arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2602.20052cs.CL
收藏

大语言模型中的熵

Entropy in Large Language Models

Carl von Ossietzky Universität Oldenburg(奥尔登堡卡尔·冯·奥西特齐克大学)

Marco Scharringhausen

AI总结:

本研究比较了大语言模型与自然语言的词熵,发现LLM的词熵低于自然口语,旨在评估LLM训练数据对信息和不确定性的影响。

摘要:

在本研究中,将大语言模型(LLM)的输出视为一个信息源,生成从有限字母表中抽取的无限符号序列。鉴于现代LLM的概率性质,我们假设这些LLM遵循一个恒定的随机分布,因此源本身是平稳的。我们将此源的熵(每词)与自然语言(书面或口语)作为开放美国国家语料库(OANC)所代表的熵进行比较。我们的结果表明,此类LLM的词熵低于自然口语(无论是书面还是口语)的词熵。此类研究的长期目标是正式化大规模语言训练中信息和不确定性的直觉,以评估从LLM生成的训练数据训练LLM的影响。这特别指的是来自全球互联网的文本。

英文摘要:

In this study, the output of large language models (LLM) is considered an information source generating an unlimited sequence of symbols drawn from a finite alphabet. Given the probabilistic nature of modern LLMs, we assume a probabilistic model for these LLMs, following a constant random distribution and the source itself thus being stationary. We compare this source entropy (per word) to that of natural language (written or spoken) as represented by the Open American National Corpus (OANC). Our results indicate that the word entropy of such LLMs is lower than the word entropy of natural speech both in written or spoken form. The long-term goal of such studies is to formalize the intuitions of information and uncertainty in large language training to assess the impact of training an LLM from LLM generated training data. This refers to texts from the world wide web in particular.

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。

↑