Cheap and Effective Personalization of Foundation Language Models for Imitating a User's Writing Style

Armand Mihai Nicolicioiu, Eugenia Iofinova, Andrej Jovanovic, Eldar Kurtic, Mahdi Nikdan, Andrei Panferov, Ilia Markov, Nir N Shavit, Dan Alistarh

Keywords: Generative AI, LMs, Personalization, Efficient ML

TL;DR: We propose a method for cheap and effective adaptation of generative language models for imitating a user's writing style. We evaluate existing personalization metrics and release three novel datasets for this use-case.

Abstract: The availability of powerful open-source foundation language models (colloquially, LMs) opens exciting use-cases, such as using personal data to fine-tune these models to imitate a user’s unique writing style. Two key requirements for such assistants are personalization-in the sense that the assistant should recognizably reflect the user’s own writing style—and privacy–users may justifiably be wary of uploading extremely personal data, such as their email archive, to a third-party service. In this paper, we present a new design and evaluation for such an automated assistant, for the specific use case of email generation, which we call Panza. Panza’s personalization features are based on a combination of fine-tuning using a variant of the Reverse Instructions technique and Retrieval Augmented Generation (RAG). We demonstrate that this combination allows us to fine-tune an LM to reflect a user’s writing style using limited data, while executing on extremely limited resources, e.g. on a free Google Colab instance. Our key methodological contribution is the first detailed study of evaluation metrics for this personalized writing task, and of how different choices of system components–the use of RAG and of different fine-tuning approaches–impact the system’s performance. Additionally, we demonstrate that very little data - under 100 email samples - are sufficient to create models that convincingly imitate humans. This finding showcases a previously-unknown attack vector in language models - that access to a small number of writing samples can allow a bad actor to cheaply create generative models that imitate a target’s writing style. We are releasing the full Panza code as well as three new email datasets licensed for research use.

Submission Number: 11