InterrogateLLM: Zero-Resource Hallucination Detection in LLM-Generated Answers (2403.02889v3)
Abstract: Despite the many advances of LLMs and their unprecedented rapid evolution, their impact and integration into every facet of our daily lives is limited due to various reasons. One critical factor hindering their widespread adoption is the occurrence of hallucinations, where LLMs invent answers that sound realistic, yet drift away from factual truth. In this paper, we present a novel method for detecting hallucinations in LLMs, which tackles a critical issue in the adoption of these models in various real-world scenarios. Through extensive evaluations across multiple datasets and LLMs, including Llama-2, we study the hallucination levels of various recent LLMs and demonstrate the effectiveness of our method to automatically detect them. Notably, we observe up to 87% hallucinations for Llama-2 in a specific experiment, where our method achieves a Balanced Accuracy of 81%, all without relying on external knowledge.
- Amos Azaria and Tom Mitchell. 2023. The internal state of an llm knows when its lying.
- Frederic Charles Bartlett. 1995. Remembering: A study in experimental and social psychology. Cambridge university press.
- Beliefs and data on the relationship between consistency and accuracy of eyewitness testimony. Applied Cognitive Psychology: The Official Journal of the Society for Applied Research in Memory and Cognition, 13(4):297–313.
- Language models are few-shot learners. In Advances in Neural Information Processing Systems, volume 33, pages 1877–1901. Curran Associates, Inc.
- Quora question pairs.
- Palm: Scaling language modeling with pathways.
- BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 4171–4186, Minneapolis, Minnesota. Association for Computational Linguistics.
- Rachel E Dianiska and Christian A Meissner. 2023. The effect of credibility assessment techniques on consistency and subsequent memory for the truth. Frontiers in Psychology, 14.
- Gptscore: Evaluate as you desire.
- Pär Anders Granhag and Leif A Strömwall. 2001. Deception detection based on repeated interrogations. Legal and Criminological Psychology, 6(1):85–101.
- Survey of hallucination in natural language generation. ACM Comput. Surv., 55(12).
- Language models (mostly) know what they know.
- Evaluating the factual consistency of abstractive text summarization. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 9332–9346, Online. Association for Computational Linguistics.
- TruthfulQA: Measuring how models mimic human falsehoods. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 3214–3252, Dublin, Ireland. Association for Computational Linguistics.
- A token-level reference-free hallucination detection benchmark for free-form text generation. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 6723–6737, Dublin, Ireland. Association for Computational Linguistics.
- Mqag: Multiple-choice question answering and generation for assessing information consistency in summarization.
- Selfcheckgpt: Zero-resource black-box hallucination detection for generative large language models.
- On faithfulness and factuality in abstractive summarization. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 1906–1919, Online. Association for Computational Linguistics.
- Nils Reimers and Iryna Gurevych. 2019. Sentence-BERT: Sentence embeddings using Siamese BERT-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 3982–3992, Hong Kong, China. Association for Computational Linguistics.
- Dialogue in the wild: Learning from a deployed role-playing game with humans and bots. In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 611–624, Online. Association for Computational Linguistics.
- Llama: Open and efficient foundation language models.
- Llama 2: Open foundation and fine-tuned chat models.
- Yakir Yehuda (3 papers)
- Itzik Malkiel (19 papers)
- Oren Barkan (29 papers)
- Jonathan Weill (8 papers)
- Royi Ronen (5 papers)
- Noam Koenigstein (31 papers)