|
--- |
|
license: apache-2.0 |
|
base_model: runwayml/stable-diffusion-v1-5 |
|
datasets: |
|
- Oysiyl/google-android-toy |
|
language: |
|
- en |
|
--- |
|
|
|
### Demo |
|
You can try the demo [here](https://sdloraandroidtoy.streamlit.app/). |
|
|
|
For hosting the [frontend](https://github.com/dmitriy-kisil/sd_lora_android_toy_frontend) part [Streamlit Community Cloud](https://streamlit.io/cloud) and [Cerebrium](https://www.cerebrium.ai/) for the [backend](https://github.com/dmitriy-kisil/sd_lora_android_toy_backend) part were used. |
|
|
|
### Model card |
|
|
|
Finetuned from SD 1.5 using LoRA. |
|
|
|
W&B [run](https://wandb.ai/logart1995/text2image-fine-tune/runs/2o98mhc7?workspace=user-logart1995). |
|
|
|
### Inference |
|
|
|
```py |
|
from diffusers import AutoPipelineForText2Image |
|
import torch |
|
|
|
pipe = AutoPipelineForText2Image.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16) |
|
pipe.load_lora_weights("Oysiyl/sd-lora-android-google-toy", weights="pytorch_lora_weights.safetensors") |
|
pipe = pipe.to("cuda") |
|
|
|
g = torch.Generator(device="cuda").manual_seed(42) |
|
|
|
image = pipe("An android toy near Eiffel tower", |
|
num_inference_steps=50, |
|
num_images_per_prompt=1, |
|
guidance_scale=7.5, |
|
temperature=1.0, |
|
generator=g).images[0] |
|
|
|
image.save("android_toy.png") |
|
``` |
|
|
|
### Output |
|
|
|
 |
|
|
