arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2407.16604cs.CL
收藏

共享想象:LLM产生相似的幻觉

Shared Imagination: LLMs Hallucinate Alike

Salesforce AI Research(Salesforce AI研究部门)

Yilun Zhou, Caiming Xiong, Silvio Savarese, Chien-Sheng Wu

AI总结:

本文提出虚构问答(IQA)设定以探究LLM相似性,发现不同模型能成功回答彼此生成的纯虚构问题,揭示了模型在幻觉中存在“共享想象空间”,并讨论了其对模型同质性与计算创造力的启示。

摘要:

尽管大型语言模型(LLMs)近期大量涌现,但它们的训练配方——模型架构、预训练数据和优化算法——通常非常相似。这自然引发了关于这些生成模型之间相似性的疑问。在本文中,我们提出了一种新设定,即虚构问答(IQA),以更好地理解模型相似性。在IQA中,我们让一个模型生成纯虚构的问题(例如关于物理学中完全捏造的概念),并提示另一个模型进行回答。令人惊讶的是,尽管这些问题完全是虚构的,但所有模型都能以极高的成功率回答彼此的问题,这表明这些模型在此类幻觉期间运行于一个“共享想象空间”中。我们对这一现象进行了一系列调查,并讨论了其对模型同质性、幻觉和计算创造力的启示。

英文摘要:

Despite the recent proliferation of large language models (LLMs), their training recipes -- model architecture, pre-training data and optimization algorithm -- are often very similar. This naturally raises the question of the similarity among the resulting models. In this paper, we propose a novel setting, imaginary question answering (IQA), to better understand model similarity. In IQA, we ask one model to generate purely imaginary questions (e.g., on completely made-up concepts in physics) and prompt another model to answer. Surprisingly, despite the total fictionality of these questions, all models can answer each other's questions with remarkable success, suggesting a "shared imagination space" in which these models operate during such hallucinations. We conduct a series of investigations into this phenomenon and discuss implications on model homogeneity, hallucination, and computational creativity.

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。

↑