This repository hosts a GGUF quantized version of the Mistral-7B-Instruct-v0.2 model, which has been fine-tuned to serve as a conversational therapist. The purpose of this project was to create a language model that can provide empathetic, encouraging, and thought-provoking responses in a supportive, role-playing context. By fine-tuning the powerful Mistral-7B-Instruct model on high-quality mental health conversation data, the aim was to imbue it with the conversational patterns and empathetic tone characteristic of a professional therapist.

This model is intended for experimental use and role-playing applications, allowing users to explore conversational AI in a controlled, supportive environment.

The model was fine-tuned using a diverse set of publicly available datasets focused on mental health and counseling conversations. These sources provided the conversational examples necessary to shape the model's empathetic and supportive persona. The datasets used are:

nbertagnolli/counsel-chat: A collection of conversations between users seeking mental health support and volunteer counselors.

EmoCareAI/Psych8k: A dataset of conversational logs from online mental health support forums.

AnnoMI-full.csv: A dataset from the AnnoMI project, containing annotated mental illness conversations.

AMOD Mental Health Counseling Conversations Data: A Kaggle dataset with conversational examples from mental health counseling sessions.