Instructions to use uglasses/IdentiFace-PixArt-FFHQ with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use uglasses/IdentiFace-PixArt-FFHQ with Diffusers:
pip install -U diffusers transformers accelerate
from diffusers import ControlNetModel, StableDiffusionControlNetPipeline controlnet = ControlNetModel.from_pretrained("uglasses/IdentiFace-PixArt-FFHQ") pipe = StableDiffusionControlNetPipeline.from_pretrained( "fill-in-base-model", controlnet=controlnet ) - Notebooks
- Google Colab
- Kaggle
IdentiFace PixArt FFHQ
PixArt-α ControlNet weights fine-tuned on FFHQ for identity-preserving face generation.
Files
controlnet_FFHQ_base_model_30epochs_lpr03/checkpoints/latest.pth— trained checkpoint (30 epochs)controlnet_FFHQ_base_model_30epochs_lpr03/config.py— training configurationcontrolnet_FFHQ_base_model_30epochs_lpr03/train_log.log— training log
Training notes
- Base:
PixArtMS_XL_2(PixArt-XL-2-1024-MS.pth) - Dataset: ID-FFHQ
- Control: edge + low-quality features (
edges/lq_feature_1024) - Epochs: 30
- Long-prompt ratio: 0.3
- Identity loss: AdaFace (
adaface_loss_weight=0.1)
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support