TofuEval: Evaluating Hallucinations of LLMs on Topic-Focused Dialogue Summarization

arXiv:2402.13249v1 Announce Type: new
Abstract: Single document news summarization has seen substantial progress on faithfulness in recent years, driven by research on the evaluation of factual consistency, or hallucinations. We ask whether these advances carry over to other text summarization domains. We propose a new evaluation benchmark on topic-focused dialogue summarization, generated by LLMs of varying sizes. We provide binary sentence-level human annotations of the factual consistency of these summaries along with detailed explanations of factually inconsistent sentences. Our analysis …

abstract advances arxiv benchmark cs.ai cs.cl dialogue document domains evaluation hallucinations llms progress research summarization text text summarization type