Commit
•
25230d9
1
Parent(s):
271d743
Update README.md
Browse files
README.md
CHANGED
|
@@ -34,7 +34,7 @@ extra_gated_heading: Please read the LICENSE to access this model
|
|
| 34 |
|
| 35 |
# BK-SDM Model Card
|
| 36 |
Block-removed Knowledge-distilled Stable Diffusion Model (BK-SDM) is an architecturally compressed SDM for efficient general-purpose text-to-image synthesis. This model is bulit with (i) removing several residual and attention blocks from the U-Net of [Stable Diffusion v1.4]( https://huggingface.co/CompVis/stable-diffusion-v1-4) and (ii) distillation pretraining on only 0.22M LAION pairs (fewer than 0.1% of the full training set). Despite being trained with very limited resources, our compact model can imitate the original SDM by benefiting from transferred knowledge.
|
| 37 |
-
- **Resources for more information**: [Paper](https://arxiv.org/abs/2305.15798), [Demo]( https://huggingface.co/spaces/nota-ai/compressed-stable-diffusion).
|
| 38 |
|
| 39 |
|
| 40 |
|
|
@@ -189,8 +189,8 @@ The intended use of this model is with the [Safety Checker](https://github.com/h
|
|
| 189 |
|
| 190 |
# Acknowledgments
|
| 191 |
- We express our gratitude to [Microsoft for Startups Founders Hub](https://www.microsoft.com/en-us/startups) for generously providing the Azure credits used during pretraining.
|
| 192 |
-
- We deeply appreciate the pioneering research on Latent/Stable Diffusion conducted by [CompVis](https://github.com/CompVis/latent-diffusion)
|
| 193 |
-
- Special thanks to the contributors to [Diffusers](https://github.com/huggingface/diffusers) for their valuable support.
|
| 194 |
|
| 195 |
|
| 196 |
# Citation
|
|
|
|
| 34 |
|
| 35 |
# BK-SDM Model Card
|
| 36 |
Block-removed Knowledge-distilled Stable Diffusion Model (BK-SDM) is an architecturally compressed SDM for efficient general-purpose text-to-image synthesis. This model is bulit with (i) removing several residual and attention blocks from the U-Net of [Stable Diffusion v1.4]( https://huggingface.co/CompVis/stable-diffusion-v1-4) and (ii) distillation pretraining on only 0.22M LAION pairs (fewer than 0.1% of the full training set). Despite being trained with very limited resources, our compact model can imitate the original SDM by benefiting from transferred knowledge.
|
| 37 |
+
- **Resources for more information**: [Paper](https://arxiv.org/abs/2305.15798), [GitHub](https://github.com/Nota-NetsPresso/BK-SDM), [Demo]( https://huggingface.co/spaces/nota-ai/compressed-stable-diffusion).
|
| 38 |
|
| 39 |
|
| 40 |
|
|
|
|
| 189 |
|
| 190 |
# Acknowledgments
|
| 191 |
- We express our gratitude to [Microsoft for Startups Founders Hub](https://www.microsoft.com/en-us/startups) for generously providing the Azure credits used during pretraining.
|
| 192 |
+
- We deeply appreciate the pioneering research on Latent/Stable Diffusion conducted by [CompVis](https://github.com/CompVis/latent-diffusion), [Runway](https://runwayml.com/), and [Stability AI](https://stability.ai/).
|
| 193 |
+
- Special thanks to the contributors to [LAION](https://laion.ai/), [Diffusers](https://github.com/huggingface/diffusers), and [Gradio](https://www.gradio.app/) for their valuable support.
|
| 194 |
|
| 195 |
|
| 196 |
# Citation
|