vibhorag101 / phr-mental-chat

Llama-2's Mental Therapy Chatbot: fine-tuned 7B & 13B models for empathetic conversations. Made using HF Transformers, Gradio UI and deployed using HF TGI-Inference, FastAPI and Docker

phr-mental-chat

A mental therapy chatbot

Model and dataset details:

• The original dataset is accessible at here
• The modified dataset for llama-2 format is accessible at here
• The finetuned llama-2-7b-chat-hf model is accessible at here

To build the docker image, run following commands:

docker build -t {image_name} .
docker run --gpus all -p 5555:5555  -itd --name {container_name} {image_name}

• Chatbot will be accessible at http://{server_ip}:5555/gradio
• Chatbot API will be accessible at http://{server_ip}:5555/generate
• Roberta Emotion Classification api will be accessible at http://{server_ip}:5555/emotion
• Suicide Prediction api will be accessible at http://{server_ip}:5555/suicide
• Plutchik Emotion API available at http://{server_ip}:5000/predict_sentiment and http://{server_ip}:5000/predict_sentiment_scores
• The API docs are available at http://{server_ip}:5555/docs