arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2403.12244cs.CL
收藏

零样本多任务幻觉检测

Zero-Shot Multi-task Hallucination Detection

IIIT Hyderabad(印度信息技术学院海得拉巴分校)

Patanjali Bhamidipati, Advaith Malladi, Manish Shrivastava, Radhika Mamidi

AI总结:

本研究正式定义了幻觉问题,提出一种零样本设置下的多任务幻觉定量检测框架,在保持高计算效率的同时实现了0.78和0.61的检测准确率。

摘要:

近期研究中,大语言模型的广泛应用凸显了用于评估文本生成质量及与特定任务相关性的稳健评估方法的重要性。这揭示了一个普遍存在的问题,即幻觉,这是模型中的一种突发状况,其生成的文本对来源缺乏忠实性并偏离评估标准。在本研究中,我们正式定义了幻觉,并利用该定义以及模型输出蕴含任务和样本特定输入的假设,提出了一个在零样本设置下对其进行定量检测的框架。在检测幻觉时,我们的解决方案在模型感知设置下达到了0.78的准确率,在模型无关设置下达到了0.61的准确率。值得注意的是,我们的解决方案保持了计算效率,所需的计算资源远少于其他SOTA方法,符合轻量化和压缩模型的趋势。

英文摘要:

In recent studies, the extensive utilization of large language models has underscored the importance of robust evaluation methodologies for assessing text generation quality and relevance to specific tasks. This has revealed a prevalent issue known as hallucination, an emergent condition in the model where generated text lacks faithfulness to the source and deviates from the evaluation criteria. In this study, we formally define hallucination and propose a framework for its quantitative detection in a zero-shot setting, leveraging our definition and the assumption that model outputs entail task and sample specific inputs. In detecting hallucinations, our solution achieves an accuracy of 0.78 in a model-aware setting and 0.61 in a model-agnostic setting. Notably, our solution maintains computational efficiency, requiring far less computational resources than other SOTA approaches, aligning with the trend towards lightweight and compressed models.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑