Multimodal image-text Gastrointestinal (GI) dataset, with VLM-generated descriptive responses, expert-labeled tags identifying hallucinated sentences, and their corresponding corrections. Extensive ...