AI Assistant with Persistent Memory

This AI assistant uses Redis Stack for persistent memory storage, vector similarity search, and JSON document storage. It includes RedisInsight for visual management and monitoring.

Prerequisites

• Docker and Docker Compose
• Python 3.8+
• Git
• At least 4GB RAM available
• Internet connection for first-time model download
• Hugging Face account (optional, for faster downloads)

Initial Setup

1. Install Required Python Packages

   pip install -r config/requirements.txt
   pip install -U sentence-transformers huggingface-hub[cli]

2. Download Sentence Transformer Model

   # Optional: Login to Hugging Face for faster downloads
   huggingface-cli login
   
   # Download and cache the model (requires internet connection first time)
   python -c '
   from sentence_transformers import SentenceTransformer
   model = SentenceTransformer("sentence-transformers/all-mpnet-base-v2")
   print(f"Model downloaded and cached. Vector size: {len(model.encode('Test'))}")
   '

   The model (approximately 1.2GB) will be cached in:

   • Linux/Mac: ~/.cache/torch/sentence_transformers/
   • Windows: C:\Users\<username>\.cache\torch\sentence_transformers\

   Note: The first run requires an internet connection to download the model. Subsequent runs will use the cached model.

3. Install Docker and Docker Compose