Comparison of Hybrid Fine-Tuning and Retrieval Approaches for Cross-Domain Generalization in Mental Health Datasets

This study presents a systematic comparison of three approaches for the analysis of mental health text using large language models (LLMs): prompt engineering, retrieval augmented generation (RAG), and fine-tuning. Using LLaMA 3, we evaluate these approaches on emotion classification and mental health condition detection tasks across two datasets. Fine-tuning achieves the highest accuracy (91\% for emotion classification, 80\% for mental health conditions) but requires substantial computational resources and large training sets, while prompt engineering and RAG offer more flexible deployment with

Research goal: How do hybrid approaches combining fine-tuning and retrieval (e.g., fine-tuned RAG) compare to pure fine-tuning or RAG in cross-domain generalization, measured by accuracy on out-of-domain mental health datasets?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.5/10.