Document detail
ID
oai:arXiv.org:2409.04574
Topic
Computer Science - Computation and...Author
Liu, Xinyue Diddee, Harshita Ippolito, DaphneCategory
Computer Science
Year
2024
listing date
9/11/2024
Keywords
style llms languageAbstract
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.
Liu, Xinyue,Diddee, Harshita,Ippolito, Daphne, 2024, Customizing Large Language Model Generation Style using Parameter-Efficient Finetuning
HYDRA: Hybrid Data Multiplexing and Run-time Layer Configurable DNN Accelerator
configurable proposed dnn hydra architecture science computer
Effect of Selective Lesions of Nucleus Accumbens µ-Opioid Receptor-Expressing Cells on Heroin Self-Administration in Male and Female Rats: A Study with Novel Oprm1-Cre Knock-in Rats
dorsal µ-opioid opioid brain male models effects nac behaviors rat mor-expressing cells heroin
