This LoRA model is trained on a dataset of 1,200 images featuring Eastern female cosplay costumes and styles. It demonstrates balanced and effective responsiveness to both Chinese and English prompts.