# PEARL: Personalizing Large Language Model Writing Assistants with Generation-Calibrated Retrievers

**Authors:** Sheshera Mysore, Zhuoran Lu, Mengting Wan, Longqi Yang, Steve Menezes, Tina Baghaee, Emmanuel Barajas Gonzalez, Jennifer Neville, Tara Safavi  
**arXiv:** [2311.09180](https://arxiv.org/abs/2311.09180)  
**Date:** 2023-11-15  
**Summary page:** https://www.summarizepaper.com/en/paper/2311.09180v1/

## Key points

- Authors address the challenge of personalization in large language models (LLMs) for writing assistance
- Proposal of PEARL, a retrieval-augmented LLM writing assistant personalized with a generation-calibrated retriever
- Retrieval method selects historic user-authored documents to personalize LLM generations
- Training novelties include data selection method and scale-calibrating KL-divergence objective
- Effectiveness demonstrated through generating personalized workplace social media posts and Reddit comments
- Generation-calibrated retriever can double as a performance predictor and enhance low-quality generations through LLM chaining

## Summary

, , , , 

In their paper titled "PEARL: Personalizing Large Language Model Writing Assistants with Generation-Calibrated Retrievers," authors Sheshera Mysore, Zhuoran Lu, Mengting Wan, Longqi Yang, Steve Menezes, Tina Baghaee, Emmanuel Barajas Gonzalez, Jennifer Neville, and Tara Safavi address the challenge of personalization in large language models (LLMs) used for writing assistance. LLMs have shown promise in improving the quality and efficiency of composition and communication. However, they often lack personalization to the author's communication style and specialized knowledge. To overcome this limitation, the authors propose PEARL, a retrieval-augmented LLM writing assistant personalized with a generation-calibrated retriever. This retriever is trained to select historic user-authored documents that are likely to best personalize LLM generations for a user request. The key novelties in training the retriever include a data selection method that identifies user requests benefiting from personalization and documents providing that benefit. Additionally, a scale-calibrating KL-divergence objective ensures accurate tracking of document benefits for personalized generation. The effectiveness of PEARL is demonstrated through its ability to generate personalized workplace social media posts and Reddit comments. Moreover, the authors showcase how a generation-calibrated retriever can double as a performance predictor and further enhance low-quality generations through LLM chaining. This innovative approach holds significant potential for improving writing assistance tools by tailoring outputs to individual users' needs and preferences.