Hi @sayakpaul @qgallouedec @DN6 , I’m currently working on a FLUX.1-dev text-to-image pipeline and have been experimenting with InstantX/FLUX.1-dev-controlnet-canny. While the model produces impressive styling, it's trained on a different dataset than what my use case requires. Therefore, I’m looking to train the ControlNet pipeline on my own data.
please provide guidance or share the training script? It would be extremely helpful in understanding how the model is trained, particularly within the Flux Dev framework.