重新审视子词分词:以大型语言模型中的词缀否定为例
Revisiting subword tokenization: A case study on affixal negation in large language models
AI总结:
本文通过多种子词分词方式的 LLM 实验,考察词缀否定识别能力,揭示分词准确率与否定敏感性的关系,并发现模型总体上能可靠理解词缀否定。
摘要:
在这项工作中,我们衡量了词缀否定对现代英语大型语言模型(LLM)的影响。在词缀否定中,否定意义通过否定语素表达,这对 LLM 可能具有挑战性,因为它们的分词器通常在形态上并不合理。我们使用采用不同子词分词方法的 LLM 开展了大量实验,由此得出关于分词性能与否定敏感性之间相互作用的若干洞见。尽管分词准确率与否定检测性能之间存在一些有趣的不匹配,但我们表明,总体而言,模型能够可靠地识别词缀否定的含义。
英文摘要:
In this work, we measure the impact of affixal negation on modern English large language models (LLMs). In affixal negation, the negated meaning is expressed through a negative morpheme, which is potentially challenging for LLMs as their tokenizers are often not morphologically plausible. We conduct extensive experiments using LLMs with different subword tokenization methods, which lead to several insights on the interaction between tokenization performance and negation sensitivity. Despite some interesting mismatches between tokenization accuracy and negation detection performance, we show that models can, on the whole, reliably recognize the meaning of affixal negation.
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。