GenAI Retrieval Augmented Generation (RAG) application with locally running models using Ollama, Langchain, Langfuse and Ragas

Setup

In the root directory run

docker-compose up --build

Python3 Environment

Create a new Python3 virtual environment

python3 -m venv venv

Activate the new Python3 virtual environment

source venv/bin/activate

Install the required dependencies into the new Python3 virtual environment

pip install -r requirements.txt

Langfuse

Once the Langfuse-server container is up and running;

If this is the first time you have setup the application:

• visit http://localhost:4000 in your browser
• Create a new account
• Click the 'New Project' button
• Name the project 'coreai-techspike-evaluations'
• Click 'Settings' in the left hand panel
• Click 'Create new API keys'
• Copy the secret key and paste it into your .env file 'LANGFUSE_SECRET_KEY'
• Copy the public key and paste it into your .env file 'LANGFUSE_PUBLIC_KEY'

PGVector

Reference: https://api.python.langchain.com/en/latest/vectorstores/langchain_postgres.vectorstores.PGVector.html#langchain_postgres.vectorstores.PGVector