Update README.md
Browse files
README.md
CHANGED
@@ -1,8 +1,26 @@
|
|
|
|
1 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
|
3 |
-
TODO:
|
4 |
- The fp16 base model is converted to openvino in fp32, which is unnecessary.
|
5 |
|
|
|
|
|
|
|
6 |
|
7 |
```python
|
8 |
import torch
|
@@ -29,4 +47,4 @@ ov_pipeline = OVStableDiffusionPipeline.from_pretrained(
|
|
29 |
export=True,
|
30 |
)
|
31 |
ov_pipeline.save_pretrained(save_ov_folder)
|
32 |
-
```
|
|
|
1 |
+
## yujiepan/dreamshaper-8-lcm-openvino
|
2 |
|
3 |
+
This model applies `latent-consistency/lcm-lora-sdv1-5` to base model `Lykon/dreamshaper-8`, and is converted to OpenVINO format.
|
4 |
+
|
5 |
+
#### Usage
|
6 |
+
|
7 |
+
```python
|
8 |
+
from optimum.intel.openvino.modeling_diffusion import OVStableDiffusionPipeline
|
9 |
+
pipeline = OVStableDiffusionPipeline.from_pretrained(
|
10 |
+
'yujiepan/dreamshaper-8-lcm-openvino',
|
11 |
+
device='CPU',
|
12 |
+
)
|
13 |
+
prompt = 'Cute Dog Typing at a Typewriter German Style'
|
14 |
+
images = pipeline(prompt=prompt, num_inference_steps=8, guidance_scale=1.0).images
|
15 |
+
```
|
16 |
+
|
17 |
+
#### TODO
|
18 |
|
|
|
19 |
- The fp16 base model is converted to openvino in fp32, which is unnecessary.
|
20 |
|
21 |
+
#### Scripts
|
22 |
+
|
23 |
+
The model is generated by the following codes:
|
24 |
|
25 |
```python
|
26 |
import torch
|
|
|
47 |
export=True,
|
48 |
)
|
49 |
ov_pipeline.save_pretrained(save_ov_folder)
|
50 |
+
```
|