Self-Specialization: Uncovering Latent Expertise within Large Language Models

TL;DR

This paper introduces self-specialization, a method for efficiently adapting large language models to specific expert domains like biomedicine and finance, outperforming generalist models and other adaptation techniques.

Contribution

It proposes a novel self-specialization approach that enables effective domain-specific model tuning with minimal data and parameters, improving upon existing instruction-tuning methods.

Findings

Self-specialized models outperform base models in biomedical and financial tasks.

Self-specialization achieves significant performance gains with few labeled seeds.

The method surpasses instruction-tuned and domain-adapted models in experiments.

Abstract

Recent works have demonstrated the effectiveness of self-alignment in which a large language model is aligned to follow general instructions using instructional data generated from the model itself starting from a handful of human-written seeds. Instead of general alignment, in this work, we focus on self-alignment for expert domain specialization (e.g., biomedicine, finance). As a preliminary, we quantitively show the marginal effect that generic instruction-following training has on downstream expert domains' performance. To remedy this, we propose self-specialization - allowing for effective model specialization while achieving cross-task generalization by leveraging only a few labeled seeds. Self-specialization offers a data- and parameter-efficient way of "carving out" an expert model out of a generalist pre-trained LLM. Exploring a variety of popular open large models as a base for specialization, our experimental results in both biomedical and financial domains show that our self-specialized models outperform their base models by a large margin, and even larger models that are generally instruction-tuned or that have been adapted to the target domain by other means.