Train a ControlNet on paired conditioning images
A Diffusers implementation of ControlNet training for teaching an image model to follow an auxiliary image condition.
Published by Hugging Face Diffusers
How to use it
The example adapts the original ControlNet coloring demonstration to Diffusers. Its Fill50K dataset provides the target image, conditioning image, and text input used during training.
Check image alignment before training and test on conditions outside the training set. The publisher provides several memory configurations, with progressively more aggressive settings or offloading for smaller GPUs.
- Framework
- Hugging Face Diffusers
- Dataset
- fusing/fill50k
- Hardware
- Publisher reports about 38 GB VRAM for the default setup and documents lower-memory alternatives.
- Access
- Public publisher script and examples; checkpoint and dataset terms apply.