arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.04574cs.CL
收藏

使用参数高效微调定制大语言模型的生成风格

Customizing Large Language Model Generation Style using Parameter-Efficient Finetuning

Carnegie Mellon University(卡内基梅隆大学)

Xinyue Liu, Harshita Diddee, Daphne Ippolito

AI总结:

本文研究如何用基于低秩适配的PEFT将LLaMA-2定制为十位作者风格,证明其可在词汇、句法和表层特征上对齐目标作者,但内容记忆仍有限。

摘要:

一刀切的大语言模型(LLM)正越来越多地被用于帮助人们写作。然而,这些模型被训练出来的写作风格可能并不适合所有用户或使用场景。如果LLM的个人语言风格能够被定制以匹配每位用户,它们作为写作助手将更有用。本文探讨了采用低秩适配的参数高效微调(PEFT)能否有效引导LLM生成内容的风格。我们使用该方法将LLaMA-2定制为十位不同作者的风格,并表明生成的文本在词汇、句法和表层层面与目标作者保持一致,但在内容记忆方面存在困难。我们的发现凸显了PEFT支持高效、用户级LLM定制的潜力。

英文摘要:

One-size-fits-all large language models (LLMs) are increasingly being used to help people with their writing. However, the style these models are trained to write in may not suit all users or use cases. LLMs would be more useful as writing assistants if their idiolect could be customized to match each user. In this paper, we explore whether parameter-efficient finetuning (PEFT) with Low-Rank Adaptation can effectively guide the style of LLM generations. We use this method to customize LLaMA-2 to ten different authors and show that the generated text has lexical, syntactic, and surface alignment with the target author but struggles with content memorization. Our findings highlight the potential of PEFT to support efficient, user-level customization of LLMs.

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。

↑