datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Tennis_Betting_CorrelationCorrelationQA
CorrelationQA
This dataset is from the paper: "The Instinctive Bias: Spurious Images lead to Hallucination in MLLMs".
Dataset Description
CorrelationQA is a benchmark for evaluating hallucination in Multimodal Large Language Models (MLLMs) caused by spurious image-text correlations. The dataset contains questions paired with misleading or irrelevant images that may trigger hallucinated responses.
Dataset Structure
image: The image associated with the question… See the full description on the dataset page: https://huggingface.co/datasets/MM-Hallu/CorrelationQA.