Flux Dev2Pro Fp8 E4m3fn

🚀 Flux-Dev2Pro

Flux-Dev2Pro finetunes the transformer of Flux-Dev to enhance LoRA training performance.

The model has been converted to fp8_e4m3fn, and the rest remains the same as the upstream version. Check it out here.

You can also access the Colab Flux-dev2pro fp8 script.

🚀 Quick Start

To use this model, you can run the following code:

from diffusers import FluxTransformer2DModel

transformer = FluxTransformer2DModel.from_pretrained("rockerBOO/Flux-Dev2Pro-fp8_e4m3fn")

✨ Features

Flux-Dev2Pro finetunes the transformer of Flux-Dev to make LoRA training better. As discussed in this blog, LoRA trained on Flux-Dev often yields bad results because without guidance distillation, the LoRA training diverges from the original training process. Flux-Dev2Pro recovers Flux-pro from Flux-dev by finetuning the model for many steps. Two epochs of 3M high-quality images have been trained. The LoRA trained on Flux-Dev2pro yields much better results when applied to Flux-dev, similar to how LoRA trained on SDXL performs well when applied to SDXL-turbo/lightning.

📄 License

“The FLUX.1 [dev] Model is licensed by Black Forest Labs. Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs. Inc.

IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.”