This dataset is derived from our COLM 2024 paper, Iteratively Prompting Multimodal LLMs to Reproduce Natural and AI-Generated Images. The paper studies whether multimodal language models can infer prompts that generate images visually similar to target images produced by text-to-image systems or found in stock image collections, highlighting the relationship between real-world prompts and generated images as well as broader economic and security… See the full description on the dataset page:
https://huggingface.co/datasets/AliN96/midjourney-prompts-embeddings.
