The tendency of hallucinated responses to be longer reflects two mechanisms: models attempting to maintain coherence while generating incorrect information, and a 'snowball effect' where initial errors cascade into further mistakes, increasing verbosity.

Authors

• Person: Not available Organization: arXiv
  Re-evaluating Hallucination Detection in LLMs - arXiv

Sources

• Re-evaluating Hallucination Detection in LLMs - arXiv arxiv.org arXiv via serper

Referenced by nodes (1)

• Large Language Models concept