MALTO at SemEval-2024 Task 6: Leveraging Synthetic Data for LLM Hallucination Detection

arXiv:2403.00964v1 Announce Type: cross
Abstract: In Natural Language Generation (NLG), contemporary Large Language Models (LLMs) face several challenges, such as generating fluent yet inaccurate outputs and reliance on fluency-centric metrics. This often leads to neural networks exhibiting "hallucinations". The SHROOM challenge focuses on automatically identifying these hallucinations in the generated text. To tackle these issues, we introduce two key components, a data augmentation pipeline incorporating LLM-assisted pseudo-labelling and sentence rephrasing, and a voting ensemble from three models pre-trained on Natural …

abstract arxiv challenge challenges cs.cl cs.lg data detection face hallucination hallucinations language language generation language models large language large language models leads llm llm hallucination llms metrics natural natural language natural language generation networks neural networks nlg reliance synthetic synthetic data type