Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
97 tokens/sec
GPT-4o
11 tokens/sec
Gemini 2.5 Pro Pro
47 tokens/sec
o3 Pro
5 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

A Comprehensive Survey of Hallucination Mitigation Techniques in Large Language Models (2401.01313v3)

Published 2 Jan 2024 in cs.CL

Abstract: As LLMs continue to advance in their ability to write human-like text, a key challenge remains around their tendency to hallucinate generating content that appears factual but is ungrounded. This issue of hallucination is arguably the biggest hindrance to safely deploying these powerful LLMs into real-world production systems that impact people's lives. The journey toward widespread adoption of LLMs in practical settings heavily relies on addressing and mitigating hallucinations. Unlike traditional AI systems focused on limited tasks, LLMs have been exposed to vast amounts of online text data during training. While this allows them to display impressive language fluency, it also means they are capable of extrapolating information from the biases in training data, misinterpreting ambiguous prompts, or modifying the information to align superficially with the input. This becomes hugely alarming when we rely on language generation capabilities for sensitive applications, such as summarizing medical records, financial analysis reports, etc. This paper presents a comprehensive survey of over 32 techniques developed to mitigate hallucination in LLMs. Notable among these are Retrieval Augmented Generation (Lewis et al, 2021), Knowledge Retrieval (Varshney et al,2023), CoNLI (Lei et al, 2023), and CoVe (Dhuliawala et al, 2023). Furthermore, we introduce a detailed taxonomy categorizing these methods based on various parameters, such as dataset utilization, common tasks, feedback mechanisms, and retriever types. This classification helps distinguish the diverse approaches specifically designed to tackle hallucination issues in LLMs. Additionally, we analyze the challenges and limitations inherent in these techniques, providing a solid foundation for future research in addressing hallucinations and related phenomena within the realm of LLMs.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (51)
  1. Fleek: Factual error detection and correction with evidence retrieved from external knowledge.
  2. A step closer to comprehensive answers: Constrained multi-stage question decomposition with large language models.
  3. Dress: Instructing large vision-language models to align and interact with humans via natural language feedback. arXiv preprint arXiv:2311.10081.
  4. Uprise: Universal prompt retrieval for improving zero-shot evaluation.
  5. George Chrysostomou and Nikolaos Aletras. 2021. Enjoy the salience: Towards better transformer-based faithful explanations with word salience. arXiv preprint arXiv:2108.13759.
  6. Dola: Decoding by contrasting layers improves factuality in large language models.
  7. Scaling instruction-finetuned language models.
  8. Chain-of-verification reduces hallucination in large language models.
  9. Halo: Estimation and reduction of hallucinations in open-source weak large language models.
  10. Trapping llm hallucinations using tagged context prompts.
  11. Rarr: Researching and revising what language models say, using language models. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 16477–16508.
  12. Opt-iml: Scaling language model instruction meta learning through the lens of generalization.
  13. RHO: Reducing hallucination in open-domain dialogues with knowledge grounding. In Findings of the Association for Computational Linguistics: ACL 2023, pages 4504–4522, Toronto, Canada. Association for Computational Linguistics.
  14. Towards mitigating hallucination in large language models via self-reflection.
  15. Teaching language models to hallucinate less with synthetic tasks.
  16. Ever: Mitigating hallucination in large language models through real-time verification and rectification.
  17. Hallucination augmented recitations for language models.
  18. Albert: A lite bert for self-supervised learning of language representations. In International Conference on Learning Representations.
  19. Mateusz Lango and Ondrej Dusek. 2023. Critic-driven decoding for mitigating hallucinations in data-to-text generation. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 2853–2862, Singapore. Association for Computational Linguistics.
  20. Chain of natural language inference for reducing large language model ungrounded hallucinations.
  21. The power of scale for parameter-efficient prompt tuning. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 3045–3059, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics.
  22. Retrieval-augmented generation for knowledge-intensive nlp tasks. arXiv preprint arXiv:2005.11401v4.
  23. Inference-time intervention: Eliciting truthful answers from a language model. arXiv preprint arXiv:2306.03341.
  24. Self-checker: Plug-and-play modules for fact-checking with large language models.
  25. Mind’s mirror: Distilling self-evaluation capability and comprehensive thinking from large language models.
  26. Self-refine: Iterative refinement with self-feedback.
  27. Self-contradictory hallucinations of large language models: Evaluation, detection and mitigation. arXiv preprint arXiv:2305.15852.
  28. Self-contradictory hallucinations of large language models: Evaluation, detection and mitigation.
  29. Check your facts and try again: Improving large language models with external knowledge and automated feedback.
  30. Think while you write: Hypothesis verification promotes faithful knowledge-to-text generation. arXiv preprint arXiv:2311.09467.
  31. Detecting and mitigating hallucinations in multilingual summarisation.
  32. The troubling emergence of hallucination in large language models – an extensive definition, quantification, and prescriptive remediations.
  33. Partha Pratim Ray. 2023. Chatgpt: A comprehensive review on background, applications, key challenges, bias, ethics, limitations and future scope. Internet of Things and Cyber-Physical Systems, 3:121–154.
  34. Dial BeInfo for Faithfulness: Improving factuality of information-seeking dialogue via behavioural fine-tuning.
  35. Distilbert, a distilled version of bert: smaller, faster, cheaper and lighter. arXiv preprint arXiv:1910.01108.
  36. Trusting your evidence: Hallucinate less with context-aware decoding.
  37. Prompting gpt-3 to be reliable. arXiv preprint arXiv:2210.09150.
  38. Head-to-tail: How knowledgeable are large language models (llm)? aka will llms replace knowledge graphs? arXiv preprint arXiv:2308.10168.
  39. Principle-driven self-alignment of language models from scratch with minimal human supervision.
  40. Fine-tuning language models for factuality.
  41. Llama 2: Open foundation and fine-tuned chat models.
  42. A stitch in time saves nine: Detecting and mitigating hallucinations of llms by validating low-confidence generation.
  43. Freshllms: Refreshing large language models with search engine augmentation.
  44. Self-instruct: Aligning language models with self-generated instructions.
  45. A prompt pattern catalog to enhance prompt engineering with chatgpt.
  46. Bloom: A 176b-parameter open-access multilingual language model. arXiv preprint arXiv:2211.05100.
  47. Wizardlm: Empowering large language models to follow complex instructions.
  48. On what basis? predicting text preference via structured comparative reasoning.
  49. Information-theoretic text hallucination reduction for video-grounded dialogue. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 4182–4193, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
  50. R-tuning: Teaching large language models to refuse unknown questions. arXiv preprint arXiv:2311.09677.
  51. The knowledge alignment problem: Bridging human and external knowledge for large language models.
Citations (124)
View on Semantic Scholar

Summary

  • The paper categorizes various methods, including prompting techniques and Retrieval-Augmented Generation, to integrate verified knowledge and reduce hallucinations.
  • It demonstrates the use of self-refinement and supervised fine-tuning approaches like Knowledge Injection to improve text accuracy in critical applications.
  • It highlights challenges such as data reliability and cross-domain adaptability while proposing future directions like hybrid architectures for enhanced model safety.

Introduction

Hallucination in LLMs is a recognized problem that occurs when these models generate text containing inaccurate or unfounded information. This presents substantial challenges in applications like summarizing medical records or providing financial advice, where accuracy is vital. The survey explored in this discussion tackles over thirty-two different techniques developed to mitigate hallucinations in LLMs.

Hallucination Mitigation Techniques

The survey categorizes these techniques into several groups. Prompting methods focus on optimizing instructions to generate more accurate responses. For instance, Retrieval-Augmented Generation (RAG) incorporates external knowledge to update and enrich model responses. Techniques that unfold through self-refinement leverage feedback to improve subsequent outputs, such as the Self-Reflection Methodology that iteratively refines medical QA responses.

Furthermore, studies have also proposed novel model architectures specifically designed to tackle hallucinations, including decoding strategies like Context-Aware Decoding (CAD), which emphasizes context-relevant information, and the utilization of Knowledge Graphs (KGs) that enable models to ground responses in verified information.

Supervised Fine-Tuning

Supervised fine-tuning refines the model on task-specific data, which can significantly improve the relevance and reliability of text produced by LLMs. For example, Knowledge Injection techniques infuse domain-specific knowledge, while others like Refusal-Aware Instruction Tuning (R-Tuning) teach the model when to avoid responding to certain prompts due to knowledge limitations.

Challenges and Future Directions

The survey addresses the challenges and limitations associated with current hallucination mitigation techniques. These include the varying reliability of tagged datasets and the complexity of implementing solutions that could work across different language domains and tasks. Looking forward, potential directions include hybrid models that integrate multiple mitigation approaches, unsupervised learning methods to reduce reliance on labeled data, and the development of models with inherent safety features to tackle hallucinations.

Conclusion

The thorough survey presented in this discussion offers a structured categorization of hallucination mitigation techniques, providing a basis for future research. It underscores the need for continued advancement in this area, as the reliability and accuracy of LLMs are critical for their practical application. With ongoing review and development of mitigation strategies, we move closer to the goal of creating LLMs that can consistently produce coherent and contextually relevant information, while minimizing the risk and impact of hallucination.

File Document Download Save Streamline Icon: https://streamlinehq.com PDF File Document Download Save Streamline Icon: https://streamlinehq.com Markdown
Youtube Logo Streamline Icon: https://streamlinehq.com

YouTube