[论文解读] A Comprehensive Survey of Hallucination Mitigation Techniques in Large Language Models

S. M Towhidul Islam Tonmoy, S M Mehedi Zaman+5 more|arXiv (Cornell University)|Jan 2, 2024

COVID-19 diagnosis using AI被引用 109

一句话总结

本综述汇总了超过三十二种用于缓解大语言模型（LLMs）幻觉的技术，提出一个分类法和洞见以指导未来研究。

ABSTRACT

As Large Language Models (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.

研究动机与目标

• 描述在不同模型和任务下，LLMs 幻觉缓解方法的谱系。
• 提供一个结构化的分类法来组织检索、提示、模型开发和评估方法。
• 分析挑战、局限性及潜在方向，以指导未来研究与部署。
• 强调在LLM输出中的可信性、可靠性和安全性的实际考量。

提出的方法

• 构建覆盖提示、模型开发和评估的幻觉缓解技术的全面分类法。
• 综合包括检索增强生成、自我改进、提示微调、知识图谱、可信性损失以及有监督微调等技术。
• 讨论影响事实性和基础性的端到端 RAG 系统与解码策略。
• 总结著名系统与框架（如 RAG、D&Q、EVER、CoVe、CoNLI）及其在减少幻觉方面的报道作用。
• 提供对数据集使用、反馈机制和检索器类型的跨领域分析，以指导未来研究。

实验结果

研究问题

• RQ1 LLMs 幻觉缓解的主要类别和技术有哪些？
• RQ2 检索、提示、模型开发和评估策略在有效性与适用性方面如何比较？
• RQ3 当前幻觉缓解方法的关键局限与挑战有哪些，未来的有前景的方向是什么？
• RQ4 如何在不同模态和任务中量化并提升可信性与 grounding？

主要发现

• 给出一个涵盖提示工程、端到端检索、自我改进和模型架构变更的广泛幻觉缓解技术分类。
• 检索增强生成（RAG）和知识绑定/知识基础的方式被反复强调为对输出进行扎根的有效机制。
• 若干自检查和基于反馈的策略（如 EVER、CoVe、CoNLI、SC 推理）在多个任务中显示出幻觉的减少。
• 知识图谱和基于可信性的损失有助于改善事实对齐，提出了若干架构级方法。
• 该综述指出数据质量、评估挑战以及对稳健、任务和领域特定解决方案的需求等局限性。