Psychotherapy-LLM/PsyCoPref-Llama3-8B
Psychotherapy-LLM/PsyCoPref-Llama3-8B is an 8 billion parameter instruction-tuned Llama 3.1 model developed by Psychotherapy-LLM, specifically fine-tuned using preference learning on the PsyCoPref dataset. This model is optimized for psycho-counseling sessions, aiming to generate high-quality, therapeutically aligned responses. It achieves an 87% win rate against GPT-4o in counseling contexts, making it highly effective for mental health support applications.
Loading preview...
Model Overview
Psychotherapy-LLM/PsyCoPref-Llama3-8B is an 8 billion parameter language model, fine-tuned from meta-llama/Llama-3.1-8B-Instruct. Its development is detailed in the paper "Preference Learning Unlocks LLMs' Psycho-Counseling Skills". The model leverages preference learning on the specialized PsyCoPref dataset, which comprises 36,000 high-quality comparison pairs aligned with professional psychotherapist preferences.
Key Capabilities
- Psycho-Counseling Optimization: Specifically trained to improve the quality and therapeutic alignment of responses in psycho-counseling scenarios.
- Preference Learning: Utilizes a unique preference learning approach to align with expert psychotherapist judgments.
- High Performance: Demonstrates an 87% win rate against GPT-4o in psycho-counseling contexts, indicating superior performance in its specialized domain.
Good For
- Applications requiring empathetic and therapeutically sound AI-driven counseling.
- Research and development in AI for mental health support.
- Generating responses that adhere to professional psychotherapy standards.