Trained with faster unet and text encoder learning rates for even better likeness. Will be the final version of this style lora.
Version 2 of my ThePit style lora, trained with better training settings, gives better style likeness and image composition imo, moreso than the previous version
