|
--- |
|
license: other |
|
license_name: flux-1-dev-non-commercial-license |
|
license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md |
|
base_model: |
|
- ashen0209/Flux-Dev2Pro |
|
- black-forest-labs/FLUX.1-schnell |
|
tags: |
|
- text-to-image |
|
- flux |
|
- merge |
|
widget: |
|
- text: Example 1 |
|
output: |
|
url: images/0015.webp |
|
- text: Example 2 |
|
output: |
|
url: images/0022.webp |
|
- text: Example 3 |
|
output: |
|
url: images/0012.webp |
|
- text: Example 4 |
|
output: |
|
url: images/0014.webp |
|
- text: Example 5 |
|
output: |
|
url: images/0033.webp |
|
- text: Example 6 |
|
output: |
|
url: images/0032.webp |
|
--- |
|
|
|
# Flux D+S F8 Diffusers |
|
|
|
A + B merge, meant for gradio inference, possibly build adapters on top with [diffusers](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth_lora_flux.py). |
|
|
|
<Gallery /> |
|
|
|
## Inference |
|
|
|
[Instruction](https://huggingface.co/twodgirl/flux-dev-fp8-e4m3fn-diffusers) |
|
|
|
This model needs 10 steps, otherwise the images remain blurry. |
|
|
|
## ComfyUI |
|
|
|
I convert all my checkpoints locally. There are 5+ file formats out there, they take up too much space. |
|
|
|
Download the contents of this repo, then enter the directory. A script will convert the diffusers format to something that is compatible with ComfyUI. |
|
|
|
``` |
|
cd flux-devpro-schnell-merge-fp8-e4m3fn-diffusers |
|
pip install safetensors torch |
|
python flux_devpro_ckpt.py transformer/diffusion_pytorch_model.safetensors flux-devpro-schnell.safetensors |
|
mv flux-devpro-schnell.safetensors path/to/your/ComfyUI/models/unet/flux-devpro-schnell.safetensors |
|
``` |
|
|
|
In the workflow, the LoadDiffusionModel node should take two values: flux-devpro-schnell.safetensors, fp8_e4m3fn. |
|
|
|
Use of this code requires citation and attribution to the author via a link to their Hugging Face profile in all resulting work. |
|
|
|
## Disclaimer |
|
|
|
Sharing, reusing the model weights requires a link back to authors and their Hugging Face profile. The source models were uploaded by Ashen0209 and BFL. |
|
|
