Optimizing Psychological Counseling with Instruction-Tuned Large Language Models

TL;DR

The paper addresses the challenge of delivering empathetic, context-aware psychological counseling with large language models amid rising demand for mental health support. It introduces instruction tuning using a counseling-focused prompt dataset refined by professional feedback and evaluated with GPT-4-based judgments, complemented by a dedicated evaluation dataset. The approach yields an instruction-tuned model that outperforms baselines such as LLaMA-7B, LLaMA-2-7B, and Qwen-7B across empathy, relevance, and crisis-handling metrics, validated by both automatic metrics and human expert ratings. This work demonstrates a scalable, accessible tool for mental health support and underscores the importance of domain-specific prompts and rigorous evaluation in deploying therapeutic AI systems.

Abstract

The advent of large language models (LLMs) has significantly advanced various fields, including natural language processing and automated dialogue systems. This paper explores the application of LLMs in psychological counseling, addressing the increasing demand for mental health services. We present a method for instruction tuning LLMs with specialized prompts to enhance their performance in providing empathetic, relevant, and supportive responses. Our approach involves developing a comprehensive dataset of counseling-specific prompts, refining them through feedback from professional counselors, and conducting rigorous evaluations using both automatic metrics and human assessments. The results demonstrate that our instruction-tuned model outperforms several baseline LLMs, highlighting its potential as a scalable and accessible tool for mental health support.