shahin-canary's picture
Update README.md
d884bc2 verified
|
raw
history blame
1.73 kB
---
base_model: stabilityai/stable-diffusion-xl-base-1.0
library_name: diffusers
license: openrail++
tags:
- text-to-image
- text-to-image
- diffusers-training
- diffusers
- lora
- template:sd-lora
- stable-diffusion-xl
- stable-diffusion-xl-diffusers
instance_prompt: a photo of sks charctr_kyz
widget:
- text: A photo of sks charctr_kyz
output:
url: image_0.png
- text: A photo of sks charctr_kyz
output:
url: image_2.png
- text: A photo of sks charctr_kyz
output:
url: image_3.png
---
<!-- This model card has been generated automatically according to the information the training script had access to. You
should probably proofread and complete it, then remove this comment. -->
# SDXL LoRA DreamBooth - shahin-canary/sdxl-base-1.0-greenchair-dreambooth-lora
<Gallery />
## Model description
These are shahin-canary/sdxl-base-1.0-greenchair-dreambooth-lora LoRA adaption weights for stabilityai/stable-diffusion-xl-base-1.0.
The weights were trained using [DreamBooth](https://dreambooth.github.io/).
LoRA for the text encoder was enabled: True.
Special VAE used for training: madebyollin/sdxl-vae-fp16-fix.
## Trigger words
You should use a photo of sks charctr_kyz to trigger the image generation.
## Download model
Weights for this model are available in Safetensors format.
[Download](shahin-canary/sdxl-base-1.0-greenchair-dreambooth-lora/tree/main) them in the Files & versions tab.
## Intended uses & limitations
#### How to use
```python
# TODO: add an example code snippet for running this diffusion pipeline
```
#### Limitations and bias
[TODO: provide examples of latent issues and potential remediations]
## Training details
[TODO: describe the data used to train the model]