📖 Project Overview

EmpathicAI Companion demonstrates how to fine-tune an open-source LLM on therapeutic dialogues, build an Elasticsearch-backed RAG pipeline, and run a simple agentic loop via

✨ Features

• Fine-Tuning: One-epoch demo training of gpt2 on custom dialogue data via Hugging Face’s Trainer API.
• RAG Pipeline: Index and retrieve context snippets with Elasticsearch, then condition your LLM’s output.
• Agentic Loop: FastAPI service that retrieves, generates, and suggests follow-up actions.
• Containerized: Docker & Docker-Compose setup for both Elasticsearch and the FastAPI

🚀 Getting Started

Prerequisites

• macOS/Linux with Bash
• Docker & Docker-Compose
• Python 3.10+
• GitHub account with SSH key configured

Installation

git clone git@github.com:maddykws/empathicai-companion.git  # clone repo  
cd empathicai-companion                                       # enter directory  
cp .env.example .env                                          # copy env template  
docker-compose up --build -d