This project demonstrates why a hybrid chat agent (classifier + LLM) can be more robust or secure than a pure LLM approach. This example uses intent classification for core tasks (like "check balance") ensuring accuracy and control of the LLM actions.

1. Install Ollama from ollama.com
2. Pull and run the llama3.2 model:

ollama pull llama3.2:latest
ollama run llama3.2:latest

The Python API server will connect to Ollama running on localhost:11434 by default.

1. Install Python dependencies:

cd classify_query
pip install -r requirements.txt

2. Train the classifier:

python fine_tune.py

This will train the classifier on the financial intents dataset and save the model for use by the API server.