Commit
•
24e8e8b
1
Parent(s):
46fe2aa
Update README.md
Browse files
README.md
CHANGED
@@ -1,8 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
```python
|
2 |
from diffusers import UnCLIPPipeline
|
3 |
import torch
|
4 |
-
import random
|
5 |
-
import numpy as np
|
6 |
|
7 |
pipe = UnCLIPPipeline.from_pretrained("fusing/karlo_unclip", torch_dtype=torch.float16)
|
8 |
pipe = pipe.to('cuda')
|
@@ -11,7 +16,12 @@ prompt = "a high-resolution photograph of a big red frog on a green leaf."
|
|
11 |
|
12 |
image = pipe([prompt]).images[0]
|
13 |
|
14 |
-
image.save("
|
15 |
```
|
16 |
|
17 |
-
![img](https://huggingface.co/datasets/patrickvonplaten/images/resolve/main/frog.png)
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
## Karlo v1 alpha
|
2 |
+
|
3 |
+
Karlo is a text-conditional image generation model based on OpenAI's unCLIP architecture with the improvement over the standard super-resolution model from 64px to 256px, recovering high-frequency details only in the small number of denoising steps.
|
4 |
+
|
5 |
+
|
6 |
+
Karlo is available in diffusers!
|
7 |
+
|
8 |
```python
|
9 |
from diffusers import UnCLIPPipeline
|
10 |
import torch
|
|
|
|
|
11 |
|
12 |
pipe = UnCLIPPipeline.from_pretrained("fusing/karlo_unclip", torch_dtype=torch.float16)
|
13 |
pipe = pipe.to('cuda')
|
|
|
16 |
|
17 |
image = pipe([prompt]).images[0]
|
18 |
|
19 |
+
image.save("./frog.png")
|
20 |
```
|
21 |
|
22 |
+
![img](https://huggingface.co/datasets/patrickvonplaten/images/resolve/main/frog.png)
|
23 |
+
|
24 |
+
[Original codebase](https://github.com/kakaobrain/karlo)
|
25 |
+
|
26 |
+
This alpha version of Karlo is trained on 115M image-text pairs, including [COYO](https://github.com/kakaobrain/coyo-dataset)-100M high-quality subset, CC3M, and CC12M. For those who are interested in a better version of Karlo trained on more large-scale high-quality datasets, please visit the landing page of our application [B^DISCOVER](https://bdiscover.kakaobrain.com/).
|
27 |
+
|