Fine-Tuning Medical Language Models for Enhanced Long-Contextual Understanding and Domain Expertise

TL;DR

This paper investigates why fine-tuned medical language models perform poorly on long-context understanding despite improved domain knowledge, and explores optimal data composition to balance these capabilities.

Contribution

It identifies the decline in long-context understanding in medical LLMs and proposes data tuning strategies to optimize both domain knowledge and long-context performance.

Findings

Optimal data composition improves long-context understanding.

Fine-tuning data balance affects domain expertise and context comprehension.

Experiments demonstrate trade-offs between knowledge and context performance.

Abstract

Large Language Models (LLMs) have been widely applied in various professional fields. By fine-tuning the models using domain specific question and answer datasets, the professional domain knowledge and Q\&A abilities of these models have significantly improved, for example, medical professional LLMs that use fine-tuning of doctor-patient Q\&A data exhibit extraordinary disease diagnostic abilities. However, we observed that despite improvements in specific domain knowledge, the performance of medical LLM in long-context understanding has significantly declined, especially compared to general language models with similar parameters. The purpose of this study is to investigate the phenomenon of reduced performance in understanding long-context in medical LLM. We designed a series of experiments to conduct open-book professional knowledge exams on all models to evaluate their ability to read long-context. By adjusting the proportion and quantity of general data and medical data in the process of fine-tuning, we can determine the best data composition to optimize the professional model and achieve a balance between long-context performance and specific domain knowledge.