting | [Stable Diffusion XL Long Weighted Prompt Pipeline](#stable-diffusion-xl-long-weighted-prompt-pipeline) | - | [Andrew Zhu](https://xhinker.medium.com/) |
FABRIC - Stable Diffusion with feedback Pipeline | pipeline supports feedback from liked and disliked images | [Stable Diffusion Fabric Pipeline](#stable-diffusion-fabric-pipeline) | - | [Shauray Singh](https://shauray8.github.io/about_shauray/) |
sketch inpaint - Inpainting with non-inpaint Stable Diffusion | sketch inpaint much like in automatic1111 | [Masked Im2Im Stable Diffusion Pipeline](#stable-diffusion-masked-im2im) | - | [Anatoly Belikov](https://github.com/noskill) |
prompt-to-prompt | change parts of a prompt and retain image structure (see [paper page](https://prompt-to-prompt.github.io/)) | [Prompt2Prompt Pipeline](#prompt2prompt-pipeline) | - | [Umer H. Adil](https://twitter.com/UmerHAdil) |
Stable Diffusion XL Long Weighted Prompt Pipeline | A pipeline support unlimited length of prompt and negative prompt, use A1111 style of prompt weighting | [Stable Diffusion XL Long Weighted Prompt Pipeline](#stable-diffusion-xl-long-weighted-prompt-pipeline) | - | [Andrew Zhu](https://xhinker.medium.com/) |
FABRIC - Stable Diffusion with feedback Pipeline | pipeline supports feedback from liked and disliked images | [Stable Diffusion Fabric Pipeline](#stable-diffusion-fabric-pipeline) | - | [Shauray Singh](https://shauray8.github.io/about_shauray/) |
sketch inpaint - Inpainting with non-inpaint Stable Diffusion | sketch inpaint much like in automatic1111 | [Masked Im2Im Stable Diffusion Pipeline](#stable-diffusion-masked-im2im) | - | [Anatoly Belikov](https://github.com/noskill) |
prompt-to-prompt | change parts of a prompt and retain image structure (see [paper page](https://prompt-to-prompt.github.io/)) | [Prompt2Prompt Pipeline](#prompt2prompt-pipeline) | - | [Umer H. Adil](https://twitter.com/UmerHAdil) |
The output of the `walk(...)` function returns a list of images saved under the folder as defined in `output_dir`. You can use these images to create videos of stable diffusion.
The output of the `walk(...)` function returns a list of images saved under the folder as defined in `output_dir`. You can use these images to create videos of stable diffusion.
> **Please have a look at https://github.com/nateraw/stable-diffusion-videos for more in-detail information on how to create videos using stable diffusion as well as more feature-complete functionality.**
Allows you to edit an image using stable diffusion.
Allows you to edit an image using stable diffusion.
```python
importrequests
...
...
@@ -539,7 +539,7 @@ image = res.images[0]
image.save('./imagic/imagic_image_alpha_2.png')
```
### Seed Resizing
### Seed Resizing
Test seed resizing. Originally generate an image in 512 by 512, then generate image with same seed at 512 by 592 using seed resizing. Finally, generate 512 by 592 using original stable diffusion pipeline.
Based https://arxiv.org/abs/2208.04202, this is used for diffusion on discrete data - eg, discreate image data, DNA sequence data. An unconditional discreate image can be generated like this:
### Bit Diffusion
Based https://arxiv.org/abs/2208.04202, this is used for diffusion on discrete data - eg, discreate image data, DNA sequence data. An unconditional discreate image can be generated like this:
```python
fromdiffusersimportDiffusionPipeline
...
...
@@ -837,8 +837,8 @@ Usage:-
```python
fromdiffusersimportDiffusionPipeline
#Return a CheckpointMergerPipeline class that allows you to merge checkpoints.
#The checkpoint passed here is ignored. But still pass one of the checkpoints you plan to
#Return a CheckpointMergerPipeline class that allows you to merge checkpoints.
#The checkpoint passed here is ignored. But still pass one of the checkpoints you plan to
This Diffusion Pipeline takes two prompts and interpolates between the two input prompts using spherical interpolation ( slerp ). The input prompts are converted to text embeddings by the pipeline's text_encoder and the interpolation is done on the resulting text_embeddings over the number of steps specified. Defaults to 5 steps.
This Diffusion Pipeline takes two prompts and interpolates between the two input prompts using spherical interpolation ( slerp ). The input prompts are converted to text embeddings by the pipeline's text_encoder and the interpolation is done on the resulting text_embeddings over the number of steps specified. Defaults to 5 steps.
```python
importtorch
...
...
@@ -1086,7 +1086,7 @@ The resulting images in order:-
### UnCLIP Image Interpolation Pipeline
This Diffusion Pipeline takes two images or an image_embeddings tensor of size 2 and interpolates between their embeddings using spherical interpolation ( slerp ). The input images/image_embeddings are converted to image embeddings by the pipeline's image_encoder and the interpolation is done on the resulting image_embeddings over the number of steps specified. Defaults to 5 steps.
This Diffusion Pipeline takes two images or an image_embeddings tensor of size 2 and interpolates between their embeddings using spherical interpolation ( slerp ). The input images/image_embeddings are converted to image embeddings by the pipeline's image_encoder and the interpolation is done on the resulting image_embeddings over the number of steps specified. Defaults to 5 steps.
```python
importtorch
...
...
@@ -1127,8 +1127,8 @@ The resulting images in order:-
#### **Research question: What visual concepts do the diffusion models learn from each noise level during training?**
The [P2 weighting (CVPR 2022)](https://arxiv.org/abs/2204.00227) paper proposed an approach to answer the above question, which is their second contribution.
#### **Research question: What visual concepts do the diffusion models learn from each noise level during training?**
The [P2 weighting (CVPR 2022)](https://arxiv.org/abs/2204.00227) paper proposed an approach to answer the above question, which is their second contribution.
The approach consists of the following steps:
1. The input is an image x0.
...
...
@@ -1170,7 +1170,7 @@ Here is the result of this pipeline (which is DDIM) on CelebA-HQ dataset.
### CLIP Guided Img2Img Stable Diffusion
CLIP guided Img2Img stable diffusion can help to generate more realistic images with an initial image
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.
...
...
@@ -1322,8 +1322,8 @@ target_prompt = "A golden retriever"
### Stable Diffusion XL Long Weighted Prompt Pipeline
This SDXL pipeline support unlimited length prompt and negative prompt, compatible with A1111 prompt weighted style.
This SDXL pipeline support unlimited length prompt and negative prompt, compatible with A1111 prompt weighted style.
You can provide both `prompt` and `prompt_2`. if only one prompt is provided, `prompt_2` will be a copy of the provided `prompt`. Here is a sample code to use this pipeline.
You can provide both `prompt` and `prompt_2`. if only one prompt is provided, `prompt_2` will be a copy of the provided `prompt`. Here is a sample code to use this pipeline.
[LDM3D-VR](https://arxiv.org/pdf/2311.03226.pdf) is an extended version of LDM3D.
[LDM3D-VR](https://arxiv.org/pdf/2311.03226.pdf) is an extended version of LDM3D.
The abstract from the paper is:
*Latent diffusion models have proven to be state-of-the-art in the creation and manipulation of visual outputs. However, as far as we know, the generation of depth maps jointly with RGB is still limited. We introduce LDM3D-VR, a suite of diffusion models targeting virtual reality development that includes LDM3D-pano and LDM3D-SR. These models enable the generation of panoramic RGBD based on textual prompts and the upscaling of low-resolution inputs to high-resolution RGBD, respectively. Our models are fine-tuned from existing pretrained models on datasets containing panoramic/high-resolution RGB images, depth maps and captions. Both models are evaluated in comparison to existing related methods*
This pipelines combines both ControlNet and T2IAdapter into a single pipeline, where the forward pass is executed once.
It receives `control_image` and `adapter_image`, as well as `controlnet_conditioning_scale` and `adapter_conditioning_scale`, for the ControlNet and Adapter modules, respectively. Whenever `adapter_conditioning_scale = 0` or `controlnet_conditioning_scale = 0`, it will act as a full ControlNet module or as a full T2IAdapter module, respectively.
This pipelines combines both ControlNet and T2IAdapter into a single pipeline, where the forward pass is executed once.
It receives `control_image` and `adapter_image`, as well as `controlnet_conditioning_scale` and `adapter_conditioning_scale`, for the ControlNet and Adapter modules, respectively. Whenever `adapter_conditioning_scale = 0` or `controlnet_conditioning_scale = 0`, it will act as a full ControlNet module or as a full T2IAdapter module, respectively.
In the Cols, Rows mode, you can split the screen vertically and horizontally and assign prompts to each region. The split ratio can be specified by 'div', and you can set the division ratio like '3;3;2' or '0.1;0.5'. Furthermore, as will be described later, you can also subdivide the split Cols, Rows to specify more complex regions.
In this image, the image is divided into three parts, and a separate prompt is applied to each. The prompts are divided by 'BREAK', and each is applied to the respective region.
In this image, the image is divided into three parts, and a separate prompt is applied to each. The prompts are divided by 'BREAK', and each is applied to the respective region.
@@ -2644,7 +2644,7 @@ If only one input is given for multiple regions, they are all assumed to be the
The difference is that in Prompt, duplicate regions are added, whereas in Prompt-EX, duplicate regions are overwritten sequentially. Since they are processed in order, setting a TARGET with a large regions first makes it easier for the effect of small regions to remain unmuffled.
### Accuracy
In the case of a 512 x 512 image, Attention mode reduces the size of the region to about 8 x 8 pixels deep in the U-Net, so that small regions get mixed up; Latent mode calculates 64*64, so that the region is exact.
In the case of a 512 x 512 image, Attention mode reduces the size of the region to about 8 x 8 pixels deep in the U-Net, so that small regions get mixed up; Latent mode calculates 64*64, so that the region is exact.
```
girl hair twintail frills,ribbons, dress, face BREAK
girl, ,face
...
...
@@ -2675,13 +2675,13 @@ Negative prompts are equally effective across all regions, but it is possible to
To activate Regional Prompter, it is necessary to enter settings in rp_args. The items that can be set are as follows. rp_args is a dictionary type.
### Input Parameters
Parameters are specified through the `rp_arg`(dictionary type).
Parameters are specified through the `rp_arg`(dictionary type).
```
rp_args = {
"mode":"rows",
"div": "1;1;1"
}
}
pipe(prompt =prompt, rp_args = rp_args)
```
...
...
@@ -2760,7 +2760,7 @@ The Pipeline supports `compel` syntax. Input prompts using the `compel` structur
def get_kernel(self):
return self.k
self.kernel_size = kernel_size
self.conv = Blurkernel(blur_type='gaussian',
kernel_size=kernel_size,
...
...
@@ -2835,7 +2835,7 @@ The Pipeline supports `compel` syntax. Input prompts using the `compel` structur
This folder contains various research projects using 🧨 Diffusers.
They are not really maintained by the core maintainers of this library and often require a specific version of Diffusers that is indicated in the requirements file of each folder.
This folder contains various research projects using 🧨 Diffusers.
They are not really maintained by the core maintainers of this library and often require a specific version of Diffusers that is indicated in the requirements file of each folder.
Updating them to the most recent version of the library will require some work.
**IMPORTART: This research project is deprecated. Multi Token Textual Inversion is now supported natively in [the officail textual inversion example](https://github.com/huggingface/diffusers/tree/main/examples/textual_inversion#running-locally-with-pytorch).**
**IMPORTART: This research project is deprecated. Multi Token Textual Inversion is now supported natively in [the official textual inversion example](https://github.com/huggingface/diffusers/tree/main/examples/textual_inversion#running-locally-with-pytorch).**
The author of this project is [Isamu Isozaki](https://github.com/isamu-isozaki) - please make sure to tag the author for issue and PRs as well as @patrickvonplaten.
...
...
@@ -17,9 +17,9 @@ Feel free to add these options to your training! In practice num_vec_per_token a
[Textual inversion](https://arxiv.org/abs/2208.01618) is a method to personalize text2image models like stable diffusion on your own images using just 3-5 examples.
The `textual_inversion.py` script shows how to implement the training procedure and adapt it for stable diffusion.
## Running on Colab
## Running on Colab
Colab for training
Colab for training
[](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_textual_inversion_training.ipynb)
Colab for inference
...
...
@@ -53,7 +53,7 @@ accelerate config
### Cat toy example
You need to accept the model license before downloading or using the weights. In this example we'll use model version `v1-5`, so you'll need to visit [its card](https://huggingface.co/runwayml/stable-diffusion-v1-5), read the license and tick the checkbox if you agree.
You need to accept the model license before downloading or using the weights. In this example we'll use model version `v1-5`, so you'll need to visit [its card](https://huggingface.co/runwayml/stable-diffusion-v1-5), read the license and tick the checkbox if you agree.
You have to be a registered user in 🤗 Hugging Face Hub, and you'll also need to use an access token for the code to work. For more information on access tokens, please refer to [this section of the documentation](https://huggingface.co/docs/hub/security-tokens).
...
...
@@ -63,7 +63,7 @@ Run the following command to authenticate your token
huggingface-cli login
```
If you have already cloned the repo, then you won't need to go through these steps.
If you have already cloned the repo, then you won't need to go through these steps.
**This research project is not actively maintained by the diffusers team. For any questions or comments, please contact Prathik Rao (prathikr), Sunghoon Choi (hanbitmyths), Ashwini Khade (askhade), or Peng Wang (pengwa) on github with any questions.**
This aims to provide diffusers examples with ONNXRuntime optimizations for training/fine-tuning unconditional image generation, text to image, and textual inversion. Please see individual directories for more details on how to run each task using ONNXRuntime.
\ No newline at end of file
This aims to provide diffusers examples with ONNXRuntime optimizations for training/fine-tuning unconditional image generation, text to image, and textual inversion. Please see individual directories for more details on how to run each task using ONNXRuntime.
You need to accept the model license before downloading or using the weights. In this example we'll use model version `v1-4`, so you'll need to visit [its card](https://huggingface.co/CompVis/stable-diffusion-v1-4), read the license and tick the checkbox if you agree.
You need to accept the model license before downloading or using the weights. In this example we'll use model version `v1-4`, so you'll need to visit [its card](https://huggingface.co/CompVis/stable-diffusion-v1-4), read the license and tick the checkbox if you agree.
You have to be a registered user in 🤗 Hugging Face Hub, and you'll also need to use an access token for the code to work. For more information on access tokens, please refer to [this section of the documentation](https://huggingface.co/docs/hub/security-tokens).
[Textual inversion](https://arxiv.org/abs/2208.01618) is a method to personalize text2image models like stable diffusion on your own images using just 3-5 examples.
The `textual_inversion.py` script shows how to implement the training procedure and adapt it for stable diffusion.
## Running on Colab
## Running on Colab
Colab for training
Colab for training
[](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_textual_inversion_training.ipynb)
Colab for inference
...
...
@@ -39,7 +39,7 @@ accelerate config
### Cat toy example
You need to accept the model license before downloading or using the weights. In this example we'll use model version `v1-5`, so you'll need to visit [its card](https://huggingface.co/runwayml/stable-diffusion-v1-5), read the license and tick the checkbox if you agree.
You need to accept the model license before downloading or using the weights. In this example we'll use model version `v1-5`, so you'll need to visit [its card](https://huggingface.co/runwayml/stable-diffusion-v1-5), read the license and tick the checkbox if you agree.
You have to be a registered user in 🤗 Hugging Face Hub, and you'll also need to use an access token for the code to work. For more information on access tokens, please refer to [this section of the documentation](https://huggingface.co/docs/hub/security-tokens).
...
...
@@ -49,7 +49,7 @@ Run the following command to authenticate your token
huggingface-cli login
```
If you have already cloned the repo, then you won't need to go through these steps.
If you have already cloned the repo, then you won't need to go through these steps.
This section adjusts the data types of the model parameters.
We convert all parameters to `bfloat16` to speed-up the computation with model weights.
**Note** that the scheduler parameters are **not** converted to `blfoat16` as the loss
We convert all parameters to `bfloat16` to speed-up the computation with model weights.
**Note** that the scheduler parameters are **not** converted to `blfoat16` as the loss
in precision is degrading the pipeline's performance too significantly.
**3. Define Inputs to Pipeline**
...
...
@@ -146,12 +146,12 @@ For this we will be using a JAX feature called [Ahead of Time](https://jax.readt
In [sdxl_single_aot.py](./sdxl_single_aot.py) we give a simple example of how to write our own parallelization logic for text-to-image generation pipeline in JAX using [StabilityAI's Stable Diffusion XL](stabilityai/stable-diffusion-xl-base-1.0)
We add a `aot_compile` function that compiles the `pipeline._generate` function
We add a `aot_compile` function that compiles the `pipeline._generate` function
telling JAX which input arguments are static, that is, arguments that
are known at compile time and won't change. In our case, it is num_inference_steps,
are known at compile time and won't change. In our case, it is num_inference_steps,
height, width and return_latents.
Once the function is compiled, these parameters are omitted from future calls and
Once the function is compiled, these parameters are omitted from future calls and
cannot be changed without modifying the code and recompiling.
```python
...
...
@@ -205,9 +205,9 @@ def generate(
g = jnp.array([guidance_scale] * prompt_ids.shape[0], dtype=jnp.float32)
g = g[:, None]
images = p_generate(
prompt_ids,
p_params,
rng,
prompt_ids,
p_params,
rng,
g,
None,
neg_prompt_ids)
...
...
@@ -220,7 +220,7 @@ def generate(
The first forward pass after AOT compilation still takes a while longer than
subsequent passes, this is because on the first pass, JAX uses Python dispatch, which
Fills the C++ dispatch cache.
When using jit, this extra step is done automatically, but when using AOT compilation,
When using jit, this extra step is done automatically, but when using AOT compilation,
it doesn't happen until the function call is made.
To better track our training experiments, we're using the following flags in the command above:
*`report_to="wandb` will ensure the training runs are tracked on Weights and Biases. To use it, be sure to install `wandb` with `pip install wandb`.
*`validation_image`, `validation_prompt`, and `validation_steps` to allow the script to do a few validation inference runs. This allows us to qualitatively check if the training is progressing as expected.
*`validation_image`, `validation_prompt`, and `validation_steps` to allow the script to do a few validation inference runs. This allows us to qualitatively check if the training is progressing as expected.
Our experiments were conducted on a single 40GB A100 GPU.
@@ -36,7 +36,7 @@ Note also that we use PEFT library as backend for LoRA training, make sure to ha
### Pokemon example
You need to accept the model license before downloading or using the weights. In this example we'll use model version `v1-4`, so you'll need to visit [its card](https://huggingface.co/CompVis/stable-diffusion-v1-4), read the license and tick the checkbox if you agree.
You need to accept the model license before downloading or using the weights. In this example we'll use model version `v1-4`, so you'll need to visit [its card](https://huggingface.co/CompVis/stable-diffusion-v1-4), read the license and tick the checkbox if you agree.
You have to be a registered user in 🤗 Hugging Face Hub, and you'll also need to use an access token for the code to work. For more information on access tokens, please refer to [this section of the documentation](https://huggingface.co/docs/hub/security-tokens).
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.
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:
...
...
@@ -167,7 +167,7 @@ You can find [this project on Weights and Biases](https://wandb.ai/sayakpaul/tex
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.
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.
## Training with LoRA
...
...
@@ -186,7 +186,7 @@ on consumer GPUs like Tesla T4, Tesla V100.
### Training
First, you need to set up your development environment as is explained in the [installation section](#installing-the-dependencies). Make sure to set the `MODEL_NAME` and `DATASET_NAME` environment variables. Here, we will use [Stable Diffusion v1-4](https://hf.co/CompVis/stable-diffusion-v1-4) and the [Pokemons dataset](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions).
First, you need to set up your development environment as is explained in the [installation section](#installing-the-dependencies). Make sure to set the `MODEL_NAME` and `DATASET_NAME` environment variables. Here, we will use [Stable Diffusion v1-4](https://hf.co/CompVis/stable-diffusion-v1-4) and the [Pokemons dataset](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions).
**___Note: Change the `resolution` to 768 if you are using the [stable-diffusion-2](https://huggingface.co/stabilityai/stable-diffusion-2) 768x768 model.___**
For this example we want to directly store the trained LoRA embeddings on the Hub, so
For this example we want to directly store the trained LoRA embeddings on the Hub, so
we need to be logged in and add the `--push_to_hub` flag.
```bash
...
...
@@ -225,11 +225,11 @@ The above command will also run inference as fine-tuning progresses and log the
The final LoRA embedding weights have been uploaded to [sayakpaul/sd-model-finetuned-lora-t4](https://huggingface.co/sayakpaul/sd-model-finetuned-lora-t4). **___Note: [The final weights](https://huggingface.co/sayakpaul/sd-model-finetuned-lora-t4/blob/main/pytorch_lora_weights.bin) are only 3 MB in size, which is orders of magnitudes smaller than the original model.___**
You can check some inference samples that were logged during the course of the fine-tuning process [here](https://wandb.ai/sayakpaul/text2image-fine-tune/runs/q4lc0xsw).
You can check some inference samples that were logged during the course of the fine-tuning process [here](https://wandb.ai/sayakpaul/text2image-fine-tune/runs/q4lc0xsw).
### Inference
Once you have trained a model using above command, the inference can be done simply using the `StableDiffusionPipeline` after loading the trained LoRA weights. You
Once you have trained a model using above command, the inference can be done simply using the `StableDiffusionPipeline` after loading the trained LoRA weights. You
need to pass the `output_dir` for loading the LoRA weights which, in this case, is `sd-pokemon-model-lora`.
```python
...
...
@@ -248,9 +248,9 @@ image.save("pokemon.png")
If you are loading the LoRA parameters from the Hub and if the Hub repository has
a `base_model` tag (such as [this](https://huggingface.co/sayakpaul/sd-model-finetuned-lora-t4/blob/main/README.md?code=true#L4)), then
To run on your own training files prepare the dataset according to the format required by `datasets`, you can find the instructions for how to do that in this [document](https://huggingface.co/docs/datasets/v2.4.0/en/image_load#imagefolder-with-metadata).
...
...
@@ -321,5 +321,5 @@ According to [this issue](https://github.com/huggingface/diffusers/issues/2234#i
## Stable Diffusion XL
* We support fine-tuning the UNet shipped in [Stable Diffusion XL](https://huggingface.co/papers/2307.01952) via the `train_text_to_image_sdxl.py` script. Please refer to the docs [here](./README_sdxl.md).
* We also support fine-tuning of the UNet and Text Encoder shipped in [Stable Diffusion XL](https://huggingface.co/papers/2307.01952) with LoRA via the `train_text_to_image_lora_sdxl.py` script. Please refer to the docs [here](./README_sdxl.md).
* We support fine-tuning the UNet shipped in [Stable Diffusion XL](https://huggingface.co/papers/2307.01952) via the `train_text_to_image_sdxl.py` script. Please refer to the docs [here](./README_sdxl.md).
* We also support fine-tuning of the UNet and Text Encoder shipped in [Stable Diffusion XL](https://huggingface.co/papers/2307.01952) with LoRA via the `train_text_to_image_lora_sdxl.py` script. Please refer to the docs [here](./README_sdxl.md).
[Textual inversion](https://arxiv.org/abs/2208.01618) is a method to personalize text2image models like stable diffusion on your own images using just 3-5 examples.
The `textual_inversion.py` script shows how to implement the training procedure and adapt it for stable diffusion.
## Running on Colab
## Running on Colab
Colab for training
Colab for training
[](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_textual_inversion_training.ipynb)
@@ -72,7 +72,7 @@ In a nutshell, LoRA allows adapting pretrained models by adding pairs of rank-de
### Prior Training
First, you need to set up your development environment as explained in the [installation](#Running-locally-with-PyTorch) section. Make sure to set the `DATASET_NAME` environment variable. Here, we will use the [Pokemon captions dataset](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions).
First, you need to set up your development environment as explained in the [installation](#Running-locally-with-PyTorch) section. Make sure to set the `DATASET_NAME` environment variable. Here, we will use the [Pokemon captions dataset](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions).
