Update README.md (#5)
Browse files- Update README.md (3e578ac39387f384f98aa494a12cfd814afafb80)
README.md
CHANGED
|
@@ -40,7 +40,7 @@ image.save("./frog.png")
|
|
| 40 |
Karlo is a text-conditional diffusion model based on unCLIP, composed of prior, decoder, and super-resolution modules. In this repository, we include the improved version of the standard super-resolution module for upscaling 64px to 256px only in 7 reverse steps, as illustrated in the figure below:
|
| 41 |
|
| 42 |
<p float="left">
|
| 43 |
-
<img src="https://raw.githubusercontent.com/kakaobrain/karlo/main/assets/improved_sr_arch.
|
| 44 |
</p>
|
| 45 |
|
| 46 |
In specific, the standard SR module trained by DDPM objective upscales 64px to 256px in the first 6 denoising steps based on the respacing technique. Then, the additional fine-tuned SR module trained by [VQ-GAN](https://compvis.github.io/taming-transformers/)-style loss performs the final reverse step to recover high-frequency details. We observe that this approach is very effective to upscale the low-resolution in a small number of reverse steps.
|
| 40 |
Karlo is a text-conditional diffusion model based on unCLIP, composed of prior, decoder, and super-resolution modules. In this repository, we include the improved version of the standard super-resolution module for upscaling 64px to 256px only in 7 reverse steps, as illustrated in the figure below:
|
| 41 |
|
| 42 |
<p float="left">
|
| 43 |
+
<img src="https://raw.githubusercontent.com/kakaobrain/karlo/main/assets/improved_sr_arch.jpg"/>
|
| 44 |
</p>
|
| 45 |
|
| 46 |
In specific, the standard SR module trained by DDPM objective upscales 64px to 256px in the first 6 denoising steps based on the respacing technique. Then, the additional fine-tuned SR module trained by [VQ-GAN](https://compvis.github.io/taming-transformers/)-style loss performs the final reverse step to recover high-frequency details. We observe that this approach is very effective to upscale the low-resolution in a small number of reverse steps.
|