anvilarth commited on
Commit
fb6e70c
1 Parent(s): 5c3e1fc

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +2 -2
README.md CHANGED
@@ -2,7 +2,7 @@
2
  license: apache-2.0
3
  ---
4
 
5
- # Kandinsky-3: Text-to-image diffusion model
6
 
7
  ![](assets/title.jpg)
8
 
@@ -21,7 +21,7 @@ We present Kandinsky 3.1, the follow-up to the Kandinsky 3.0 model, a large-scal
21
  </figure>
22
 
23
 
24
- Diffusion models have problems with fast image generation. To address this problem, we trained a Kandinksy Flash model based on the [Adversarial Diffusion Distillation](https://arxiv.org/abs/2311.17042) approach with some modifications: we trained the model on latents, which reduced the memory overhead and removed distillation loss as it did not affect the training.
25
 
26
  ### Architecture
27
 
 
2
  license: apache-2.0
3
  ---
4
 
5
+ # Kandinsky-3.1: Text-to-image diffusion model
6
 
7
  ![](assets/title.jpg)
8
 
 
21
  </figure>
22
 
23
 
24
+ Diffusion models have problems with fast image generation. To address this problem, we trained a Kandinksy Flash model based on the [Adversarial Diffusion Distillation](https://arxiv.org/abs/2311.17042) approach with some modifications: we trained the model on latents, which reduced the memory overhead and removed distillation loss as it did not affect the training. Also we used Kandinsky Flash model to improve visual quality of generation from Kandinsky 3.0.
25
 
26
  ### Architecture
27