Upload 8–60 images to fine-tune a FLUX LoRA model.
A unique token the LoRA learns. Used at inference time.