MM-Hallu/

MM-UPD

MM-UPD (Multimodal Understanding Preference Dataset) is a benchmark for evaluating whether multimodal models can distinguish between hallucinated and truthful descriptions of images. It includes three sub-tasks: AAD (Attribute Anomaly Detection), IASD (Inappropriate Answer Selection Detection), and IVQD (Incorrect Visual Question Detection).

	Fields




	

Field Description

task Sub-dataset name (e.g., mmaad_20240303_base)

image The input image

question The… See the full description on the dataset page: https://huggingface.co/datasets/MM-Hallu/MM-UPD.