Update README.md
Browse files
README.md
CHANGED
@@ -4,4 +4,56 @@ datasets:
|
|
4 |
library_name: diffusers
|
5 |
tags:
|
6 |
- art
|
7 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
library_name: diffusers
|
5 |
tags:
|
6 |
- art
|
7 |
+
---
|
8 |
+
|
9 |
+
# SDXL LoRA DreamBooth - imomayiz/moroccan_sdxl_lora
|
10 |
+
|
11 |
+
<Gallery />
|
12 |
+
|
13 |
+
## Model description
|
14 |
+
|
15 |
+
### These are LoRA weights of sdxl-base-1.0 finetuned on modern moroccan cities images.
|
16 |
+
|
17 |
+
The weights were trained using [DreamBooth](https://dreambooth.github.io/).
|
18 |
+
|
19 |
+
VAE used for training: madebyollin/sdxl-vae-fp16-fix.
|
20 |
+
|
21 |
+
## Trigger words
|
22 |
+
|
23 |
+
You should use "moroccan city" to trigger the image generation.
|
24 |
+
|
25 |
+
## Download model
|
26 |
+
|
27 |
+
Weights for this model are available [here](imomayiz/moroccan_sdxl_lora/tree/main).
|
28 |
+
|
29 |
+
## Dataset
|
30 |
+
|
31 |
+
The training images can be found [here](https://huggingface.co/datasets/imomayiz/morocco-img/tree/main/data/cities).
|
32 |
+
|
33 |
+
## How to use the model
|
34 |
+
````
|
35 |
+
import torch
|
36 |
+
from diffusers import DiffusionPipeline, AutoencoderKL
|
37 |
+
|
38 |
+
repo_id = "imomayiz/moroccan_sdxl_lora"
|
39 |
+
|
40 |
+
# load lora weights and pipeline
|
41 |
+
vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
|
42 |
+
pipe = DiffusionPipeline.from_pretrained(
|
43 |
+
"stabilityai/stable-diffusion-xl-base-1.0",
|
44 |
+
vae=vae,
|
45 |
+
torch_dtype=torch.float16,
|
46 |
+
variant="fp16",
|
47 |
+
use_safetensors=True
|
48 |
+
)
|
49 |
+
pipe.load_lora_weights(repo_id)
|
50 |
+
_ = pipe.to("cuda")
|
51 |
+
|
52 |
+
|
53 |
+
prompt = "a photo of a modern moroccan city"
|
54 |
+
|
55 |
+
# generate the image
|
56 |
+
image = pipe(prompt=prompt, num_inference_steps=20).images[0]
|
57 |
+
image
|
58 |
+
|
59 |
+
````
|