Open-Source LLMs for Text Annotation: A Practical Guide for Model Setting and Fine-Tuning

TL;DR

This paper evaluates open-source LLMs for political science text classification, demonstrating that fine-tuning enhances their performance to rival proprietary models and providing practical guidance for researchers.

Contribution

It offers a comprehensive assessment of open-source LLMs in text annotation, highlighting the benefits of fine-tuning over zero-shot and few-shot approaches.

Findings

Fine-tuning improves open-source LLM performance.

Fine-tuned open-source LLMs can match or surpass GPT-3.5 in some tasks.

A Python notebook is provided for practical application.

Abstract

This paper studies the performance of open-source Large Language Models (LLMs) in text classification tasks typical for political science research. By examining tasks like stance, topic, and relevance classification, we aim to guide scholars in making informed decisions about their use of LLMs for text analysis. Specifically, we conduct an assessment of both zero-shot and fine-tuned LLMs across a range of text annotation tasks using news articles and tweets datasets. Our analysis shows that fine-tuning improves the performance of open-source LLMs, allowing them to match or even surpass zero-shot GPT-3.5 and GPT-4, though still lagging behind fine-tuned GPT-3.5. We further establish that fine-tuning is preferable to few-shot training with a relatively modest quantity of annotated text. Our findings show that fine-tuned open-source LLMs can be effectively deployed in a broad spectrum of text annotation applications. We provide a Python notebook facilitating the application of LLMs in text annotation for other researchers.