File size: 1,100 Bytes
15b6805 0af31ce e1df254 5a41e68 51d96cf eb9d3d4 15b6805 51d96cf 15b6805 0af31ce eb9d3d4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 |
---
license: apache-2.0
library_name: diffusers
pipeline_tag: text-to-image
---
<img src="https://huggingface.co/ostris/OpenFLUX.1/resolve/main/assets/banner_0_0_9.png" style="max-width: 100%;">
<div style="color: red;">
# <span style="color: red;"> Not Ready </span>
This is a WIP. It is nearing a state I am going to call "early pre alpha". You are welcome to start messing with it at this point, but it still has a ways to go before it is ready for people to start finetuning on it. New updates are being pushed every few days as it continues to train. I will update this message when it is ready for that.
</div>
## What is this?
This project started as a way to train adapters for the [FLUX.1-schnell](https://huggingface.co/black-forest-labs/FLUX.1-schnell) model. But has evolved over time. The purpose of this project now, is to create an open source version of FLUX.1 that uses classic CFG and is finetunable for the community. Since it requires CFG, it will require a different pipeline than the original FLUX.1. This pipeline can be found in open_flux_pipeline.py in this repo.
|