arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2203.07618cs.CLcs.AI
收藏

语言模型抄袭吗？

Do Language Models Plagiarize?

Penn State University（宾夕法尼亚州立大学）；University of Mississippi（密西西比大学）

Jooyoung Lee, Thai Le, Jinghui Chen, Dongwon Lee

备注：
Accepted to WWW'23
阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

摘要：

过去的文献已经表明，语言模型（LM）经常记住部分训练实例，并在自然语言生成（NLG）过程中再现它们。然而，目前还不清楚LM在多大程度上“重用”训练语料库。例如，模型可以生成与训练样本在上下文上相似的释义句子。因此，在这项工作中，我们研究了三种类型的剽窃（即，逐字、释义和想法）与其训练数据进行比较，并进一步分析了在实践中广泛使用的特定领域语料库中微调LM的剽窃模式。我们的研究结果表明：（1）三种类型的剽窃广泛存在于记忆之外的语言记忆中;（2）语言记忆的大小和解码方式与剽窃程度密切相关;（3）微调后的语言记忆的剽窃模式因其语料相似性和同质性而异。鉴于LM的大部分训练数据都是从网络上抓取的，而没有通知内容所有者，他们将单词、短语甚至核心思想从训练集中重复到生成的文本中具有伦理意义。随着LM的大小和它们的训练数据的增加，它们的模式可能会加剧，这引起了人们对不加区别地追求具有更大训练语料库的更大模型的担忧。剽窃内容还可能包含个人的个人信息和敏感信息。这些研究结果总体上对当前LM在关键任务写作任务中的实用性提出了质疑，并敦促围绕观察到的现象进行更多的讨论。数据和源代码可在https://github.com/Brit7777/LM-plagiarism上获得。

英文摘要：

Past literature has illustrated that language models (LMs) often memorize parts of training instances and reproduce them in natural language generation (NLG) processes. However, it is unclear to what extent LMs "reuse" a training corpus. For instance, models can generate paraphrased sentences that are contextually similar to training samples. In this work, therefore, we study three types of plagiarism (i.e., verbatim, paraphrase, and idea) among GPT-2 generated texts, in comparison to its training data, and further analyze the plagiarism patterns of fine-tuned LMs with domain-specific corpora which are extensively used in practice. Our results suggest that (1) three types of plagiarism widely exist in LMs beyond memorization, (2) both size and decoding methods of LMs are strongly associated with the degrees of plagiarism they exhibit, and (3) fine-tuned LMs' plagiarism patterns vary based on their corpus similarity and homogeneity. Given that a majority of LMs' training data is scraped from the Web without informing content owners, their reiteration of words, phrases, and even core ideas from training sets into generated texts has ethical implications. Their patterns are likely to exacerbate as both the size of LMs and their training data increase, raising concerns about indiscriminately pursuing larger models with larger training corpora. Plagiarized content can also contain individuals' personal and sensitive information. These findings overall cast doubt on the practicality of current LMs in mission-critical writing tasks and urge more discussions around the observed phenomena. Data and source code are available at https://github.com/Brit7777/LM-plagiarism.

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。

↑