Bk Sdm Small

🚀 BK-SDM Model Card

The Block-removed Knowledge-distilled Stable Diffusion Model (BK-SDM) is an architecturally compressed SDM designed for efficient general-purpose text-to-image synthesis. It achieves this by removing several residual and attention blocks from the U-Net of Stable Diffusion v1.4 and performing distillation pretraining on only 0.22M LAION pairs (less than 0.1% of the full training set). Despite limited training resources, this compact model can imitate the original SDM through transferred knowledge.

• More Information Resources: Paper, GitHub, Demo.

🚀 Quick Start

✨ Features

• Architectural Compression: Removes specific residual and attention blocks from the U-Net of Stable Diffusion v1.4.
• Knowledge Distillation: Conducts distillation pretraining on a small subset of LAION pairs.
• Efficient Synthesis: Capable of efficient general-purpose text-to-image synthesis with limited resources.

📦 Installation

This section does not contain specific installation steps, so it is skipped.

💻 Usage Examples

Basic Usage

Here are examples using the 🤗Diffusers library.

The following is an inference code using the default PNDM scheduler and 50 denoising steps:

import torch
from diffusers import StableDiffusionPipeline

pipe = StableDiffusionPipeline.from_pretrained("nota-ai/bk-sdm-small", torch_dtype=torch.float16)
pipe = pipe.to("cuda")

prompt = "a tropical bird sitting on a branch of a tree"
image = pipe(prompt).images[0]  
    
image.save("example.png")

Advanced Usage

The following code is also runnable because the U-Net of Stable Diffusion v1.4 is compressed while keeping the other parts (i.e., Text Encoder and Image Decoder) unchanged:

import torch
from diffusers import StableDiffusionPipeline, UNet2DConditionModel

pipe = StableDiffusionPipeline.from_pretrained("CompVis/stable-diffusion-v1-4", torch_dtype=torch.float16)
pipe.unet = UNet2DConditionModel.from_pretrained("nota-ai/bk-sdm-small", subfolder="unet", torch_dtype=torch.float16)
pipe = pipe.to("cuda")

prompt = "a tropical bird sitting on a branch of a tree"
image = pipe(prompt).images[0]  
    
image.save("example.png")

📚 Documentation

Compression Method

U-Net Architecture

Certain residual and attention blocks were removed from the U-Net of SDM-v1.4:

• 1.04B-param SDM-v1.4 (0.86B-param U-Net): The original source model.
• 0.76B-param BK-SDM-Base (0.58B-param U-Net): Obtained by removing fewer blocks in the outer stages.
• 0.66B-param BK-SDM-Small (0.49B-param U-Net): Obtained by removing blocks in the outer and middle stages.
• 0.50B-param BK-SDM-Tiny (0.33B-param U-Net): Obtained by removing blocks in the outer, middle, and inner stages.

Distillation Pretraining

The compact U-Net was trained to mimic the behavior of the original U-Net. Feature-level and output-level distillation were used, along with the denoising task loss.

• Training Data: 212,776 image-text pairs (i.e., 0.22M pairs) from LAION-Aesthetics V2 6.5+.
• Hardware: A single NVIDIA A100 80GB GPU
• Gradient Accumulations: 4
• Batch: 256 (=4×64)
• Optimizer: AdamW
• Learning Rate: A constant learning rate of 5e-5 for 50K-iteration pretraining

Experimental Results

The following table shows the zero-shot results on 30K samples from the MS-COCO validation split. After generating 512×512 images with the PNDM scheduler and 25 denoising steps, the images were downsampled to 256×256 for evaluating generation scores. The models were drawn at the 50K-th training iteration.

The following figure depicts synthesized images with some MS-COCO captions.

Uses

Note: This section is taken from the Stable Diffusion v1 model card (which was based on the DALLE-MINI model card) and applies in the same way to BK-SDMs.

Direct Use

The model is intended for research purposes only. Possible research areas and tasks include:

• Safe deployment of models which have the potential to generate harmful content.
• Probing and understanding the limitations and biases of generative models.
• Generation of artworks and use in design and other artistic processes.
• Applications in educational or creative tools.
• Research on generative models.

Excluded uses are described below.

Misuse, Malicious Use, and Out-of-Scope Use

The model should not be used to intentionally create or disseminate images that create hostile or alienating environments for people. This includes generating images that people would foreseeably find disturbing, distressing, or offensive; or content that propagates historical or current stereotypes.

Out-of-Scope Use

The model was not trained to be factual or true representations of people or events, and therefore using the model to generate such content is out-of-scope for the abilities of this model.

Misuse and Malicious Use

Using the model to generate content that is cruel to individuals is a misuse of this model. This includes, but is not limited to:

• Generating demeaning, dehumanizing, or otherwise harmful representations of people or their environments, cultures, religions, etc.
• Intentionally promoting or propagating discriminatory content or harmful stereotypes.
• Impersonating individuals without their consent.
• Sexual content without consent of the people who might see it.
• Mis- and disinformation
• Representations of egregious violence and gore
• Sharing of copyrighted or licensed material in violation of its terms of use.
• Sharing content that is an alteration of copyrighted or licensed material in violation of its terms of use.

Limitations and Bias

Limitations

• The model does not achieve perfect photorealism.
• The model cannot render legible text.
• The model does not perform well on more difficult tasks which involve compositionality, such as rendering an image corresponding to “A red cube on top of a blue sphere”.
• Faces and people in general may not be generated properly.
• The model was trained mainly with English captions and will not work as well in other languages.
• The autoencoding part of the model is lossy.
• The model was trained on a large-scale dataset LAION-5B which contains adult material and is not fit for product use without additional safety mechanisms and considerations.
• No additional measures were used to deduplicate the dataset. As a result, some degree of memorization for images that are duplicated in the training data is observed. The training data can be searched at https://rom1504.github.io/clip-retrieval/ to possibly assist in the detection of memorized images.

Bias

While the capabilities of image generation models are impressive, they can also reinforce or exacerbate social biases. Stable Diffusion v1 was trained on subsets of LAION-2B(en), which consists of images that are primarily limited to English descriptions. Texts and images from communities and cultures that use other languages are likely to be insufficiently accounted for. This affects the overall output of the model, as white and western cultures are often set as the default. Further, the ability of the model to generate content with non-English prompts is significantly worse than with English-language prompts.

Safety Module

The intended use of this model is with the Safety Checker in Diffusers. This checker works by checking model outputs against known hard-coded NSFW concepts. The concepts are intentionally hidden to reduce the likelihood of reverse-engineering this filter. Specifically, the checker compares the class probability of harmful concepts in the embedding space of the CLIPTextModel after generation of the images. The concepts are passed into the model with the generated image and compared to a hand-engineered weight for each NSFW concept.

🔧 Technical Details

This section contains sufficient technical details, so it is included. The details are mainly about the compression method (U-Net architecture and distillation pretraining), experimental results, and the limitations and bias of the model.

📄 License

This model is open access and available to all, with a CreativeML OpenRAIL-M license further specifying rights and usage.

The CreativeML OpenRAIL License specifies:

1. You can't use the model to deliberately produce nor share illegal or harmful outputs or content.
2. The authors claim no rights on the outputs you generate, you are free to use them and are accountable for their use which must not go against the provisions set in the license.
3. You may re-distribute the weights and use the model commercially and/or as a service. If you do, please be aware you have to include the same use restrictions as the ones in the license and share a copy of the CreativeML OpenRAIL-M to all your users (please read the license entirely and carefully).

Please read the full license carefully here: https://huggingface.co/spaces/CompVis/stable-diffusion-license

Acknowledgments

• We express our gratitude to Microsoft for Startups Founders Hub for generously providing the Azure credits used during pretraining.
• We deeply appreciate the pioneering research on Latent/Stable Diffusion conducted by CompVis, Runway, and Stability AI.
• Special thanks to the contributors to LAION, Diffusers, and Gradio for their valuable support.

Citation

@article{kim2023architectural,
  title={BK-SDM: A Lightweight, Fast, and Cheap Version of Stable Diffusion},
  author={Kim, Bo-Kyeong and Song, Hyoung-Kyu and Castells, Thibault and Choi, Shinkook},
  journal={arXiv preprint arXiv:2305.15798},
  year={2023},
  url={https://arxiv.org/abs/2305.15798}
}

@article{kim2023bksdm,
  title={BK-SDM: Architecturally Compressed Stable Diffusion for Efficient Text-to-Image Generation},
  author={Kim, Bo-Kyeong and Song, Hyoung-Kyu and Castells, Thibault and Choi, Shinkook},
  journal={ICML Workshop on Efficient Systems for Foundation Models (ES-FoMo)},
  year={2023},
  url={https://openreview.net/forum?id=bOVydU0XKC}
}

This model card was written by Bo-Kyeong Kim and is based on the Stable Diffusion v1 model card.