This is the same training data as Flux. What kind of results will we get with XL?:) The sample prompts are all as follows: a girl { 2$$ glasses | earrings | hair ribbon}