Removing sensitive information from LLM inputs
Getting captions and image previews from a given dataset