Self-Specialization: Uncovering Latent Expertise within Large Language Models
This addresses the need for efficient domain-specific AI expertise, offering a data- and parameter-efficient method for creating expert models from generalist LLMs, though it is incremental as it builds on existing self-alignment techniques.
The paper tackles the problem of specializing large language models for expert domains like biomedicine and finance by proposing self-specialization, which uses few labeled seeds to achieve large performance gains, outperforming base models and other adapted models by a significant margin.
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.