This Python application implements a Retrieval-Augmented Generation (RAG) system using open-source embeddings and ChromaDB for vector storage. Key features include:

• Document processing with text splitting
• Sentence Transformers for local embeddings (no API keys needed)
• ChromaDB vector store with cosine similarity search
• Conversation memory with automatic compression
• OpenAI integration for response generation (optional)

• Uses Sentence Transformers models (all-MiniLM-L6-v2 by default)
• Supports both document and query embeddings
• Embedding dimension detection

• Loads text documents from directory (supports .txt, .md, .markdown)
• Splits documents into chunks (1000 chars with 200 char overlap)
• Preserves metadata including source file information

• Persistent ChromaDB storage with cosine similarity
• Batch document ingestion with progress tracking
• Similarity search with relevance scoring
• Collection management functions

• Maintains conversation history
• Automatic compression when exceeding 8 messages
• OpenAI-based summarization (with fallback)
• Context management for LLM prompts

• Main application class combining all components
• Document loading and indexing
• Context-aware response generation
• Interactive chat interface

• Python 3.8+
• 4GB+ RAM (recommended for embedding models)
• 2GB+ disk space for models and vector database

chromadb>=0.4.0
sentence-transformers>=2.2.0
langchain>=0.1.0
openai>=1.0.0
numpy>=1.21.0

git clone https://github.com/avinash00134/rag-chat-ai
cd rag-chat-ai

python -m venv rag_env
source rag_env/bin/activate # On Windows: rag_env\Scripts\activate

pip install -r requirements.txt

For full response generation capabilities, set your OpenAI API key:

export OPENAI_API_KEY="your-api-key-here"

Or create a .env file:

OPENAI_API_KEY=your-api-key-here

Create a documents folder and add your text files:

mkdir documents
# Add your .txt, .md, or .markdown files to this folder

python rag_chat.py

The application will automatically:

• Load and process documents from the documents folder
• Create embeddings using Sentence Transformers
• Start an interactive chat session

You: What is machine learning?
Assistant: Machine learning is a subset of artificial intelligence that enables computers to learn and make decisions from data without being explicitly programmed...

You: search artificial intelligence
# Returns relevant document chunks with similarity scores

• quit - Exit the application
• clear - Clear chat history
• search <query> - Search documents without generating response
• info - Show collection information
• history - Display chat history
• responses - Show stored LLM responses with metadata
• tokens - Display total token usage