Introduction

This project demonstrates one way of leveraging LLM as a copilot in assisting data record classification task. The approach employs prompt engineering technique called "ReAct" as a reasoning module that enables our Agent to be able to interact with external information beyond LLM knowledge scope.

Setup

1. make sure you have conda installed

2. create .env in this folder containing the following keys to set LLM API key:

TYPHOON_API_KEY=...

1. setup python environment

$ make setup-env

1. setup python dependencies

$ make setup-deps

1. attach agent_llm.ipynb to conda env genai_share to run the notebook.