datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
KoEVD
KoEVD
KoEVD is a Korean benchmark linking five evaluation or analysis targets through source utterances: utterance-risk judgment, candidate-response safety choice, direct-generation response harmfulness, descriptive response strategies, and a pre-execution mock tool/action-choice diagnostic.
Contents and scope
The canonical corpus contains 13,552 sources and 71,395 response candidates: 30,740 accepted, 27,104 rejected, and 13,551 strongly rejected. Three… See the full description on the dataset page: https://huggingface.co/datasets/KETI-NLP/KoEVD.K-prism
K-Prism
Korean diagnostic benchmark data for evaluating hallucination in vision-language
models. Evaluation code and protocol documentation are available at
alsgur0720/K-Prism.
Files
File
Contents
Text_track.json
504 text-track questions
Image_track.json
498 image-track questions
images/
165 original images referenced by the image track
The release contains 1,002 questions and approximately 303 MB of annotations and
images. Keep both JSON files… See the full description on the dataset page: https://huggingface.co/datasets/KETI-NLP/K-prism.