|
| 1 | +<!--Copyright 2022 The HuggingFace Team. All rights reserved. |
| 2 | + |
| 3 | +Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with |
| 4 | +the License. You may obtain a copy of the License at |
| 5 | + |
| 6 | +http://www.apache.org/licenses/LICENSE-2.0 |
| 7 | + |
| 8 | +Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on |
| 9 | +an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the |
| 10 | +specific language governing permissions and limitations under the License. |
| 11 | +--> |
| 12 | + |
| 13 | +# Safe Stable Diffusion |
| 14 | + |
| 15 | +Safe Stable Diffusion was proposed in [Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models](https://arxiv.org/abs/2211.05105) and mitigates the well known issue that models like Stable Diffusion that are trained on unfiltered, web-crawled datasets tend to suffer from inappropriate degeneration. For instance Stable Diffusion may unexpectedly generate nudity, violence, images depicting self-harm, or otherwise offensive content. |
| 16 | +Safe Stable Diffusion is an extension to the Stable Diffusion that drastically reduces content like this. |
| 17 | + |
| 18 | +The abstract of the paper is the following: |
| 19 | + |
| 20 | +*Text-conditioned image generation models have recently achieved astonishing results in image quality and text alignment and are consequently employed in a fast-growing number of applications. Since they are highly data-driven, relying on billion-sized datasets randomly scraped from the internet, they also suffer, as we demonstrate, from degenerated and biased human behavior. In turn, they may even reinforce such biases. To help combat these undesired side effects, we present safe latent diffusion (SLD). Specifically, to measure the inappropriate degeneration due to unfiltered and imbalanced training sets, we establish a novel image generation test bed-inappropriate image prompts (I2P)-containing dedicated, real-world image-to-text prompts covering concepts such as nudity and violence. As our exhaustive empirical evaluation demonstrates, the introduced SLD removes and suppresses inappropriate image parts during the diffusion process, with no additional training required and no adverse effect on overall image quality or text alignment.* |
| 21 | + |
| 22 | + |
| 23 | +*Overview*: |
| 24 | + |
| 25 | +| Pipeline | Tasks | Colab | Demo |
| 26 | +|---|---|:---:|:---:| |
| 27 | +| [pipeline_stable_diffusion_safe.py](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion_safe/pipeline_stable_diffusion_safe.py) | *Text-to-Image Generation* | [](https://colab.research.google.com/github/ml-research/safe-latent-diffusion/blob/main/examples/Safe%20Latent%20Diffusion.ipynb) | - |
| 28 | + |
| 29 | +## Tips |
| 30 | + |
| 31 | +- Safe Stable Diffusion may also be used with weights of [Stable Diffusion](./api/pipelines/stable_diffusion). |
| 32 | + |
| 33 | +### Run Safe Stable Diffusion |
| 34 | + |
| 35 | +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). |
| 36 | + |
| 37 | +### Interacting with the Safety Concept |
| 38 | + |
| 39 | +To check and edit the currently used safety concept, use the `safety_concept` property of [`StableDiffusionPipelineSafe`] |
| 40 | +```python |
| 41 | +>>> from diffusers import StableDiffusionPipelineSafe |
| 42 | + |
| 43 | +>>> pipeline = StableDiffusionPipelineSafe.from_pretrained("AIML-TUDA/stable-diffusion-safe") |
| 44 | +>>> pipeline.safety_concept |
| 45 | +``` |
| 46 | +For each image generation the active concept is also contained in [`StableDiffusionSafePipelineOutput`]. |
| 47 | +
|
| 48 | +### Using pre-defined safety configurations |
| 49 | +
|
| 50 | +You may use the 4 configurations defined in the [Safe Latent Diffusion paper](https://arxiv.org/abs/2211.05105) as follows: |
| 51 | +
|
| 52 | +```python |
| 53 | +>>> from diffusers import StableDiffusionPipelineSafe |
| 54 | +>>> from diffusers.pipelines.stable_diffusion_safe import SafetyConfig |
| 55 | +
|
| 56 | +>>> pipeline = StableDiffusionPipelineSafe.from_pretrained("AIML-TUDA/stable-diffusion-safe") |
| 57 | +>>> prompt = "the four horsewomen of the apocalypse, painting by tom of finland, gaston bussiere, craig mullins, j. c. leyendecker" |
| 58 | +>>> out = pipeline(prompt=prompt, **SafetyConfig.MAX) |
| 59 | +``` |
| 60 | + |
| 61 | +The following configurations are available: `SafetyConfig.WEAK`, `SafetyConfig.MEDIUM`, `SafetyConfig.STRONg`, and `SafetyConfig.MAX`. |
| 62 | + |
| 63 | +### How to load and use different schedulers. |
| 64 | + |
| 65 | +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. |
| 66 | +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: |
| 67 | + |
| 68 | +```python |
| 69 | +>>> from diffusers import StableDiffusionPipelineSafe, EulerDiscreteScheduler |
| 70 | + |
| 71 | +>>> pipeline = StableDiffusionPipelineSafe.from_pretrained("AIML-TUDA/stable-diffusion-safe") |
| 72 | +>>> pipeline.scheduler = EulerDiscreteScheduler.from_config(pipeline.scheduler.config) |
| 73 | + |
| 74 | +>>> # or |
| 75 | +>>> euler_scheduler = EulerDiscreteScheduler.from_pretrained("AIML-TUDA/stable-diffusion-safe", subfolder="scheduler") |
| 76 | +>>> pipeline = StableDiffusionPipelineSafe.from_pretrained( |
| 77 | +... "AIML-TUDA/stable-diffusion-safe", scheduler=euler_scheduler |
| 78 | +... ) |
| 79 | +``` |
| 80 | + |
| 81 | + |
| 82 | +## StableDiffusionSafePipelineOutput |
| 83 | +[[autodoc]] pipelines.stable_diffusion_safe.StableDiffusionSafePipelineOutput |
| 84 | + |
| 85 | +## StableDiffusionPipelineSafe |
| 86 | +[[autodoc]] StableDiffusionPipelineSafe |
| 87 | + - __call__ |
| 88 | + - enable_attention_slicing |
| 89 | + - disable_attention_slicing |
| 90 | + |
0 commit comments