Psych-LLM: Advanced RAG Pipeline for Psychology Research

📖 Overview

A Retrieval-Augmented Generation (RAG) system designed for psychology research documentation, developed for the CASML Generative AI Hackathon. This solution combines advanced document retrieval with generative AI to provide context-aware answers from academic literature (OpenStax Psychology (2e), CC BY 4.0).

Key Features:

• PDF document parsing with page-aware chunking
• Hybrid vector database integration (ChromaDB)
• Metadata-enriched text embeddings
• LLM-powered response generation with source attribution

🛠️ Technical Stack

• Embeddings: TogetherAI/m2-bert-80M-2k-retrieval
• Vector DB: ChromaDB
• LLM: meta-llama/Llama-3.3-70B-Instruct-Turbo-Free
• Text Processing: pdfplumber, custom chunking pipeline

⚙️ Installation

Prerequisites

• Python 3.9+
• Together API key

Setup Instructions

1. Clone repository

git clone https://github.com/yash4agr/Psych-LLM.git
cd psych-llm

2. Create a virtual environment

python -m venv .venv

3. Activate the virtual environment

# On Windows:
.venv\Scripts\activate

# On macOS/Linux:
source .venv/bin/activate

4. Install dependencies

pip install -r requirements.txt

5. Register and obtain your API key

   • Visit Together API Key Settings
   • Generate an API key

6. Export your Together API Key

export TOGETHER_API_KEY=<your_api_key>  # On macOS/Linux
$env:TOGETHER_API_KEY="<your_api_key>"  # On Windows (Powershell)

7. Run the application

python main.py