Support on dataset creation

Hi all,
Some time ago, we published a Github repository to craft a dataset for ControlNet fine-tuning. We also used it to train an interior design model. Unfortunately, it wasn’t very useful in the past few weeks since the LAION dataset is offline.

We have published a dataset on Huggingface based on the Datacomp-12.8M dataset, and updated our code base to use this dataset. You can start using the pipelines there to craft a new dataset for fine-tuning ControlNet. Let us know if you run into any issues. We’re happy to help!

GitHub repository: GitHub - ml6team/fondant-usecase-controlnet: Example Fondant pipeline preparing data to train a Controlnet model
HuggingFace dataset: fondant-ai/datacomp-small-clip · Datasets at Hugging Face