accept t5xxl style prompts
training performed on a large dataset with llm-generated captions on the base dpo sd1.5 model
may produce NSFW material
training performed on a large dataset with llm-generated captions on the base dpo sd1.5 model
may produce NSFW material