Dataset Card for HalluciGen-Detection
Dataset Summary
This is a dataset for hallucination detection in the paraphrase generation and machine translation scenario. Each example in the dataset consists of a source sentence, a correct hypothesis, and an incorrect hypothesis containing an intrinsic hallucination. A hypothesis is considered to be a hallucination if it is not entailed by the "source" either by containing additional or contradictory information with respect to… See the full description on the dataset page: https://huggingface.co/datasets/NLP-RISE/HalluciGen.
