I used the sd xl base model to create the datasets for a base model and then merged all of that with the base model to create a checkpoint. I then used that checkpoint to create datasets, merging them with the checkpoint to create another base model called Kodak. I then created a 30 image_data set with the Kodak checkpoint, named it 56_Faces and trained this LoRA with that image_data set, the reasoning behind the name being "Kodak_LoRA_Faces".